Matters that turned on the evidence — and what running it early changes

Case studies · what the evidence delivered

Matters that turned on the evidence — and what running it early changes.

These are worked matters — public cases, plus clearly-labelled anonymised composites where the real work is confidential — read for one thing: what the evidence delivered, and what it would have delivered run earlier. They are examples, not legal precedent.

Each was investigated, litigated, arbitrated or resolved across the UK, EU and Switzerland. Read together they make one practical argument for the people who commission this work — general counsel, heads of compliance, boards, and the firms that instruct us: the work that decides these matters is evidential, and almost all of it can be done earlier, faster, and at a fraction of the price.

Figures below are drawn from judgments, regulatory reports and reputable press — sources linked on each real case. Composites are illustrative, drawn from public patterns, and carry no confidential detail.

Headline stats

Civil & corporate fraud — in litigation

A dispute that hardens into a contested trial makes the evidence both the cost and the delay. Commissioned early — by a board, an audit committee or an acquirer — the same assembly is a fraction of the price.

ACL Netherlands BV v Lynch (Autonomy)

Civil fraud · England & Wales

English High Court — 93-day trial before Hildyard J

The parties

The matter, in brief. HP bought the UK software company Autonomy for $11bn, then alleged its accounts had been dressed up: hardware sold at a loss but booked as software revenue, sales pushed through resellers, reciprocal deals. HP sued the two executives personally for the difference between what it paid and what the business was really worth.

What needed to be done

1. Reconstruct hundreds of individual transactions to show how each was booked versus what it actually was.
2. Prove from emails and internal records that both executives knew.
3. Quantify what Autonomy was worth absent the misstatements — the damages case.
4. Defend every one of those transactions through 93 days of trial.

What the money bought. HP's costs alone ran past £120m, and bought: disclosure platforms and review teams working a review corpus the judge put in the millions of documents (the extracted trial bundle alone exceeded 36,000), forensic accountants re-deriving hundreds of deals, two full legal teams with experts and leading counsel through a 93-day trial, then separate liability, damages and appeal phases stretched over a decade.

Who paid — and when the money actually moved. HP carried its own costs for 13 years with no insurance recourse — W&I policies carve out exactly this kind of fraud. The court assessed HP's loss at ~£730m in 2025 and added a further ~£177m of interest in March 2026, against Hussain and the Lynch estate; following Lynch's death in 2024, actual collection remains uncertain. Judgment is not payment.

No hindsight required — knowable at the time. Analyst and short-seller doubts about Autonomy's revenue quality were published before the 2011 acquisition. The trial was decided on documents dated 2009–2011 — deal files, emails, audit queries — every one of which existed and was obtainable at close. Nothing material was discovered later; it was assembled later.

The clock — conduct to cash. Damages 2025 · collection still open

• Conduct → writedown (2009–12) — 3y
• Claim to trial (2012–19) — 7y
• Trial → damages (2019–25) — 6y
• Recovery (open) — 2y

Done pre-litigation. The case was decided on deal files that existed at acquisition. A focused revenue-quality review of the top deals pre-close — weeks of work — would have surfaced the same pattern for well under 1% of what the litigation cost.

Sources: Judgment — [2022] EWHC 1178 (Ch) (primary) · Bloomberg — HPE lost £730m on Autonomy, judge says (2025) (public)

Wirecard — Braun trial & the EY claims

Criminal fraud · Germany

Munich I Regional Court — criminal trial since Dec 2022; civil claims against EY

The parties

The matter, in brief. A DAX-30 payments company reported €1.9bn held in Asian escrow accounts. Journalists doubted it for years; when a special audit finally asked the banks directly in 2020, the money did not exist. The company collapsed within days, the CEO was arrested, and the criminal trial has run past a hundred sitting days.

What needed to be done

1. Confirm the escrow balances directly with the banks — past management.
2. Reconstruct the third-party acquiring business that supposedly earned them.
3. Trace what cash genuinely existed, and where it went.
4. Sustain a multi-year criminal trial and the parallel auditor litigation.

What the money bought. A KPMG special audit, an insolvency administrator reconstructing records across Asia, a criminal trial past 100 days — and EY defending civil claims over audits that accepted confirmationsOne block built on top of the block containing a transaction. More confirmations, more settled. at face value. The professional costs sit in the hundreds of millions; the losses at ~€20bn.

Who paid — and when the money actually moved. Shareholders lost roughly €20bn with little prospect of meaningful recovery; lenders wrote off billions. EY faces investor claims, but German auditor-liability caps mean any payout will be partial. Five years on, essentially nothing has flowed back.

No hindsight required — knowable at the time. The FT published detailed allegations from 2015; internal whistleblowers reported in 2016 and 2019. The escrow claim was checkable with a bank letter at any moment — when KPMG finally sent one in 2020, the answer took weeks.

The clock — conduct to cash. Collapse 2020 · verdict pending · recovery thin

• Allegations unresolved (2015–20) — 5y
• Collapse → charges (2020–22) — 2y
• Trial (2022– ) — 4y

Done pre-litigation. A four-week independent escrow confirmationOne block built on top of the block containing a transaction. More confirmations, more settled. in 2016 — direct to the banks, past management — ends the fraud four years and most of €20bn earlier.

Sources: ESMA — Fast Track Peer Review: Wirecard (primary) · FT — Dan McCrum on exposing Wirecard (public)

NMC Health — the administrators v EY

Civil fraud & audit negligence · England

English High Court — US$2.5bn claim against EY tried 2025, settled 2026; ADGM proceedings against the founders

The parties

The matter, in brief. A FTSE-100 healthcare group carried more than US$4bn of borrowing hidden across dozens of facilities that no single lender or auditor ever aggregated. A short-seller's report in December 2019 collapsed it within four months. The administrators then sued the auditors for the audits that missed it.

