1 The heart of this appeal concerns the proper determination of the consequences of the fraud perpetrated by a relationship manager of a bank over a period of more than a decade. However, that relationship manager was not an employee of the appellant, Credit Suisse Trust Limited (“CS Trust”). He was at all material times a senior employee of Credit Suisse AG (“CS Bank”) where the respondents’ assets were deposited (the “Trust Assets”) and over which CS Trust owed a duty to protect and safeguard. At the risk of stating the obvious, CS Bank and CS Trust are part of the Credit Suisse Group.

2 The ultimate analysis of the consequences of the fraud depends on the nature of the breach of duty by CS Trust. CS Trust does not dispute that it was in breach of its duty to safeguard the Trust Assets — although that admission was only made about ten days into the trial in the court below. However, the parties remain divided as regards the nature of the breach. CS Trust seeks to frame the breach as a tortious breach of its duty to safeguard the respondents’ assets. On that premise, it relies on the usual tortious principles concerning causation, burden of proof, scope of duty with reference to the contractual terms, remoteness of damages, and contributory negligence, to bear on the quantification of the damages. The respondents’ case is instead premised on CS Trust’s breach of fiduciary duty which if accepted — as was found by the Judge of the Singapore International Commercial Court (the “Judge”) — would consign these points as background facts and no more.

3 As we explain below, an admission by CS Trust of a breach of its duty to safeguard the Trust Assets does not preclude a separate finding that it was nonetheless also in breach of its fiduciary duties. In our judgment, it is clear that CS Trust was indeed in breach of its fiduciary duties to the respondents with the result that the main focus of this judgment is on the proper quantification of the respondents’ claim. To that end, both parties adduced conflicting and contrasting expert evidence to construct an alternative investment portfolio on the basis that the respondents would have moved the Trust Assets to an alternative competent and professional financial advisor, had CS Trust timeously informed the respondents of the numerous unauthorised transactions. This judgment will thus examine the threshold for appellate intervention in relation to expert evidence where such expert evidence is sensitive to and subject to the objective factual evidence before the court.

4 The background facts have been canvassed in detail by the Judge in Ivanishvili, Bidzina and others v Credit Suisse Trust Ltd [2023] 5 SLR 59 (the “Judgment”). We therefore briefly set out only the salient facts for this appeal.

5 The first respondent is Mr Bidzina Ivanishvili (“Mr Ivanishvili”), a wealthy French-Georgian businessman. Credit Suisse promoted their trust services to Mr Ivanishvili in 2004 who then agreed to their proposal to establish a trust over assets exceeding US$1.1bn (the “Mandalay Trust”). The Mandalay Trust had the objective of “Inheritance Planning and Asset Holding” and was to be for the benefit of Mr Ivanishvili and his family, the second to fifth respondents (collectively, the “respondents”).

6 The appellant, CS Trust, was appointed the trustee of the Mandalay Trust. CS Trust is a wholly owned subsidiary of Credit Suisse Trust AG (“CS Trust AG”). The Trust Assets were deposited with CS Bank, in accounts at branches in Geneva (“CS Bank (CH)”) and Singapore (“CS Bank (SG)”).

7 The settled assets of the Mandalay Trust comprised bankable assets and various artworks. The Trust Assets with which we are concerned were derived from the funds that were originally deposited in March 2005 into accounts opened by Meadowsweet Assets Limited (“Meadowsweet”) at CS Bank (CH) (the “Meadowsweet Accounts”), and accounts opened by Soothsayer Limited (“Soothsayer”) at CS Bank (SG) (the “Soothsayer Accounts”). We refer to the various accounts of the Mandalay Trust collectively as the “Trust Accounts”. Meadowsweet and Soothsayer are owned by nominee companies that are in turn ultimately owned by CS Trust.

8 In 2005, shortly after the establishment of the Mandalay Trust, Meadowsweet and Soothsayer entered into discretionary portfolio management agreements with CS Bank (CH) and CS Bank (SG) respectively, granting CS Bank the mandate and authorisation to manage the assets within the Trust Accounts. Clementi Limited (“Clementi”), another related company wholly owned by CS Trust, was appointed the authorised signatory of the Trust Accounts. This meant that payments could (ostensibly) only be made out of Trust Accounts upon Clementi’s signed instructions. Mr Ivanishvili, on the other hand, was never appointed as an authorised signatory of the Trust Accounts.

9 It is undisputed that Mr Ivanishvili was responsible for: (a) certain Russian investments from 2005 to 2008; (b) recommending a US$100m investment by Meadowsweet into the Georgian Cooperation Fund in 2014; and (c) directing a US$100m loan to certain third-parties in 2014: Judgment at [478]-–[479].

10 The fraudster, one Mr Patrice Lescaudron (“Mr Lescaudron”), was Mr Ivanishvili’s relationship manager at CS Bank from 2006 until 2015, when his fraud was discovered. Mr Lescaudron was then convicted by the Swiss courts and was found to have not only misappropriated from the Mandalay Trust, but also to have covertly transferred and manipulated the Trust Assets for the purposes of concealing and covering losses in other clients’ accounts, losses which Mr Lescaudron had also caused. Amongst other acts of subterfuge, Mr Lescaudron executed fraudulent investments orders by forging Mr Ivanishvili’s signature.

11 A significant (but by no means the only) aspect of Mr Lescaudron’s fraud took the form of Unauthorised Payments Away (“UPAs”), a term used internally by Credit Suisse to refer to a type of high-risk transaction. By Credit Suisse’s own internal guidelines, transactions that are UPAs include payment transactions out of a bank account held by a structure (such as the Mandalay Trust) under a Credit Suisse trustee (such as CS Trust) that are effected by a relationship manager of CS Bank (such as Mr Lescaudron) without the necessary approval. In other words, UPAs represent unauthorised direct removals of Trust Assets and a corresponding reduction to the Trust Accounts.

12 Credit Suisse’s internal guidelines therefore required UPAs to be reduced to a minimum and required UPAs to be swiftly addressed within ten working days by obtaining adequate documentation to support the transaction. UPAs are to be identified from “Debit Advices” from CS Bank, which record payments out of a bank account. If the Debit Advices fail to adequately describe the transaction, identify the recipient of the transaction, or state the reason and nature of the transaction, the transaction will be flagged as a possible UPA. The “Trust Manager” at Credit Suisse would then need to make inquiries with the relevant relationship manager to obtain evidence of the beneficiary’s approval of the transaction. Transactions that remain UPAs are therefore self-evidently not in the best interests of the beneficiary. We therefore focus on UPAs in our brief narrative of the relevant facts, both because of the patent risk of harm generated by UPAs, as well as the undeniable fact that CS Trust would have known, and did know, about the occurrence of the UPAs from the Trust Accounts.

13 Before summarising the litany of red-flags which surfaced during Mr Lescaudron’s tenure as Mr Ivanishvili’s relationship manager, it is important to highlight that Mr Lescaudron was never formally authorised to deal with the Trust Assets. Mr Lescaudron was not authorised to withdraw or transfer funds from the Trust Accounts, open new sub-accounts, or trade on the accounts. Nevertheless, Mr Lescaudron was apparently allowed to do so with impunity between 2006 to 2015.

14 In November 2006, several months after Mr Lescaudron first took over as Mr Ivanishvili’s relationship manager in July 2006, six UPAs totalling US$35.412m were carried out by Mr Lescaudron. This staggering sum flowed out of the Trust Accounts within a very short timeframe, between 10 and 23 November 2006. CS Trust was evidently aware of this, as Ms Sim I-May Joni (“Ms Sim”), the Trust Manager of the Mandalay Trust, wrote to Mr Lescaudron on 5 December 2006 seeking clarification on the unauthorised transactions and requiring him to furnish supporting invoices. After being informed that the funds had already been paid from the Trust Accounts despite the lack of CS Trust’s prior authorisation, Ms Sim cautioned that signed instructions by Clementi were necessary and that payments should not be made based on Mr Ivanishvili’s signature. Mr Michael Low (“Mr Low”), the head of Trust Administration at CS Trust, and Ms Lina Teng (“Ms Teng”), a senior compliance officer at CS Trust, were copied. Mr Low emphasised to Mr Lescaudron that Mr Ivanishvili was not to operate the Trust Accounts without CS Trust’s permission.

15 This first set of UPAs, occurring right at the inception of Mr Lescaudron’s involvement with the Mandalay Trust, was the cause of palpable alarm within CS Trust. Ms Teng prepared a report on Mr Lescaudron’s UPAs for senior members of CS Trust and CS Trust AG. Thereafter, internal communications expressed concern at the fact that the sum of US$35.412m had been paid out to third-parties, rather than to the beneficiaries of the Mandalay Trust (ie, the respondents), and warned that UPAs were clearly prohibited and posed a grave risk of employee fraud. UPAs were said to pose “considerable regulatory and legal risks” as they made it “impossible” for CS Trust to safeguard the interests of its beneficiaries.

16 Rather than treat this unauthorised episode as a wake-up call, the evidence shows that CS Trust did nothing about Mr Lescaudron’s repeated recalcitrance over a period of approximately nine years thereafter.

17 In 2007, Mr Lescaudron expanded his repertoire of unauthorised dealings. In a single day on 11 May 2007, Mr Lescaudron made further UPAs exceeding US$46.6m. Over the course of the year, he also set up new bank accounts for Meadowsweet, despite lacking the authority to create new accounts, transferred Trust Assets between the Trust Accounts, despite lacking the authority to transfer assets, and began investing in shares and engaging in a high level of trading on the Trust Accounts, despite lacking any authority to trade. The level of trading was so high that it was described by Credit Suisse’s own subsequent internal audit as “churning”. Furthermore, Mr Lescaudron invested the Trust Assets against prevailing market sentiment and in volatile industries. From August to October 2007, Mr Lescaudron fraudulently used the Trust Assets to purchase securities at an overvalue, benefitting personally from the trades. CS Trust was aware of the transactions but failed to take any steps to enquire into the transaction details or to look into the identity of the recipient of the Trust Assets, or even to simply check in with Mr Ivanishvili to inquire whether he had approved of the transactions.

18 In March and November 2008, Mr Lescaudron carried out a total of 21 UPAs involving tens of millions of Trust Assets. The Debit Advices for these extraordinarily large and unauthorised payments did not provide any details about the nature and purpose of the transactions, and did not identify the recipient of the Trust Assets. CS Trust was by then certainly well aware of the established pattern of UPAs. A draft e-mail from Ms Sim recognised “the disturbing UPA trend” for sums that “[ran] into the millions”. Remarkably, in our view, nothing was done by CS Trust to address or even question these UPAs until 19 February 2009, almost an entire year after the Trust Assets had been paid out without authorisation.

19 On 19 February 2009, CS Trust finally wrote to Mr Lescaudron, seeking supporting documents for the 2008 UPAs. This request was effectively ignored for a further two years thereafter. To describe CS Trust’s response as desultory or lackadaisical would be charitable. During that span of time, CS Trust was only able to issue sporadic and inept reminders to Mr Lescaudron, with a lapse of several months between each reminder. Despite having proven to be demonstrably ineffective, reminders were sent time and again on 11 March 2009, 3 November 2009, 3 February 2010, 25 November 2010, and 9 December 2010. The Judge found that CS Trust was deeply concerned about being left in the dark about the recipient of millions of dollars of Trust Assets for an entire year, and found that CS Trust had been “well aware of the highly unsatisfactory trend of UPAs in which millions of dollars were transferred out of the Trust accounts without authorisation … well before 2010”: Judgment at [197].

20 It took until February 2012 for CS Trust to consider seeking an explanation from CS Bank in relation to Mr Lescaudron’s UPAs. However, CS Trust first notified Ms Josephine Novoa Sampaoli (“Ms Sampaoli”), an employee of CS Trust AG who worked closely with Mr Lescaudron, of CS Trust’s intention to write to CS Bank. In response, Ms Sampaoli conveyed Mr Lescaudron’s blatantly self-serving opinion that it was unnecessary for CS Trust to seek an explanation for the UPAs from CS Bank. Notably, Ms Sampaoli also emphasised CS Bank’s intention to grow the relationship with Mr Ivanishvili. The Judge found that Ms Sampaoli did this to emphasise the importance of Mr Lescaudron’s opinion, and that the effect of this letter was to inform CS Trust that it should not “rock the boat” lest Credit Suisse lose out on business: Judgment at [204], [378] and [380]. In the event, CS Trust did not ultimately seek any explanation from or question CS Bank.

21 In this way, Mr Lescaudron was allowed to continue to effect unauthorised transactions and deal with the Trust Assets. Despite having been made aware of Mr Lescaudron’s continued execution of the clearly prohibited UPAs, and despite being concerned about this, nothing was done by CS Trust to investigate Mr Lescaudron or prevent his unauthorised access to the Trust Assets. This abysmal state of affairs was allowed to persist unabated until 2015.

22 In 2015, the price of certain shares that Mr Lescaudron had purchased using the Trust Assets (“Raptor Shares”) collapsed, and margin calls were triggered on the Trust Accounts. Mr Lescaudron’s fraud was thereafter exposed, and he was removed as the relationship manager. Notably, internal audits by Credit Suisse in 2011 and 2012 had already identified Mr Lescaudron’s investments in Raptor Shares as suspicious, and a subsequent internal investigation concluded that Mr Lescaudron had misused information for personal gain. These findings, however, were apparently insufficient to cause or compel any change in Mr Lescaudron’s position until the collapse in 2015.

23 Incredibly, CS Trust did not immediately notify the respondents of this very significant development. A trust accountant, Mr Patrick Guldimann (“Mr Guldimann”), was instead engaged by CS Trust in October 2015 to “review” the financial statements of the Mandalay Trust. This was done discreetly and without informing the respondents. Mr Guldimann produced restated accounts for the period between 2006 to 2014, accounts which the Judge found to be inaccurate and incomplete.

24 The Judge held that CS Trust had breached its duties by 30 March 2008. The losses that flowed therefrom were not limited to the direct defalcations by Mr Lescaudron, US$79,430,773 of which had been repaid by CS Bank pursuant to a settlement agreement dated 1 December 2022. In permitting Mr Lescaudron’s continued access to the Trust Assets, CS Trust had left all of the Trust Assets vulnerable, and this allowed Mr Lescaudron to use and manipulate all of the Trust Assets for the purposes of his fraudulent schemes. Had CS Trust discharged its duty to advise the respondents by at least 30 March 2008 of Mr Lescaudron’s UPAs, Mr Lescaudron’s access to the Trust Assets would have ended, Mr Ivanishvili would have terminated his relationship with Credit Suisse, and the whole portfolio would have been removed to the management of another professional trustee.

25 In a separate Judgment on Costs and Final Orders dated 19 September 2023, the Judge awarded the respondents compensation for their losses in the sum of US$742.73m together with interest and costs (see Ivanishvili, Bidzina and others v Credit Suisse Trust Ltd [2024] 3 SLR 78 (the “Costs Judgment”). The Judge’s reasons for arriving at this amount are considered below together with the parties’ respective submissions on appeal with respect to the quantification.

26 CS Trust admits that it breached its duty to safeguard the Trust Assets by failing to take steps to detect and prevent Mr Lescaudron’s misappropriations and admits that Mr Lescaudron would have been removed from his position by 31 December 2008 had such steps been taken. CS Trust admits that it should have contacted Mr Ivanishvili directly and should have ensured that movements out of the Trust Accounts were authorised.

27 However, CS Trust’s appeal is premised on the argument that a trustee that has no duties or powers of investment cannot be liable for bad investments carried out by a third party, even if the trustee is in breach of its duty to safeguard the trust assets from misappropriation by that third party. On this premise, CS Trust submits that it did not have a duty to monitor and review the investment decisions and therefore cannot be liable for losses arising from bad or improper investments. According to CS Trust, the duty that it breached was only the duty to safeguard trust assets, a tortious duty of care, such that the ordinary tortious principles apply. CS Trust does not contend that it did exercise reasonable care, but instead relies on this characterisation to import tortious principles and thereby limit its liability and the quantum of compensation to be paid.

28 The respondents submit that CS Trust’s case essentially attacks a straw man because their claim is not for negligent investment, but for breaches of duty, including the breach of CS Trust’s fiduciary duty to act honestly and in good faith. The fundamental flaw in CS Trust’s appeal is that it attempts to characterise the breach as simply a breach of a tortious duty of care, rather than address its breach of fiduciary duties. In the case of a breach of fiduciary duties, the rules of causation, foreseeability, and remoteness do not apply, and the defaulting fiduciary may only avoid liability if it can establish that the respondents would have incurred the same loss in any event, ie, it can disprove the but for causation test.

29 It is of first importance that we highlight at the outset CS Trust’s misapprehension of the respondents’ case. CS Trust focuses its submissions on the duty and power to invest, because it treats the respondents’ claim as one seeking recovery for losses arising from bad or improper (in the sense of being unwise or imprudent) investments carried out by a third party.

30 That is, however, not the respondents’ case. The respondents’ claim is not for losses arising from bad or improper investments simpliciter. This is not merely a claim for negligent investment. The point is not whether CS Trust should be held liable for bad investments, or whether CS Trust ought to have monitored and reviewed investment decisions. The point is that the respondents are seeking a remedy for CS Trust’s abject failure to take any steps to act on the numerous red-flags repeatedly brought to CS Trust’s attention over a prolonged period of time, in breach of its duties. As a fiduciary, CS Trust ought to have acted to prevent the unauthorised and fraudulent — not merely bad or improper — “investments”. Indeed, Mr Lescaudron’s fraud went beyond the direct defalcations and unauthorised investments. Mr Lescaudron’s continued access to the Trust Assets allowed him to manipulate them as a whole in furtherance of his fraudulent schemes. The respondents therefore do not seek damages for negligent investment. Instead, their claim is quantified based on what would have happened had CS Trust performed its duties and timeously informed them of the UPAs: the Trust Assets would have been taken out of Mr Lescaudron’s hands and placed with another financial institution, rather than remaining with Credit Suisse.

31 Finally, it is crucial, in our view, to appreciate that a breach of the duty to safeguard the Trust Assets (which CS Trust admits to) does not preclude a finding that CS Trust had also breached its fiduciary duties, even if we assume that the duty to safeguard trust assets is tortious in nature. Tortious and fiduciary duties are neither binary nor mutually exclusive. These duties are theoretically distinct and may be owed (and breached) by the same person.

32 The crux of the appeal is accordingly on the nature of the duties that were breached.

33 On appeal, CS Trust primarily relies on two contentions to limit the extent of its liability: the “Scope of Duty Principle”, and, in the alternative, the defence of contributory negligence. However, both of these contentions are parasitic on CS Trust’s characterisation of the breach as being solely tortious.

34 First, CS Trust contends that: (a) the loss claimed by the respondents should be characterised as “investment losses” (ie, losses arising from improper investments), and that (b) as CS Trust did not owe a duty to invest the Trust Assets or to monitor, manage, or supervise their investment, (c) it therefore should not be liable because “investment losses” fall outside its scope of duty (ie, the Scope of Duty Principle). However, the relevance of the characterisation of the loss and the significance of a duty to invest is premised on a finding that only a duty of care was breached. Second, CS Trust contends that Mr Ivanishvili should be found partially liable due to his contributory negligence. This argument is also predicated entirely on a finding that the duty that was breached was not fiduciary in nature. The first issue that arises for our determination is therefore the nature of the duties breached.

35 In this regard, it is clear that before the Judge, the case brought by the respondents was a case of a breach of a trustee’s duty qua trustee. As the Judge said at [556] of the Judgment:

36 Thereafter, we consider the quantification of losses which the respondents have suffered. In this regard, the applicable principles are closely tied to the fiduciary nature of the duties that were breached.

37 CS Trust effectively submits that there are in substance only two fiduciary duties: the no-profit rule and the no-conflict rule. According to CS Trust, these two rules are the principal obligations of a fiduciary because fiduciary duties are proscriptive in nature rather than prescriptive. The duty of honesty, the duty to act in the interest of the principal, and the duty of good faith “flow” from both rules, such that “if there is no breach of the no-profit rule or the no-conflict rule, there should be no breach of the duty to act honestly and in good faith in the interest of the beneficiary”. This is CS Trust’s primary case theory. Notably, it does not argue that it did perform the trust honestly and in good faith for the benefit of the respondents.

38 On this basis, CS Trust submits that as there was neither a breach of no-profit rule nor the no-conflict rule, CS Trust was thus not in breach of its fiduciary duties.

39 With respect, CS Trust’s submission is untenable. It is beyond doubt that an express trustee owes a fiduciary duty to perform the trust honestly and in good faith for the benefit of, and in the interest of, the beneficiaries of the trust. This duty is both a fiduciary duty and part of the irreducible core obligations of a trustee: Tan Yok Koon v Tan Choo Suan and another and other appeals [2017] 1 SLR 654 (“Tan Yok Koon”) at [187], [191], [194], [205] and [212]. Indeed, this axiomatic duty is inherent in the nature of an express trust, as an express trustee voluntarily undertakes to manage the trust property for the benefit of the beneficiaries, and not for the trustee’s own benefit.

40 CS Trust relies on Christopher Hare and Vincent Ooi, Singapore Trusts Law (LexisNexis, 2021) (“Singapore Trusts Law”) at [15-35] in support of its argument that all other fiduciary duties “flow” from the no-profit rule and the no-conflict rule. In our view, the learned authors of Singapore Trusts Law could not have meant that fiduciary duties are in substance limited to the no-profit rule and the no-conflict rule. Indeed, in Tan Yok Koon (at [191]), which the learned authors cite as authority, this court affirmed that the duty to perform the trust honestly and in good faith for the benefit of the beneficiaries is itself a fiduciary duty, as it constitutes another subset of the parent duty of loyalty.

41 The duty of loyalty owed by a fiduciary to their principal is the root from which the recognised and more specific fiduciary duties stem. Each represents a facet of the parent duty of loyalty (Bristol and West Building Society v Mothew [1998] Ch 1 (“Bristol”) at 18A–C, per Millett LJ). As Bristol makes clear, the distinguishing obligation of a fiduciary is the obligation of loyalty, with the specific duties including the no-profit rule and the no-conflict rule being facets or manifestations of that fundamental duty of loyalty.

42 The no-profit rule, which prohibits a fiduciary from obtaining an advantage out of their fiduciary position without the informed consent of the principal, and the no-conflict rule, which prohibits a fiduciary from placing themselves in a position of conflict of interest with their principal, are core fiduciary duties. Equally, the fiduciary duty to act in good faith is also a core fiduciary duty: Sim Poh Ping v Winsta Holding Pte Ltd and another and other appeals [2020] 1 SLR 1199 (“Sim Poh Ping”) at [253]. The no-profit rule and the no-conflict rule are not exhaustive as to the breadth of fiduciary duties. The distinguishing obligation of a fiduciary is the obligation of loyalty, which is, and ought to be, an onerous and exacting standard. The principal relies on the fiduciary to act in their best interests, and the standard expected of a fiduciary is accordingly the highest standard known to the law. The no-conflict and no-profit rule alone are insufficient to vindicate this high standard. The point may be made with an extreme example. A fiduciary cannot be said to be loyal if they intentionally take active steps against the interests of their principal, even if this does not arise from a conflict of interests or in relation to obtaining an unauthorised advantage. To find otherwise is inimical to loyalty as a concept. Accepting CS Trust’s submission that there can be no breach of fiduciary duty without a breach of the no-profit or no-conflict rule would make a mockery of the obligation of loyalty.

43 Furthermore, we do not accept CS Trust’s use of the supposedly “proscriptive” nature of fiduciary duties as the litmus for determining whether a duty is fiduciary or otherwise. While the “proscriptive” description of fiduciary duties has been the subject of lively academic debate (discourse which has been referred to by this court in Tan Yok Koon at [192]), it is not the law in Singapore that fiduciary duties are exclusively proscriptive. An academic taxonomic exercise should not obscure the substantive content of the duties. Setting aside the ease in which the reframing of a given duty can blur the prescriptive-proscriptive dichotomy (robbing the dichotomy of practical utility), there is also no principled basis for circumscribing fiduciary duties in this way. What is required of a fiduciary depends on the specific facts and circumstances. For example, a fiduciary that does not actively place themself in a position of conflict may nevertheless subsequently discover the existence of such a conflict. In such a circumstance, the fiduciary will be duty-bound to make full disclosure of that conflict as soon as is practicable and take steps to extricate themselves from the position of conflict.

44 The core fiduciary duty to act in good faith is essential to uphold the concept of the obligation of loyalty. The no-conflict rule and no-profit rule are typically binary in nature — a fiduciary is either in a position of potential/actual conflict or they are not. Similarly, a fiduciary has either procured an unauthorised advantage or has not. In contrast, what good faith entails is invariably contextual. In the context of express trusts, this is reflected in the general duty to perform the trust honestly and in good faith for the benefit of, and in the interests of, the beneficiaries of the trust. To act in good faith, a trustee must not simply refrain from acting in bad faith, something more is required. As Paul Finn correctly observed, express trustees and directors are fiduciaries that are not generally subject to the immediate control and supervision of their principals in the exercise of their powers and performance of duties, and it should therefore be unsurprising that they are positively required to act in the interests of their principals (P D Finn, Fiduciary Obligations (The Law Book Company, 1977) at [6] and [11]).

45 In the context of company law, it is well established that directors owe a fiduciary duty to act honestly and in good faith in the best interests of their company, such a duty being the “distinguishing obligation of a director”: BIT Baltic Investment & Trading Pte Ltd (in compulsory liquidation) v Wee See Boon [2023] 1 SLR 1648 at [31]–[33]. The test involves both subjective and objective elements. The court considers whether, in entering into a transaction or exercising a discretion (or omitting to do so), the director had acted bona fide in what the director considered or believed (and not what the court considers) to be in the interests of the company. A director might not be liable for an outcome that subsequently turns out to be financially detrimental if the decision to embark upon the course of action that led to that negative outcome was an honest commercial decision. In this sense, the focus is on the decision-making process rather than on the outcome of the action. However, the director’s subjective belief (or claimed belief) is not determinative. Where the transaction or decision is not objectively in the interests of the company, the court then considers whether an intelligent and honest director could, in the circumstances, have reasonably believed that the transaction or decision was for the benefit of the company. A negative answer could lead to the inference that the director had not acted honestly: Goh Chan Peng and others v Beyonics Technology Ltd and another and another appeal [2017] 2 SLR 592 at [35]–[36]. A dishonest director will be in breach regardless of whether he claims to have been acting in the interests of the company: Ho Kang Peng v Scintronix Corp Ltd (formerly known as TTL Holdings Ltd) [2014] 3 SLR 329 at [39].

46 In our judgment, in the trust law context, the trustee’s fiduciary duty to perform the trust honestly and in good faith for the benefit of, and in the interest of, the beneficiaries of the trust has a positive dimension that can manifest in at least two ways.

47 First, the duty can be adjectival, in the sense that it attaches to or regulates the performance of non-fiduciary duties or exercise of powers. In performing these duties and in exercising these powers, trustees must act in good faith. For example, a trustee acting for a class or group of beneficiaries with potentially divergent interests must perform their duties and exercise their powers with an even hand between their beneficiaries’ interests. As with directors’ fiduciary duty (above at [45]), a trustee must properly and in good faith consider the interests of their beneficiaries. We hasten to add that the focus is not on obliging the trustee to act as a guarantor and to insure a particular outcome, but to ensure that the trustee sets his mind towards that goal and conducts himself in a manner befitting his obligation of loyalty and his undertaking to act in the beneficiaries’ interests.

48 Second, the duty can be actuating, in the sense that the duty to act in good faith (in itself) will positively require a trustee to act in circumstances where he knows that the interests of the beneficiaries are at risk of harm. In Foo Jee Seng and others v Foo Jhee Tuang and another [2012] 4 SLR 339, even though the trustee appeared to have a sole and absolute discretion to postpone the sale of the trust asset, the trustee was under a “fiduciary duty to act, and to exercise his discretionary powers, in the best interests of the beneficiaries” [emphasis added] and therefore could not postpone the sale of the trust asset indefinitely to the detriment of the beneficiaries (at [79]–[80]). In Centre for Laser and Aesthetic Medicine Pte Ltd v GPK Clinic (Orchard) Pte Ltd and others and another appeal [2018] 1 SLR 180, this court found that a director was in breach of her fiduciary duties when she knew that her fellow director was diverting clients away from her company but failed to report this fact to her company (at [75]–[77]). Again, the focus is on whether the trustee should have acted in the circumstances, not on whether the trustee achieved a particular outcome. The decision to act or not act must be made honestly and in good faith for the benefit of, and in the interest of, the beneficiaries.

49 Conceptualised as such, the fiduciary duty to act in good faith is clearly distinct from the non-fiduciary duty to act with skill, care, and diligence. Breaches of fiduciary duties are those that connote disloyalty — mere incompetence is not enough (Bristol at 18F). The difference lies in the animus: the duty to act in good faith targets disloyalty rather than carelessness. While both entail a scrutiny of the conduct of the person subject to the duty, the duty of care measures their performance against an objective standard of reasonable behaviour, whereas the duty to act in good faith assesses their state of mind in performing. Although there can be overlap between the fiduciary duty to act in good faith and the non-fiduciary duty of care in the factual circumstances of the breach, the distinction is not one of degree but a difference in kind. While a person’s incompetent best efforts might not be enough to discharge the duty of care, what matters to the duty to act in good faith is not competence but that they acted loyally. The duty to act in good faith therefore does not hold trustees invariably liable for any and all adverse outcomes. However, if the trustee is objectively negligent and so acted (or failed to act) in circumstances where they did not in good faith believe that to do so would be in the interests of their beneficiary, they would be in breach of both duties. As with directors’ fiduciary duty (above at [45]), if a trustee decides upon a course of conduct that is obviously against the interests of the beneficiary in the objective sense, such that an intelligent and honest trustee would not have reasonably believed that to so act would be in the interests of their beneficiary, this will support an inference of the trustee’s lack of good faith.

50 Stripped of its factual complexities, the distinction in this case may be illustrated as follows. Suppose that a trustee undertakes to hold property, a set of valuable gold bullion, not for their own benefit, but for the benefit of another: the beneficiary.

51 In the first situation, the trustee stores the gold bullion in a room such that it is left visible to passers-by. Unthinkingly, the trustee exits the room and leaves the room unsecured, resulting in the loss of the bullion to a stranger. In this situation, the trustee can be said to have been negligent in falling below the objective standard of care to be expected from a trustee in the circumstances. The breach of duty being unwitting, there is no connotation of disloyalty.

52 In the second situation, the trustee stores the gold bullion in a room such that it is left visible to passers-by. The trustee leaves the room unsecured, but does not leave the room. Instead, the trustee watches as a stranger enters the room, and watches as the stranger removes the bullion. The trustee knows that the stranger is not their beneficiary and knows (or must be taken to know) that the stranger is not entitled to take the bullion. The trustee is not in a position of conflict as regards the stranger (with whom the trustee has no connection) and does not profit from the stranger’s theft of the bullion, but the trustee clearly knows that permitting the stranger to take the bullion would not be for the benefit of or in the best interests of the beneficiary, as the beneficiary is alone entitled to the bullion. Can such a trustee be said to have been merely negligent?

53 In our judgment, the trustee’s conduct in the second situation can be regarded as nothing short of disloyal. The trustee’s failure to secure the room is negligent in either situation. However, the gravamen of the second situation lies in the trustee’s conscious disregard of the interests of the beneficiary, the very person whose interests the trustee has undertaken to safeguard. The trustee ought to have taken steps to protect the bullion and therefore protect the interests of the beneficiary, as the trustee knows that harm will result to the trust property if nothing is done. The fiduciary duty to act in good faith in this situation preserves and indeed upholds the prophylactic purpose of fiduciary duties in deterring breaches by fiduciaries. The fiduciary must ask himself not only whether he is placing himself in a position of conflict and whether he is making an impermissible profit, he must also ask himself whether he is acting in the best interests of the beneficiary.

54 With the above principles in mind, we find that CS Trust was in breach of its fiduciary duty to perform the trust honestly and in good faith for the benefit of, and in the interests of, the respondents, and was also in breach of the no-conflict rule.

55 First, CS Trust breached its fiduciary duty to act in good faith in the actuating sense. As the Judge held, to discharge its obligation to perform the trust honestly and in good faith for the benefit of the respondents, CS Trust ought to have informed and notified the respondents (or at least Mr Ivanishvili) that it “knew that millions and millions of dollars and euros were leaving its custody of the Trust without its authorisation”. In our view, it was clear from Ms Sim’s evidence that CS Trust knew that the UPAs carried out by Mr Lescaudron were against the interests of the respondents, yet CS Trust did nothing to protect the respondents’ interests.

56 Credit Suisse as a whole, and CS Trust in particular, knew that relationship managers could make unauthorised transactions, and knew that this posed elevated risks; Credit Suisse specifically labelled these transactions as UPAs and issued specific directives to deal with them. After all, UPAs represent, and directly result in a reduction or loss of the trust assets. As we have summarised above (at [14]–[16]), CS Trust knew and was indeed alarmed by the significant volume (in rate and quantum) of UPAs carried out by Mr Lescaudron from an early stage. The risk of fraud was expressly raised. Yet CS Trust did nothing for several years to inform Mr Ivanishvili of the ever-increasing volume of UPAs.

57 CS Trust knew that UPAs were against the respondents’ interests, knew that only Clementi was authorised to effect transactions and that Mr Lescaudron had no authority to transact, knew that Mr Lescaudron was carrying out large numbers of UPAs and had done so from essentially the moment he took over as relationship manager, knew that Mr Lescaudron was repeatedly ignoring instructions to cease from doing so, knew that CS Bank was continuing to facilitate the UPAs without signed instructions from Clementi, knew that the UPAs were recurring over a sustained period of time, and yet CS Trust did nothing but sat on its hands. It clearly failed to act in the interests of the respondents despite having actual knowledge of the UPAs and the risk of harm posed by UPAs. CS Trust thereby breached its fiduciary duties. CS Trust does not specifically appeal against the Judge’s finding that it had failed to act in good faith for the benefit of the respondents. Had it asserted that it was acting in good faith, the objectively detrimental nature of the persistent unresolved UPAs and the desultory efforts of CS Trust in the circumstances would nevertheless support the inference that it lacked good faith and did not believe that failing to act was in the interests of the respondents.

58 At the very least, CS Trust turned a blind eye and abdicated its duty to perform the trust honestly and in good faith for the benefit of, and in the interests of, the respondents. In Sim Poh Ping, this court held that Mr Sim, a director of a company alongside his daughters, was himself in breach of his fiduciary duties “when he had come to know of his daughters’ actions but took no action to stop them”. Mr Sim had therefore preferred the interests of third parties “in essentially turning a blind eye to his daughters’ breaches of fiduciary duty and failing to disclose the same to the Winsta Group or to take any action to stop his daughters” [emphasis added] (at [68] and [71]–--[72]). In much the same way, CS Trust deserted its post.

59 Second, CS Trust breached its fiduciary duty to act in good faith in the adjectival sense. As the Judge held, CS Trust had breached its fiduciary duty to act in good faith in carrying out its duty to safeguard the Trust Assets by allowing Mr Lescaudron continued access to the Trust Assets after 30 March 2008. No trustee acting honestly and in good faith in compliance with its duty to safeguard the Trust assets would have continued to allow his access beyond that point. Even if the duty to safeguard trust assets is regarded as a tortious duty to take reasonable care to ensure that trust property is kept secure, the fact that CS Trust admits that it has breached that duty does not preclude a finding that it had also breached its fiduciary duty to act in good faith in doing so. CS Trust sensibly and correctly accepts that there may be a concurrent breach of the duty to safeguard trust assets and a breach of fiduciary duties. As we have explained, the duty to act in good faith is concerned with the trustee’s state of mind, rather than whether the trustee fell below the objective standard of care. The same factual basis can therefore establish a breach of both fiduciary and tortious duties, provided that there is a lack of good faith that entails a connotation of disloyalty.

60 Third, in our judgment, CS Trust also breached the no-conflict rule. The Judge found that the effect of Ms Sampaoli’s email was to inform CS Trust that it should not “rock the boat” lest Credit Suisse should lose out on business, and that this caused CS Trust not to make further inquiries: Judgment at [378] and [380]. The Judge also found that CS Trust had “preferred the ‘importance’ of Mr Lescaudron in retaining the ‘big client’ ([Mr Ivanishvili]) with the Credit Suisse organisation to the compliance with its core obligation of keeping the Trust [A]ssets safe” and that its tolerance of Mr Lescaudron’s flagrant breaches over a significant period of time “was not in good faith and was unreasonable”: Judgment at [576]. In effect, the Judge found that CS Trust had preferred the business interests of Credit Suisse over the duties it owed to the respondents. Such a conclusion was amply supported by CS Trust’s witnesses, who testified that had the bank been an unaffiliated third-party bank rather than CS Bank, CS Trust would have acted to stem the UPAs.

61 In this regard, we reject CS Trust’s procedural objection that the respondents had failed to properly plead a breach of fiduciary duties. CS Trust acknowledges that the respondents did plead that CS Trust had failed to act in good faith in the best interests of the beneficiaries, a clear reference to the fiduciary duty. The respondents sufficiently pleaded the material facts underlying the claim for breach of fiduciary duties, and the fact that the phrase “fiduciary duties” was not explicitly pleaded is of no consequence (Tan Yok Koon at [182] and [184]). The no-conflict rule had been sufficiently put into issue as the respondents had pleaded material facts as to the relationship between CS Bank and CS Trust, and the lack of independence of CS Trust. In any event, no irreparable prejudice was caused to CS Trust, as the issue had been clearly raised and appreciated by both parties in their opening statements. We highlight that CS Trust did not object to the conflict of interest issue on the basis that it had not been pleaded or put in issue, until this appeal.

62 In conclusion, we agree with the Judge that CS Trust was in breach of its fiduciary duties, and not only its duty of care. It had breached its fiduciary duty to perform the trust honestly and in good faith for the benefit of the respondents, breached its fiduciary duty to act in good faith (in carrying out its duty to safeguard the Trust Assets), and in the process also breached the no-conflict rule.

63 What remedy is appropriate depends on whether the breach of fiduciary duty is a custodial breach of fiduciary duty (for which a substitutive remedy is appropriate) or a non-custodial breach of fiduciary duty (for which a reparative remedy is appropriate). In the case of a non-custodial breach of fiduciary duty, an available remedy is equitable compensation, a reparative monetary remedy (Sim Poh Ping at [126]). To make out its claim, the principal bears the legal burden of showing a breach of a core fiduciary duty and establishing that a loss has been sustained. This gives rise to a rebuttable presumption that the fiduciary’s breach caused the loss. The fiduciary then bears the legal burden of proving that the principal would have suffered the loss in spite of the breach. The upper limit of equitable compensation is the position that the principal would have been in had there been no breach (Sim Poh Ping at [254]). As the legal burden of proving that the principal would have suffered the loss in spite of the breach is squarely on the fiduciary, any uncertainty as to what would have happened had there been no breach will be resolved against the fiduciary (Sim Poh Ping at [242]).

64 In this case, the crux of CS Trust’s breach is its failure to act in good faith, a non-custodial breach of fiduciary duty. Even if there was also a custodial breach of fiduciary duty, a position which was not advanced by either party, the respondents would have a choice and can elect to seek a reparative remedy (Sim Poh Ping at [109]). We therefore approach the issue of quantification having regard to the applicable principles for non-custodial breaches of fiduciary duty.

65 As we have stated at the outset (above at [33]–[34]), much of CS Trust’s submissions were predicated on there being only a breach of a tortious duty of care. As we have found CS Trust to be liable for breach of its fiduciary duties, we need only address these parasitic issues in brief.

66 The Scope of Duty Principle as discussed and developed in South Australia Asset Management Corp v York Montague Ltd [1997] 1 AC 191 and Manchester Building Society v Grant Thornton UK LLP [2021] 3 WLR 81 is concerned only with breaches of the tortious duty of care and does not apply to breaches of fiduciary duties. This was rightly conceded by counsel for CS Trust, Mr Lee Eng Beng SC, at the hearing. In consequence, it is unnecessary to address CS Trust’s submissions on the proper characterisation of the losses and the proper construction of the trust documents to determine with whom the investment powers were vested.

67 We observe that while CS Trust had contended in its submissions that there was no “sufficient nexus” between the “investment losses” and its duty to safeguard the Trust Assets, this was in any event a misunderstanding of the respondents’ case. The relevant nexus is between the unchallenged consequence of moving the Trust Assets to another competent and professional financial advisor, and the loss claimed by the respondents, being the difference in value of the entire portfolio of Trust Assets had the Trust Assets been taken out of the hands of Mr Lescaudron as compared to the value of the Trust Assets remaining with CS Trust from the date of the breach.

68 As CS Trust was in breach of its fiduciary duties, it is also unnecessary to deal with this point. In our view, however, it is difficult to square the concept of contributory negligence with breaches of fiduciary duty. Contributory negligence connotes a failure by a person to take care of his own interests by failing to take reasonable care to guard against foreseeable harm (Asnah bte Ab Rahman v Li Jianlin [2016] 2 SLR 944 at [18]–[19] and [140]). It embodies a balancing of interests. The tortfeasor must take reasonable care not to harm others, but others must also take reasonable care to guard against harm to oneself. Parties are assumed to be on equal ground and are expected to be capable of taking care of their own interests. In contrast, the starting position in the case of a fiduciary and their principal is the dependence of the principal on the fiduciary. The principal is in a position of special vulnerability (Sim Poh Ping at [245]–[246]). Even if contributory negligence is available in the context of a breach of fiduciary duties, its application must account for the inherent power imbalance.

69 The remedy of equitable compensation awarded to the respondents for CS Trust’s breach of its fiduciary duties (see [62]–[64] above) seeks to “repair” the losses that have been caused to the respondents by those breaches. The next issue to be determined is thus the quantification of the losses for which CS Trust must compensate the respondents. The respondents are to be placed in the position they would have been in had there been no breach of fiduciary duty. This assessment was done at the trial below with the assistance of two sets of experts, the forensic accounting experts, Mr William Howell Wyndham Davies (“Mr Davies”) and Mr James Nicholson (“Mr Nicholson”); and the wealth management or investment management experts, Mr David Morrey (“Mr Morrey”) and Ms Esther Mayr (“Ms Mayr”). Mr Davies and Mr Morrey were the respondents’ experts, while Mr Nicholson and Ms Mayr were CS Trust’s experts.

70 As part of the assessment process, the wealth management experts were required to construct “Benchmark Portfolios” that were alternative medium-risk portfolios for the Meadowsweet Accounts, the Soothsayer Accounts, and accounts held by Meadowsweet with a subsidiary of CS Bank, Credit Suisse Life (Bermuda) Ltd (the “CS Life Meadowsweet Accounts”). The forensic accounting experts then used the Benchmark Portfolios to project what the returns would have been for each account if the Trust Assets had been invested in accordance with those Benchmark Portfolios. They applied various models (the “Models”) to calculate the compensation owed to the respondents based, inter alia, on different proportions of the Trust Assets in the assessment.

71 We summarise Mr Davies and Mr Nicholson’s contrasting assessments on the losses as follows:

72 There are significant differences in the results obtained and they largely stem from three main areas of contention, namely: (a) the choice of the Model used, (b) the differences in specifications used by the experts for the Models, and (c) the differences in the experts’ Benchmark Portfolios.

73 We first address the threshold for appellate intervention, before dealing with the three areas of contention in seriatim.

74 It is trite that appellate intervention in relation to a trial judge’s findings of fact is warranted only when the trial judge’s assessment is plainly wrong or manifestly against the weight of the evidence. This is because the trial judge is generally better placed to assess the credibility of witnesses, especially where oral testimony is concerned. Where the finding is based on the credibility of the witnesses’ testimony, the power of intervention will be sparingly exercised, as the trial judge would have had the benefit of directly observing and assessing the witnesses in court. The appellate court will only intervene where the trial judge’s assessment can be said to have been plainly wrong. Where the finding is based on an inference drawn from material that is also before the appellate court, such as the internal and external consistency of the witnesses’ evidence and any objective evidence on record, the appellate court is just as well placed as the trial judge to draw the appropriate inference and may intervene where the trial judge’s finding is against the weight of the objective evidence.

75 Similarly, in the case of a trial judge’s findings on expert evidence, an appellate court will ordinarily respect a trial judge’s findings on expert evidence, unless there are doubts as to whether the evidence had been satisfactorily sifted or assessed by the trial judge. If so, the appellate court may undertake a critical evaluation of the expert evidence, focusing on obvious errors of fact and/or deficiencies in the reasoning process (iVenture Card Ltd and others v Big Bus Singapore City Sightseeing Pte Ltd and others [2022] 1 SLR 302 at [115]). As with other factual findings, the appellate court will be slow to intervene unless the trial judge’s findings can be shown to be plainly wrong or against the weight of the evidence.

76 That said, the role of expert evidence and the context in which such evidence is presented informs the approach taken by the appellate court. The bulk of expert evidence is typically found in the tendered expert report, rather than in the form of oral testimony, which then forms part of the material on the record before the appellate court. The written expert report will contain not only the expert’s opinion, but also set out the materials, grounds, and processes used by the expert. Expert evidence is opinion evidence, rather than eye-witness evidence, and the primary objective in examining an expert is to probe and test the coherence, logic, and reasoning process employed by that expert in forming their opinion as explained in their expert report. The expert’s credentials and the relevance of their experience take far greater precedence over their demeanour in determining the weight that should be accorded to their evidence. Generally, the ability to scrutinise the expert report and thus the basis of the expert’s opinion places the appellate court in a good position to assess the expert’s evidence if intervention is warranted. Finally, where the trial judge’s findings are based on inferences drawn from the expert evidence, the appellate court is entitled to undertake its own evaluation to determine whether the trial judge’s inferential findings are justified on the facts.

77 In assessing the expert evidence, the trial judge must sift, weigh, and evaluate the expert opinion in the context of the particular factual matrix and the objective facts. Accordingly, the trial judge must carefully consider the factual or other premises undergirding the expert’s conclusions. While it is important to do so in the case of expert evidence on general matters relating to an expert’s field of expertise (for example, by considering the credibility of a scientific premise relied on by the expert), this assessment takes on paramount importance where the expert evidence is sensitive to and subject to the particular factual circumstances at hand. Expert evidence cannot fly in the face of proven objective facts and the appellate court is entitled to examine the underlying facts to see if the expert opinion is supported by the factual evidence (Poh Soon Kiat v Desert Palace Inc (trading as Caesars Palace) [2010] 1 SLR 1129 at [22]–[24]). For example, where a medical expert’s opinion is based on a person’s self-reported symptoms, the court is entitled to examine whether the person’s self-reports are in fact supported by the extrinsic objective evidence in evaluating the expert opinion (see, for example, Teo Ghim Heng v Public Prosecutor [2022] 1 SLR 1240 at [37]–[54]). Equally, where the expert opinion is predicated on the existence of a particular state of affairs, the treatment and weight of the expert evidence depends on whether the objective evidence establishes that such a state of affairs existed as a matter of fact.

78 We have discussed the applicable principles for two reasons. First, in the present case, it is material that the Judge expressly found that all experts applied themselves “diligently, professionally and honestly” in giving their evidence: Judgment at [592]. The Judge’s findings were therefore primarily based on the objective material contained in the expert reports, the credibility of which was not in her view significantly undermined during examination. This objective material is on the record before us. We are therefore able to scrutinise the basis upon which the experts arrived at their conclusions for each sub-component in the quantification exercise. In principle, this made it possible for us to adjust some aspects of the final quantification based on the evidence on record without having to remit the issue to the Judge or recalling the experts. Second, the primary purpose of the expert evidence in this case was to project what was likely to have (but did not) happened, in order to place the respondents in the position they would have been in had there been no breach of fiduciary duty (above at [63]). A significant factual premise underlying the expert opinions is therefore what Mr Ivanishvili would have done in the circumstances. The experts’ conclusions, which have to be premised on the specific facts of this case, must therefore be consistent with the objective factual evidence before the court.

79 With these observations in mind, we turn to examine the propriety of the specifications employed by the experts in constructing the “Benchmark Portfolios”.

80 The experts considered two categories of Models, the Whole Portfolio Model (“Model 1”) and the Specific Transactions Model (“Models 2, 3 and 4”). Within each category, the experts had differing views on the exact specifications of each Model. Mr Davies’ expert report (dated 26 March 2022) contains a helpful overview of these different Models:

81 As previously stated (above at [24]), the Judge held that but for CS Trust’s breach of duty, Mr Ivanishvili would probably have moved his whole portfolio to another financial institution to be managed by a prudent and professional portfolio manager. The Whole Portfolio Model was thus most consistent with this alternative. It simply placed the Trust Assets in a constructed medium-risk portfolio that estimated what returns would have been achieved but for CS Trust’s breach. It was not premised on all the investments in the Trust Accounts being inappropriate: Judgment at [689]–-[690]. The Judge observed that, in assessing the respondents’ loss, it was necessary to recognise that the investments in the Trust Accounts were sitting within a portfolio “the management of which was infected by fraud either directly or indirectly” and as such, maintaining parts or sections of this original portfolio would be inappropriate: Judgment at [677].

82 CS Trust’s case is that the Whole Portfolio Model is inappropriate because it overcompensates the respondents. First, the Whole Portfolio Model wrongly assumes that all investments made after 30 March 2008 would not have taken place. Similar to CS Trust’s framing of its case on liability (see [26]–[27] above), CS Trust frames its case on quantification as a breach of only a duty of care to protect and safeguard the Trust Assets by failing to prevent improper investments from taking place. Therefore, the respondents are only entitled to recover losses incurred from improper investments which they allege that CS Trust ought to have prevented, which is more appropriately measured by the Specific Transactions Model.

83 Second, the Whole Portfolio Model holds CS Trust liable for the underperformance of the Trust Accounts against the standard of the Benchmark Portfolios, even if such underperformance is unrelated to Mr Lescaudron’s wrongdoing and the underperformance may well have occurred for other reasons. The Whole Portfolio Model thus confers a windfall upon the respondents by putting them in a position as if the Trust Accounts never underperformed the market during the relevant period. This is not reflective of the reality that properly made investments do not always meet benchmarks, whether because of misfortune or otherwise.

84 Third, there is no reason why the removal of the Trust Assets from CS Bank necessarily means a complete divestment and reinvestment of the Trust Assets on 30 March 2008. It is entirely plausible that investments and funds could have been moved in specie to be managed by another portfolio manager.

85 The respondents’ case is that CS Trust had breached its fiduciary duty to act honestly and in good faith. Accordingly, the appropriate remedy for such a breach is for the account to be put in the position it would have been in had it not been for CS Trust’s breach, which requires “a hypothetical assessment of what a properly managed investment portfolio would have achieved, ie, one not managed by an unauthorised person or infected by fraud”. The Whole Portfolio Model is thus most appropriate. In such an assessment, there is no need for the respondents to prove that all the investments were a result of Mr Lescaudron’s wrongdoing. As a fiduciary in breach of its duty to act honestly and in good faith, CS Trust is liable for all losses unless it can disprove the but for causation test. The burden of proof lies on CS Trust to do so.

86 The Whole Portfolio Model simply represents an estimate of what a properly managed portfolio would in fact have achieved if not for CS Trust’s breaches. It does not guarantee that the Trust Accounts would have never underperformed the market. Moreover, there is no basis for CS Trust to contend that the Trust Assets would have remained with Credit Suisse had Mr Ivanishvili been aware of CS Trust’s breaches. Mr Ivanishvili’s unchallenged evidence was that if he had been told by CS Trust about the fraud on the Trust Accounts, he would have moved the management of assets away from Credit Suisse.

87 In our judgment, the appropriate Model to adopt ultimately depends on which party bears the burden of proof. Given our findings that CS Trust had breached its fiduciary duties (see above at [62]–[64]), once the respondents establish that they sustained loss, a presumption arises that CS Trust’s breaches caused those losses. We agree with the Judge that if CS Trust had advised Mr Ivanishvili directly by 30 March 2008 (at the latest) about Mr Lescaudron’s misconduct (above at [14]–[18]), as it was duty-bound to do, it was more probable than not that Mr Ivanishvili would have removed the Trust Assets to another professional trustee: Judgment at [546]. In the circumstances, we affirm the Judge’s acceptance of Mr Ivanishvili’s evidence that if he had been apprised of the true state of affairs, he would have moved the Trust Assets away from CS Trust and Credit Suisse. In fact, this was precisely what Mr Ivanishvili did after he had lost confidence in CS Trust upon discovering Mr Lescaudron’s fraud, although this was by then too late. In our view, it cannot be seriously disputed that a beneficiary in Mr Ivanishvili’s position would remove the trust assets from a trustee who has been shown to have failed to act in good faith and in their best interests. CS Trust itself does not dispute that it was possible for the Trust Assets to have been “moved to another financial institution” had its breaches been disclosed timeously. The respondents’ loss is thus the difference between the value of the portfolio had it been managed by a professional, competent trustee (and thus unaffected by fraud), and the value of the portfolio under CS Trust, from 30 March 2008 onwards: Judgment at [546]–[547]. We should add that our acceptance of the Whole Portfolio Model over the Specific Transactions Model has the most substantial impact on the quantification of the respondents’ loss.

88 Given the presumption that CS Trust’s breaches of its fiduciary duties caused these losses to the respondents (above at [63]), the Whole Portfolio Model is most appropriate. As we have explained, the Whole Portfolio Model assumes that, at a defined date, the Trust Assets were no longer invested in the Trust Accounts, but were instead invested in accordance with the Benchmark Portfolios, which is an estimation of how a competent professional trustee would have performed (above at [70] and [80]).

89 To arrive at a fair and realistic alternative investment portfolio, care must be taken to ensure that the Trust Assets which Mr Ivanishvili had himself invested or authorised are excluded in determining the value of the Trust Assets to be accounted for under the Whole Portfolio Model. The reason for this exclusion is obvious: any losses (or gains) stemming from Mr Ivanishvili’s personal investment decisions cannot be attributed to CS Trust. In other words, if it can be established that some of the Trust Assets would have been invested or used in a particular way regardless of CS Trust’s breach, this would effectively have disproved the but for causation test with respect to those assets. Such adjustments have to be made to the Whole Portfolio Model to ensure that the respondents are not overcompensated for their losses. In this regard, we agree with the Judge’s exclusion of certain transactions attributable to Mr Ivanishvili from the Whole Portfolio Model (namely, those referred to above at [9]). We therefore do not disturb these adjustments, subject to our examination as to whether any further adjustments ought to have been made to the specifications of the Whole Portfolio Model (below at [95]–[116]).

90 In relation to CS Trust’s criticisms of the Whole Portfolio Model, we disagree that the Whole Portfolio Model results in a windfall for the respondents. Mr Lee argues that the relief awarded to the respondents must be “legal losses” which CS Trust had caused (ie, that the “investment losses” were tainted by Mr Lescaudron’s wrongdoing). This relates to the suspicious transactions which have been identified by the experts. In our judgment, CS Trust’s approach is flawed and misses the point. It is based on an incorrect premise that the respondents’ claim is merely for losses arising from improper investments (for which the Specific Transactions Model would be more appropriate). It does not address the true premise of the respondents’ case that, but for CS Trust’s breaches of fiduciary duties, all of the Trust Assets would have been moved to be managed by a different competent professional trustee (see above at [30]). Having established such a loss (above at [87]), the only escape route available to CS Trust is for it to prove that the respondents would have suffered the same losses even under the management of the alternative competent professional trustee.

91 In any event, Mr Cavinder Bull SC, counsel for the respondents, rightly highlighted that the Judge found that, even though the experts opined that some investments were suitable, the investments were “sitting within a portfolio the management of which was infected by fraud either directly or indirectly” and therefore, it would be inappropriate to maintain parts or sections of the portfolio as it was under CS Trust: Judgment at [677]. The experts were not directed to investigate which parts of the Trust Assets were tainted by Mr Lescaudron’s fraud, and this was in any event not possible given that CS Bank did not give evidence at the trial. In other words, it was not possible for the experts to determine whether parts of the Trust Assets were tainted by Mr Lescaudron’s fraud. We thus agree with Mr Bull that it is wrong to say that CS Trust is being held liable for under-performance that has nothing to do with Mr Lescaudron’s wrongdoings. It follows that the Specific Transactions Model (advanced by CS Trust), which assesses losses based only on the suspicious transactions identified by the experts, is not workable and cannot accurately reflect the losses suffered by the respondents.

92 In our judgment, since the legal burden of proof lay fully and squarely on CS Trust to prove that the respondents would have suffered the loss in spite of their breach (above at [63]), its failure to discharge this burden means that it should remain liable for the entire loss (subject, of course, to any adjustments that may need to be made with respect to some of the specifications in light of the objective evidence: above at [89]). In any case, the available evidence appears to be in line with the Judge’s findings that the whole portfolio was tainted by Mr Lescaudron’s fraud (for example, see below at [112]).

93 Finally, we add that it is incorrect for CS Trust to characterise the Whole Portfolio Model as placing the respondents in a position where their portfolio is deemed to have never underperformed the market over 15 years. The Whole Portfolio Model is simply a means of determining the proportion of the Trust Assets that needs to be compared against the Benchmark Portfolios to quantify the respondents’ loss. The Specific Transactions Model, which accounts for a different proportion of Trust Assets is also compared against the Benchmark Portfolios to assess the respondents’ loss. Both categories of Models thus rely on the Benchmark Portfolios to derive their rates of return to be applied to the Trust Accounts in arriving at their final sums (see [69]–[71] and [80] above). Any complaint about the performance of the Whole Portfolio Model on this basis applies equally to the Specific Transactions Model (which CS Trust has advanced).

94 CS Trust’s complaint about the respondents’ portfolio being deemed to have never underperformed the market is therefore best understood as a complaint against the use of the Benchmark Portfolios to derive the rates of return, which both parties’ experts have done. There is no merit to this complaint. The Benchmark Portfolios were created by the experts with the benefit of hindsight (inevitably so) to estimate the “investment returns that would or should have been achieved had they been managed by a competent, professional investment manager where the [Trust Assets] w[ere] not affected by fraud”: Judgment at [627] and [689]). It is possible that a competent professional trustee managing the Trust Assets during the relevant period could have outperformed or underperformed the experts’ estimated rate of return. However, given that such a competent professional trustee never existed because of CS Trust’s breaches, the best the experts were able to do was to propose a fair estimate of what would have been achieved, with the knowledge of the actual performance of various categories of investments. Although the tracking of actual market performance in the projections invariably eliminates the chance of overperformance or underperformance of the market, this is not unfair. It is but a reasonable way of estimating what a competent professional trustee could have achieved if it had managed the Trust Assets during the relevant period.

95 On appeal, the parties dispute the specifications used by the experts for the Models, and specifications used by the experts in their differing Benchmark Portfolios. The former affects the portion of Trust Assets to be accounted for under the Whole Portfolio Model, the shortfall for which CS Trust needs to compensate the respondents. The latter affects the construction of the Benchmark Portfolios and the corresponding rates of return used. Given the number of specifications which are disputed on appeal, our analysis focuses on the specifications where we depart from the Judge’s decision. For the other specifications (where we agree with the Judge), we address them only briefly.

96 We observe at the outset that our departures from the Judge’s decision pertain to instances where we find that the expert evidence preferred by the Judge goes against the weight of the factual evidence, or where the point was not expressly addressed by the Judge in the decision below (above at [74]–[75]). In our view, in constructing an alternative investment portfolio, it is important to avoid doing so in the abstract or from the perspective of a hypothetical reasonable person. Instead, the construction of the alternative investment portfolio should be done (as far as possible) with reference specifically to Mr Ivanishvili and the factual matrix of this case. The objective here is to identify what was likely to have happened, so as to place the respondents in as close a position as they would have been in but for CS Trust’s breach of fiduciary duty. This being the underlying premise of the assessment, the quantification exercise necessarily requires a consideration of what Mr Ivanishvili himself would have done in the circumstances (above at [77]–[78]). The construction must therefore take into account what can be discerned about Mr Ivanishvili’s investment preferences (eg, the discretionary mandates) and risk appetite, and must necessarily consider any other objective facts relating to Mr Ivanishvili’s likely behaviour.

97 The Judge found that it was not appropriate to exclude the investments made in hedge funds from the Whole Portfolio Model. There was a paucity of evidence with respect to the nature of the hedge funds which were sought to be excluded, and CS Trust did not identify the specific investments which were made in hedge funds. The Judge accordingly did not accept CS Trust’s “broad-brush” submission that because Mr Ivanishvili had “approved of investments in hedge funds generally”, all investments in hedge funds should therefore be excluded “on that basis”: Judgment at [723]–[726].

98 On appeal, CS Trust’s case focuses on the evidence which shows Mr Ivanishvili’s involvement in the decision to invest in hedge funds. CS Trust contends that the Judge erred by declining to exclude any of the investments made in hedge funds from the Whole Portfolio Model. The respondents disagree. They argue that unless CS Trust is able to identify the specific investments in hedge funds which ought to be excluded, it is impossible for the court to determine whether the investments were carried out properly or whether they were tainted by Mr Lescaudron’s fraud.

99 After the hearing, the respondents sought permission to provide further submissions on this issue. Further submissions were filed on 3 May 2024, with CS Trust filing its reply on 10 May 2024. The thrust of the respondents’ submissions is that the evidence relied upon by CS Trust, being Mr Ivanishvili’s testimony at the trial, reveals that he had only agreed in principle to hedge funds as a type of investment. He did not authorise or instruct specific hedge fund investments. Moreover, a consideration of the Trade Lists (ie, a list accepted by the experts of investment transactions and the profit and loss of each transaction involving the Trust Assets) shows that there were investments in hedge funds which could not possibly have been authorised by Mr Ivanishvili.

100 In reply, CS Trust argues that Mr Ivanishvili had testified that hedge fund investments were a category of investments which he was personally involved in. He had given his specific and considered approval for hedge funds. In the light of this unqualified evidence, the burden fell on the respondents to identify the hedge fund investments which were not authorised and therefore the sums represented therein ought to remain within the Whole Portfolio Model. To this end, CS Trust accepts that some of sums represented in the hedge fund investments identified by the respondents ought not to be excluded from the Whole Portfolio Model because they had previously been deemed to be “fraudulent, unsuitable or overconcentrated”.

101 In our judgment, the Judge, with respect, omitted to consider the material factual evidence relating to the hedge fund investments. Mr Ivanishvili clearly admitted under cross-examination that he was generally involved in authorising hedge fund investments. Mr Ivanishvili testified that between 2008 to 2011, Mr Lescaudron advised him on hedge fund investments that he could do through CS Bank. Mr Ivanishvili had listened to Mr Lescaudron’s advice on hedge funds and CS Bank had “indeed carried out this advice”. This was unlike all “other investments [that] were happening without” him formally agreeing to them [emphasis added]. Essentially, Mr Ivanishvili admitted that unlike “other investments” for which he did not provide specific instructions, in the case of investment in hedge funds, he had explicitly authorised them.

102 Although the respondents argue that Mr Ivanishvili’s evidence was given in response to broad questions about investments “either through the trust accounts or [his] personal accounts”, this does not assist the respondents because the rest of Mr Ivanishvili’s testimony, when read in context, reveals that he was generally involved in the authorising of hedge fund investments. Subsequently, Mr Ivanishvili further agreed that Mr Lescaudron had offered several times to establish hedge funds and Mr Ivanishvili had agreed with this, because hedge funds “were outperforming banks in general” and “were very popular at that time”. He explained that because he “did not know much about hedge funds”, he wanted the money to remain within CS Bank for somebody to have oversight of it. Significantly, he acknowledged that he “had a big trust with [CS Bank]” and that “there was an agreement on [his] part to involve hedge funds”.

103 In fact, when the Judge sought clarification from Mr Ivanishvili about the scope of his management of the Trust Assets, he explained that Mr Lescaudron had offered “hedge funds to [him]” and after Mr Ivanishvili had gathered “sufficient knowledge about this matter”, he reminded Mr Lescaudron “to return to these hedge fund issues”. When asked if he accepted that he had some involvement in the management of the hedge fund arrangement, Mr Ivanishvili acknowledged that he had “instructed [Mr Lescaudron] to create this hedge fund”. Later on, when the Judge asked Mr Ivanishvili about the decisions he made when instructions were sought, he admitted that he made some decisions “maybe of organisational nature”, “maybe to open a hedge fund or [to] move certain amount from one account to another account”. He subsequently admitted that hedge funds were a “good instance” of when he made decisions on whether he wanted to invest the Trust Assets “in a particular place”.

104 Mr Ivanishvili’s testimony under cross-examination was clear and unambiguous. His responses suggest that at least some hedge fund investments were authorised by him. In our view, his broad concession clearly called for clarification during re-examination, but that was not done.

105 In the circumstances, there is some evidence on the record that shows that the hedge fund investments would have been made in any event. Accordingly, it is our view that, as a starting point, the sums represented in all the hedge fund investments ought to be prima facie excluded from the Whole Portfolio Model, unless the respondents are able to show otherwise. In other words, given the factual evidence of Mr Ivanishvili’s testimony, the evidential burden falls on the respondents to identify and show that there are specific hedge fund investments which were not authorised. This approach has the benefit of achieving a fairer estimation because it avoids an all or nothing scenario where the sums represented in all the hedge fund investments are either totally included or excluded from the Whole Portfolio Model, which would (given the state of the evidence) invariably result in either overcompensating or undercompensating the respondents.

106 To this end, we directed the respondents to specifically identify in their further submissions the hedge fund investments that should not be excluded from the Whole Portfolio Model. They were identified as follows:

(a) funds through which Mr Lescaudron indirectly invested into Raptor Shares (ie, Mensa, Hyperion, and Matterhorn, which were described by Mr Ivanishvili as the “Sequoia Funds”, as well as Exten and Marketview);

(b) investments in Copernic where Mr Lescaudron was the investment manager; and

(c) investments in Apache which both Mr Morrey and Ms Mayr agreed were unsuitable.

107 CS Trust accepts that the respondents ought to be compensated for the above hedge fund investments that were fraudulent, unsuitable or overconcentrated. Mr Morrey and Ms Mayr had agreed that the investments in Raptor Shares through the Sequoia Funds and the investments in Apache were unsuitable. The Judge also found that the investment in Copernic was made by Mr Lescaudron without Mr Ivanishvili’s knowledge. In our view, this concession is correctly made. Given the experts’ evidence and the Judge’s findings (Judgment at [216], [294], [481], [711]–[718]), it is, in our view, fair to include the sums represented in these three categories of hedge fund investments which the respondents have identified (above at [106]) within the Whole Portfolio Model, as CS Trust did not discharge its legal burden of proving that the respondents would have suffered these specific losses in spite of CS Trust’s breach of fiduciary duties.

108 As for the hedge fund investments that took place prior to the e-mail chain showing a discussion between Mr Ivanishvili’s assistant and Mr Lescaudron about hedge funds in August to October 2011, we find that this broad category of hedge fund investments should not be included in the Whole Portfolio Model. Mr Ivanishvili’s testimony clearly shows that even before August to October 2011, he had already received advice from Mr Lescaudron about hedge funds and had authorised such hedge fund investments (above at [101]–[103]). The correspondence therefore does not indicate that this was the first occasion hedge funds were contemplated by Mr Ivanishvili.

109 During the material time, there were various discretionary mandates over the Meadowsweet Accounts and the Soothsayer Accounts. For instance, two discretionary portfolio management agreements relating to Meadowsweet Account Nos 75-1 and 75-4 were signed in early 2005. These agreements defined the “Portfolio” as the “safekeeping account and transaction accounts”. CS Bank (CH) was authorised to manage the Portfolio “independently and without special instructions, except the standing special instructions agreed upon, on a fully discretionary basis in accordance with the agreed investment profile”. As another example, with respect to Soothsayer Account Nos 8-80 and 8-81, on 1 April 2005, Clementi signed a discretionary portfolio management agreement and a portfolio mandate for each account with CS Bank (SG). These agreements provided for CS Bank (SG) to manage the Soothsayer accounts from “time to time”. Under the “discretionary management by CS Bank (SG)”, CS Bank (SG) had the discretion to manage the investments and make investment decisions, as long as it was in accordance with guidelines given by Soothsayer, and Soothsayer was authorised to make additions to or withdrawals from the accounts, provided that CS Bank (SG) was given timely notice.

110 The Judge found that the investments under the discretionary mandates were not to be excluded from the Whole Portfolio Model: Judgment at [707]. Contrary to CS Trust’s claims, Mr Lescaudron “was involved with the assets being managed under the discretionary mandates”. Although the true extent of Mr Lescaudron’s involvement was tinged with some uncertainty, the Judge found that this was not a basis to exclude this category of investments from the Whole Portfolio Model: Judgment at [705].

111 We agree. The evidence shows that in the circumstances, Mr Lescaudron had both the opportunity and the ability to interfere with the investments in the discretionary mandate accounts. Moreover, the evidence suggests that in the circumstances, there is a need for further examination of whether Mr Lescaudron had interfered in the Trust Assets under the discretionary mandate accounts. Since the burden of proof is on CS Trust to establish that Mr Lescaudron had only acted on Mr Ivanishvili’s instructions in carrying out various actions in relation to the discretionary mandate accounts – and that he had not interfered with them otherwise, CS Trust’s failure to do so would mean that there would be no grounds to interfere with the Judge’s decision not to exclude the discretionary mandate accounts from the Whole Portfolio Model.

112 A clear example of Mr Lescaudron’s involvement in the discretionary mandate accounts can be illustrated by the evidence in relation to the closure of these accounts. CS Trust relies on alleged contemporaneous evidence of Mr Ivanishvili’s signed instructions in French by email to CS Bank directing the closure of three discretionary accounts. However, Mr Ivanishvili denied this, and we find that the instruction letter which CS Trust relies on to be problematic. Ms Sim testified that the disputed letter was strange as it did not reflect the names or numbers of the accounts to be closed. Ms Sim agreed that if Mr Ivanishvili wanted to terminate the bank’s management of his assets, she would “have expected the names of the accounts and their account numbers to be reflected”. Moreover, as the respondents have further pointed out, it was CS Trust who had the power to authorise the closure of the discretionary mandates, and not Mr Ivanishvili. Ms Sim confirmed that CS Trust did not authorise these closures and therefore, the Judge correctly held that the closures were unauthorised. From the evidence, it is clear that Mr Lescaudron was involved in the discretionary accounts – Mr Lescaudron was copied in the covering email accompanying the instruction letter, which states that “[Mr Ivanishvili] confirmed today by telephone with [Mr Lescaudron]” that “[Mr Ivanishvili] wants to close the three mandats [sic] he has with CS Geneva”. These were circumstances which called for an explanation, for which no satisfactory one was provided.

113 Mr Lescaudron’s first investment when he began his unauthorised investments on Meadowsweet Account No 75 in June 2007 was acquiring shares in Carpathian Resources Ltd (“Carpathian shares”). Mr Ivanishvili testified that he was not aware of this investment. The Judge, in adopting Mr Davies’ approach, implicitly decided that the Carpathian shares should be included in the Whole Portfolio Model: Judgment at [617] and [729].

114 In our judgment, CS Trust has provided no grounds for interfering with the Judge’s decision. There is no merit to CS Trust’s contention that the Carpathian shares should not have been included in the Whole Portfolio Model because this specific transfer of Carpathian shares was not pleaded or proven to be unauthorised. As CS Trust had breached its fiduciary duties to the respondents, the legal burden is fully and squarely on CS Trust to prove that the transactions relating to the Carpathian shares were in fact expressly authorised by Mr Ivanishvili. The respondents had adequately pleaded that CS Trust had “caused loss and damage” to the Trust Assets, being “the difference between the actual value of the [Trust Assets] and the value it would have had but for the unauthorised, imprudent, and/or unsuitable investments and/or trading on the Trust Accounts”. This sufficiently covers all the investments, including the Carpathian shares.

115 In computing the Whole Portfolio Model, the parties’ experts used different effective start dates because it was not possible for the Judge’s requested start date of 30 March 2008 to be used as it fell on a Sunday. Mr Davies used 31 March 2008 as an effective start date while Mr Nicholson adopted 28 March 2008. Given the substantial sums involved, the preferred start date assumes significance as, if Mr Nicholson’s calculations (which are in part) based on 28 March 2008 (and also in part based on figures from 31 March 2008) are preferred over that of Mr Davies’ calculations, the final figure arrived at would be reduced by around US$10m. The Judge preferred Mr Davies’ start date of 31 March 2008 as it was assessed “as the fairest and most reasonable in the circumstances” (the Costs Judgment at [9]).

116 In our judgment, the Judge’s adoption of Mr Davies’ approach is correct. Both Mr Davies’ and Mr Nicholson’s approach would not be able to determine the precise valuation as of 30 March 2008 due to a lack of available information on that day. To overcome this difficulty, the experts would inevitably have to use an alternative date to best reflect the valuation as at 30 March 2008. In our view, Mr Davies provided persuasive reasons why it is reasonable and proportionate to use the values reported in the investment reports of 31 March 2008 as compared to 28 March 2008. 31 March 2008 is closer to the valuation date of 30 March 2008. Crucially, a valuation date of 28 March 2008 would omit events that had in fact occurred on 29 and 30 March 2008, and this would have had an impact on the final valuation. Moreover, as a matter of principle, the start date should logically be a date occurring after the date of the breach.

117 Mr Morrey and Ms Mayr put forward different Benchmark Portfolios. This results in different average rates of annual return (as reproduced from CS Trust’s submissions):

118 The average rate of annual return is undoubtedly important to quantification. It is applied to the Whole Portfolio Model to estimate what the invested Trust Assets would have yielded. Given the long period of around 15 years to which the average rate of annual return is applied, its impact on the estimations is particularly pronounced because of the compounding effect. A small difference in the average rate of annual return can thus lead to significant differences. This is best illustrated by the huge difference in Mr Nicholson’s calculations when applying Mr Morrey’s and Ms Mayr’s conclusions.

119 Asset allocation choices for the Benchmark Portfolios were to be made in four categories: liquidity, bonds, equities and alternative investments, with alternative investments being a “catch-all phrase for everything else” including hedge funds and gold. Asset allocation has an important impact on the rate of return of the Benchmark Portfolios because the historical performance of the market shows that certain asset classes performed better than others. The experts differed on what the appropriate asset allocation should be for the CS Life Meadowsweet Benchmark Portfolio where, unlike the Meadowsweet and Soothsayer portfolios, no contemporaneous documents were available.

120 Mr Morrey’s and Ms Mayr’s competing asset allocations for the CS Life Meadowsweet Benchmark Portfolio were as follows:

121 As we understand it, having considered the experts’ evidence, the Judge preferred Mr Morrey’s allocation over Ms Mayr’s allocation because she found that Ms Mayr’s approach in ensuring “an absence of relevant bias” and “ensuring commitment to [Mr Ivanishvili’s] instructions in the construction of the Benchmark Portfolios”, was “constrained by timidity”. Mr Morrey’s allocation was more reflective of “the reality of achieving appropriate investment returns” in a counterfactual portfolio and the Judge therefore preferred it: Judgment at [694].

122 Before us, CS Trust contends that Ms Mayr’s allocation is fairer and should be preferred. In particular, her analysis of the discretionary mandates was “rooted in the contemporaneous available evidence” while Mr Morrey’s zero allocation to alternative investments was inconsistent with the available evidence. In response, the respondents maintain that Mr Morrey’s first principles approach of using well-known portfolio management theory in assessing asset allocation for the CS Life Meadowsweet Accounts was proper as a value judgment.

123 We agree with CS Trust that the Judge erred in preferring Mr Morrey’s allocation over Ms Mayr’s allocation. This is because the Judge’s decision was inconsistent with the available evidence on asset allocation. As we have earlier explained (above at [96]), the construction of the alternative investment portfolio should be done (as far as possible) with specific reference to Mr Ivanishvili and the factual matrix of this case. Where evidence of Mr Ivanishvili’s attitude towards asset allocation is available, such evidence must be accounted for as an important factor.

124 It is not disputed that there were contemporaneous documents to assist the experts with the construction of the Benchmark Portfolios for the Meadowsweet Accounts and the Soothsayer Accounts. Based on the available information, Mr Morrey was able to derive the asset allocation for those accounts as follows:

125 In our judgment, it is material that Mr Morrey allocated a portion of the Soothsayer Accounts and Meadowsweet Accounts to alternative investments, pursuant to their respective discretionary mandates. This is indicative that Mr Ivanishvili had an investment risk appetite that included alternative investments in his portfolios. The fact that there is no evidence showing that the asset allocations for the CS Life Meadowsweet Accounts were intended to be the same as that of the Soothsayer and Meadowsweet Accounts does not assist the respondents. It is this uncertainty that requires the experts to come up with an informed estimation. It does not take away the fact that the evidence shows that Mr Ivanishvili’s risk appetite in fact included alternative investments. This objective factual evidence is an important factor in the assessment of the experts’ estimations. In our view, it was inappropriate for Mr Morrey’s asset allocation to ignore this material fact and for him to allocate 0% to alternative investments for the CS Life Meadowsweet Accounts’ Benchmark Portfolio. With respect, the Judge erred in failing to acknowledge that this was inconsistent with the available factual evidence about Mr Ivanishvili’s investment preferences.

126 In contrast, Ms Mayr’s approach of taking an average of the known allocation to alternative investments (of 20%) from the available discretionary mandates for the Soothsayer Accounts (30% from February 2009) and Meadowsweet Accounts (10%) is the more accurate way of estimating the allocation to alternative investments for the CS Life Meadowsweet Accounts. That at least accounts for Mr Ivanishvili’s actual risk appetite and is well within the range of the exposure which he had demonstrably taken for the Meadowsweet Accounts and the Soothsayer Accounts. This is also consistent with Ms Mayr’s allocations to equity (32.5%), bonds (45%), and liquidity (2.5%) in the CS Life Meadowsweet Accounts by employing the averages of the discretionary mandates for the Soothsayer Accounts and Meadowsweet Accounts as well.

127 Ms Mayr’s approach of deriving the asset allocations for the CS Life Meadowsweet Accounts in utilising the known asset allocations for the Meadowsweet Accounts and Soothsayer Accounts minimises the risk of the experts being affected by hindsight bias when retrospectively devising the asset allocations for the Benchmark Portfolios. Such bias may well stem from the knowledge of the asset classes which ultimately performed better during the relevant time. The use of objective metrics based on contemporaneous evidence (ie, the known asset allocations for the Meadowsweet Accounts and the Soothsayer Accounts) mitigates this risk.

128 We thus hold that Ms Mayr’s asset allocation for the CS Life Meadowsweet Accounts should have been adopted.

129 After determining the asset allocation for the Benchmark Portfolios, the next step is to project the performance of each of the four asset classes (ie, liquidity, bonds, equities and alternative investments). The experts used a range of indices and exchange traded funds (“ETFs”) to represent the performance of each of the four asset classes. Mr Morrey explained that an index is “fundamentally the same as an ETF”, although “it is not possible to invest in an index as it simply tracks performance of a number of underlying securities but is not an investment product in its own right”. Likewise, “some ETFs are specifically designed to track a certain index and, by investing in these products, it can be possible for investors to gain exposure to a specific index”.

130 The different choices the experts have made in relation to the selection of indices and ETFs to estimate the performance of the various asset classes are in dispute on appeal.

131 Although the Judge preferred Mr Morrey’s approach (above at [121]), the Judgment appears to be silent on the reasons for her acceptance of Mr Morrey’s choice of the appropriate alternative investment indices to be used for his various Benchmark Portfolios.

132 A summary of the experts’ competing choices of the alternative investment indices for the various accounts (per Mr Nicholson’s presentation on 21 September 2022) is as follows:

For context, the Credit Suisse Hedge Fund Index appears to have significantly outperformed the HFRX Index over time.

133 In our judgment, Ms Mayr’s approach in choosing the alternative investment indices should have been preferred over Mr Morrey’s approach. Ms Mayr’s choice of the HFRX Index for the Soothsayer Accounts is more appropriate because it is the index specified in the Soothsayer mandate. This is unlike Mr Morrey’s choices of the HFRX ETF and the Credit Suisse Hedge Fund Index for the Soothsayer Accounts.

134 Even if we accept that Mr Morrey’s choice of the HFRX ETF may be a reasonable one for the Soothsayer Accounts as it is an ETF that essentially tracks the HFRX Index and “represents a circa 8% difference in performance over a 15 year time period”, Mr Morrey’s approach remains unsatisfactory. Pertinently, as Ms Mayr testified, Mr Morrey’s use of the HFRX ETF is problematic because it does not contain “data available for the entire period”. Therefore, Mr Morrey had to supplement it with the Credit Suisse Index, which, as we have pointed out, vastly outperformed the HFRX Index contained in the Soothsayer mandate. This methodology is less than ideal. If Mr Morrey wanted to supplement the missing data for the time periods the HFRX ETF did not cover, it makes more sense for him to have done so using the HFRX Index, which, according to him, was tracked by the HFRX ETF. In our view, it is not reasonable to employ the Credit Suisse Hedge Fund Index (which happened to vastly outperform the HFRX Index and presumably the HFRX ETF) instead. This is especially so since the HFRX Index was already specified in the Soothsayer mandates.

135 Moreover, although the Meadowsweet mandate did not specify a particular index to be used for the alternative investment benchmark (unlike the Soothsayer mandate), Ms Mayr’s approach should still have been preferred. There is no reason why the HFRX Index as used in the Soothsayer mandate should not be an appropriate alternative investment index for the Meadowsweet Accounts. Although Mr Morrey’s comments about the Credit Suisse Hedge Fund Index being produced and marketed by Credit Suisse itself goes towards showing that it may be a reasonable choice for an alternative investment index, it does not offer a legitimate reason to prefer the Credit Suisse Hedge Fund Index over the HFRX Index.

136 Accordingly, we hold that Ms Mayr’s choice of the alternative investment indices should have been adopted.

137 Similar to the selection of alternative investment indices, the Judgment appears to be silent on the issue regarding the geographical diversity for the equities indices in the Benchmark Portfolios.

138 The crux of the dispute here is whether, in seeking to achieve geographical diversity for the CS Life Meadowsweet Benchmark Portfolios, Mr Morrey’s use of three regional total return benchmarks or Ms Mayr’s use of one global benchmark is to be preferred. In our view, Ms Mayr’s approach should have been preferred over Mr Morrey’s approach.

139 Ms Mayr explained that Mr Morrey’s approach should not be accepted because the United States (“US”) turned out to be “the one market that happened to perform extremely well over the period” and Mr Morrey’s approach assumes that all the US dividends would be reinvested only in the US market, thus generating a better performance as compared to the use of a global benchmark. Ms Mayr further explained that although the difference between the two experts’ approaches is small in percentage terms, being “probably about just under half a per cent better”, the actual sums involved are “still significant enough”. From the available evidence, the actual sums involved appear to be as high as US$60m after the effects of the experts’ differences are compounded over a decade.

140 When Mr Morrey was given the opportunity to comment on Ms Mayr’s adoption of a different approach, and to address her criticisms of his approach, Mr Morrey simply replied that he had constructed what he “felt was a very globally-balanced benchmark through the use of these three indices” and that Ms Mayr was correct as well. He admitted that he “could have selected one index to do the same thing”. It is material that Mr Morrey did not controvert Ms Mayr’s criticisms, and that he also accepted that Ms Mayr’s approach would produce “almost exactly the same as [his] collection of three”. He was of the view that they would both “come to pretty much the same answer”.

141 In our judgment, Ms Mayr’s criticism of Mr Morrey’s approach appears to be valid. Even though the experts agree that the percentage difference is minor, this small percentage difference would translate into a material sum (of up to US$60m). As Ms Mayr’s criticism is uncontroverted, and Mr Morrey indeed agreed with Ms Mayr’s approach, the Judge ought to have preferred Ms Mayr’s approach.

142 We therefore hold that Ms Mayr’s choice of one global index should have been adopted to achieve geographical diversity for the CS Life Meadowsweet Accounts.

143 The Judge observed that the differences between the average rate of returns were in part due to the differences in Ms Mayr’s choice to use a price-only index, and Mr Morrey’s choice to use a total return index for equities. The total return index “reflects investment performance which will have had dividends and other distributions retained and reinvested”, while the price-only index “reflects capital growth without taking account of the reinvestment of dividends”. Ms Mayr and Mr Morrey both used the total return index for bonds and alternatives. After considering the experts’ views, the Judge preferred Mr Morrey’s approach.

144 In our judgment, we agree with the Judge’s adoption of Mr Morrey’s approach in adopting the total return indices for equities. There is no basis for CS Trust’s submission that the “standard version” of an equity index is price return and the “standard version” of a bond index is total return. Ms Mayr’s evidence does not support such an assertion by CS Trust. She simply stated that in searching for the indices, the default index she saw for equities was the price-only index, and the total return index was for all other asset classes. The reason for this is unknown and could be due to various reasons, such as the search settings of Ms Mayr’s Bloomberg terminal. Ms Mayr did not provide any expert evidence that such a “standard version” of indices exists. Neither has CS Trust adduced any evidence, even from Bloomberg itself (where these purported results were displayed) of the existence of such a “standard version” of indices.

145 Similarly, there is no merit in CS Trust’s submission that Ms Mayr “faithfully appl[ied] the standard version of the [indices] found in the mandates for equities and non-equities respectively”. From our understanding of her evidence, the mandates over the Meadowsweet Accounts did not have a stated price-only index. Ms Mayr had to use the price-only index which was stated in the Soothsayer mandates and extrapolated them to the Meadowsweet mandates based on her own judgment. The need to extrapolate the price-only index into the Meadowsweet mandates (where the mandates are known) shows that the price-only index can hardly be considered a “standard version” of the indices found in the mandates for equities and non-equities.

146 It is important that Mr Morrey’s approach in using the total return index for equities accords more with commercial reality. As the Judge observed, Mr Morrey explained that the total return index includes dividends and this accords with the reality that the return an “investor would reasonably expect to achieve would include dividends and would include reinvestment, the compound growth on their investments”. By using the price-only index instead of the total return index that accounts for the reinvestment of these dividends, the dividends would essentially “vanish into thin air” and are “nowhere in the return”. According to Mr Morrey, this would not be in the normal experience of an investor owning, sharing or holding “an ETF that’s tracking one of these indices”. Such investors “would benefit from the dividends” and “benefit from the compounding of those dividends”. Ms Mayr did not dispute that the price-only index would fail to include the underlying dividends of the shares and their reinvestment.

147 In our view, the limitation which Ms Mayr raised regarding the total return index causing the portfolio to become overweighted with equities does not show that the Judge erred in adopting Mr Morrey’s approach. This limitation is mitigated in the present case. Both experts used the total returns index for bonds and alternative investments. The dividends from bonds and alternative investments are thus taken into account and assumed to be reinvested, which is similar to the economic effect of using a total return index for equities. Even if such dividends from bonds and alternative investments are not able to keep up with the dividends from equities, they necessarily mitigate the extent to which the equities will become overweighted.

148 In our view, it also does not make any sense to have inconsistent treatment between the indices for bonds and alternative investments in contrast to equities. On Ms Mayr’s approach and following her logic about dividends (and their reinvestment) causing positions to become overweight in a portfolio – this must mean that the use of a total return index for bonds and alternative investments, and the use of a price-only index for equities, would likewise lead to bonds and alternative investments becoming overweight over time.

149 In the circumstances, we find that notwithstanding its limitations, the Judge correctly preferred Mr Morrey’s approach which accords more with commercial reality.

150 The Judgment appears to be silent on the reasons regarding Mr Morrey’s choice of a “high-growth” index.

151 The crux of this dispute is over Mr Morrey’s choice of using an inherently riskier “high growth” index for the Meadowsweet Accounts. CS Trust complains that by choosing a benchmark for high-growth equities which is inherently riskier than other equities, there is significant outperformance of the broad equities indices specified in the Soothsayer discretionary mandates.

152 In our judgment, the Judge correctly accepted Mr Morrey’s explanation of his choice of a “high growth” index for the Meadowsweet Accounts as reasonable in the circumstances. Mr Morrey’s evidence is that, for the Meadowsweet Benchmark Portfolios, his allocation consisted of 70% bonds, 20% equities and 10% alternative investments, similar to Ms Mayr’s asset allocation. According to Mr Morrey, such a portfolio with 70% bonds and 30% equities and alternative investments is not usually regarded as a medium risk portfolio. It would be considered lower risk due to the high allocation in bonds. Thus, he did not consider it appropriate to apply the indices found in the mandate for the Soothsayer Accounts (which have a completely different allocation: see above at [124]). As both experts agreed that the Meadowsweet Accounts were intended to be medium risk, Mr Morrey was thus of the view that it was appropriate to have a higher risk component for non-bond assets to supplement the conservative bond portfolio. This was what he did when selecting the riskier “high growth” index for the equities component of the Meadowsweet Accounts, and this was a common approach to “multi-asset class medium risk portfolios”.

153 In the circumstances, we find no reason to disturb the Judge’s decision.

154 The Judgment does not appear to have dealt specifically with the issue of fees for the alternative benchmark portfolios.

155 CS Trust complains that there was no justification for Mr Morrey to have “excluded any fees payable for portfolio management of assets in his alternative benchmark portfolios that were invested in ETFs”. This gives Mr Morrey’s Benchmark Portfolios extra returns which, when compounded over time, translates into significant benefits for the respondents. The respondents disagree and argue that there should be no fees payable because the Benchmark Portfolios were static portfolios that do not involve active management. The respondents claim that in any event, Mr Morrey’s figures already included fees. The ETFs included in Mr Morrey’s Benchmark Portfolios are reported to be net of fees.

156 In our judgment, the fact that the ETFs are reported to be net of fees is only relevant in a situation where a customer invests directly into the ETF, bypassing the use of a discretionary mandate. Mr Morrey agrees that he is assuming “that no fees are payable to any other active manager of the portfolio of ETFs”. This does not accord with the facts of the present case. As will be recalled, the experts were to assess what would have happened assuming that there was an active manager (ie, the “competent and professional portfolio manager” with whom the Trust Assets were to be entrusted) managing the Trust Assets. Hence, the fees for such an active portfolio manager must be included in the calculation of the returns for the Benchmark Portfolios. Failing to do so would be inconsistent with the factual premise. This must be so, even if the active portfolio manager should decide to invest in ETFs, which would already have another set of investment manager fees built into their returns. In fact, Mr Morrey also agrees that if he was to assume that there is an active manager, the additional fees might be 0.5%.

157 Moreover, in our view, Mr Morrey’s explanation about his return being from an entirely static portfolio (his Benchmark Portfolios) and not featuring any benefits of active management, misses the point. The experts may have built their Benchmark Portfolios on the assumption that it would be a static portfolio, but this does not change the fact that they were to assess what would have happened “if the Trust assets had not been affected by fraud and had been placed with a competent and professional portfolio manager as at the date of breach”. This being the premise of the quantification exercise, the fees ought to be included in Mr Morrey’s Benchmark Portfolios.

158 As such, the Judge should have accounted for and deducted the portfolio management fees from Mr Morrey’s various Benchmark Portfolios. As for the quantum of these fees to be deducted, the Judge observed that the experts had “agreed that 0.8% was an appropriate level of fees for the Soothsayer accounts”, and also found that for the Meadowsweet and CS Life Meadowsweet accounts “the fees at 0.5% are in line with the actual fees charged and should be applied”: Judgment at [666]–[668]. As CS Trust does not challenge this part of the Judge’s findings on appeal, the quantum of such fees to be deducted should be in line with the Judge’s findings.

159 In conclusion, we allow the appeal but only in part. The Judge’s award of US$742.73m in compensation to the respondents, and the methodology used by the experts to reach that sum remains: Costs Judgment at [3]–[10], save for the following adjustments:

(a) the hedge fund investments are to be excluded from the Whole Portfolio Model, save for the three categories of hedge funds which the respondents have identified to be included (see above at [97]–[108]);

(b) Ms Mayr’s asset allocation for the CS Life Meadowsweet Accounts is to be adopted in place of Mr Morrey’s asset allocation (see above at [119]–[128]);

(c) Ms Mayr’s choice of alternative investment indices is to be adopted in place of Mr Morrey’s choice of alternative investment indices for the Benchmark Portfolios (see above at [131]–[136]);

(d) Ms Mayr’s choice of one global index is to be adopted to achieve geographical diversity for the CS Life Meadowsweet Accounts (see above at [137]–[142]); and

(e) the portfolio management fees are to be accounted for and deducted from Mr Morrey’s various Benchmark Portfolios that were invested in ETFs, in line with the Judge’s findings on the quantum of such fees (see above at [154]–[158]).

160 The parties, with the assistance of their respective experts, are to work out the consequential adjustments to be made to the Judge’s award of US$742.73m to derive the revised award. In the event of any dispute over the quantum of the revised award, the parties are granted liberty to apply.

161 As for costs, both parties have asked for costs of S$120,000 for the appeal. We have dismissed the key points of CS Trust’s appeal, and have only allowed the appeal in part, in terms of the quantification of the compensation to be paid to the respondents. The respondents are therefore the more successful party. Accordingly, we award costs to the respondents fixed at $80,000 plus reasonable disbursements to be taxed if not agreed. The costs order below remains unchanged.