What needed to be done

1. Aggregate the true group debt across dozens of banks.
2. Trace related-party flows through the supply chain.
3. Establish what a competent audit would have detected, year by year.
4. Try a US$2.5bn negligence claim against a Big Four defendant.

What the money bought. Years of forensic reconstruction by the administrators, a multi-month High Court trial, and parallel ADGM litigation — estate professional costs in the hundreds of millions, all funded from what remained for creditors.

Who paid — and when the money actually moved. The EY claim was tried in 2025 and settled in February 2026 for a reported £105.5m — a fraction of the US$2.5bn claimed; the ADGM claims against the founders continue, and creditor distributions remain partial years after the collapse.

No hindsight required — knowable at the time. Muddy Waters built the collapse case from public filings in a few months; any syndicate lender aggregating group facilities would have seen covenant breaches years earlier. The gap was organisational, not informational.

The clock — conduct to cash. Collapse 2020 · EY trial 2025 · settled 2026

• Debt builds unseen (2012–19) — 7y
• Exposure → administration (2019–20) — 1y
• Claims → trial (2020–25) — 5y
• Trial → settlement (2025–26) — 1y

Done pre-litigation. Independent group-debt verification for any one lender — a facilities-mapping exercise measured in weeks — exposes the hole while it is still a lending problem rather than an insolvency.

Sources: FCA — censures NMC Health Plc for market abuse (2023) (primary) · Bloomberg Law — EY at the £2bn NMC trial (2025) (public)

Steinhoff International

Accounting fraud · Netherlands / Germany / South Africa

Global settlement (Dutch & South African schemes, 2022); Oldenburg criminal proceedings

The parties

The matter, in brief. Year-end profits were manufactured through "contribution" deals with counterparties secretly controlled by insiders, buried in an intercompany web spanning four jurisdictions. In December 2017 the auditors refused to sign; more than €10bn of market value evaporated within days.

What needed to be done

1. Prove the recurring year-end counterparties were not independent.
2. Unwind thousands of intercompany entries into a true group picture.
3. Broker one settlement across Dutch, German and South African claimant classes.
4. Pursue the individuals while the estate still had value.

What the money bought. PwC's forensic investigation ran roughly fifteen months; the settlement machinery consumed years of court time in two hemispheres; advisory costs ran to hundreds of millions — against a €1.4bn fund for claims far larger.

Who paid — and when the money actually moved. Shareholders recovered cents on the euro from the 2022 settlement; employees and pension funds bore the value destruction; the company itself ceased to exist in 2023. Individual accountability largely died with Jooste in 2024.

No hindsight required — knowable at the time. The year-end "contribution" deals recurred annually with the same opaque counterparties — registry-level independence checks would have exposed the circle at any point; German authorities had raided on related concerns as early as 2015.

The clock — conduct to cash. Collapse 2017 · settlement 2022 · company gone 2023

• Scheme years (2009–17) — 8y
• Collapse → forensics (2017–19) — 2y
• Settlement & wind-up (2019–23) — 4y

Done pre-litigation. Entity-relationship mapping on the recurring year-end counterparties — weeks of registry work — breaks the scheme in any of eight years.

Sources: FSCA — Markus Jooste penalty (primary) · PwC — Steinhoff forensic investigation overview (€6.5bn) (public)

Civil fraud — resolved in arbitration

Arbitration is private and quick, and an award is hard to unwind once made — which makes decision-ready evidence, assembled before the hearing, decisive.

Nigeria v P&ID

Fraud on the tribunal · England & Wales

English Commercial Court — s.68 Arbitration Act challenge before Knowles J

The parties

The matter, in brief. In 2010 Nigeria signed a 20-year gas-processing contract with P&ID, a company that never built anything. When the deal went unperformed, P&ID took Nigeria to arbitration for lost profits and won $6.6bn in 2017 — compounding with interest towards $11bn. Nigeria's last route out was to persuade an English court that the award itself had been obtained by fraud.

What needed to be done

1. Investigate how the 2010 contract was procured — tracingFollowing value through transfers, swaps and exchanges to show where it ended up. payments to ministry officials.
2. Prove P&ID had obtained Nigeria's privileged legal documents and used them throughout the arbitration.
3. Test P&ID's witnesses against the contemporaneous record until the story collapsed.
4. Carry all of it through an 8-week Commercial Court trial.

What the money bought. The ~£44m was Nigeria's own spend on the challenge, billed across 116 invoices from 2019 to 2024: English solicitors and counsel through the 8-week trial, investigators tracingFollowing value through transfers, swaps and exchanges to show where it ended up. bribe payments across three jurisdictions, and forensic analysis of the arbitration record proving the privileged-document leak. The court set aside the award and awarded Nigeria its costs — upheld by the Supreme Court in 2025 — but against a shell company, the real return was the $11bn that never had to be paid.

Who paid — and when the money actually moved. Nigeria avoided the $11bn payout entirely, but self-funded ~£44m of costs while the award accrued interest against it for six years. Costs were awarded against P&ID — a BVI shell backed by litigation funders, so recovery is likely minimal. Note the asymmetry: the claim was fundable and insurable; Nigeria's exposure was not.

No hindsight required — knowable at the time. At signing in 2010, P&ID had no plant, no staff, and no history of performing anything — facts confirmable then by ordinary corporate checks. The court also found P&ID held Nigeria's privileged documents throughout the arbitration; the fraud was running in plain sight of anyone who looked.

The clock — conduct to cash. Nothing paid — but $11bn at risk for 6 yrs

• Sham contract dormant (2010–17) — 7y
• Award + interest accruing (2017–19) — 2y
• Challenge → set aside (2019–23) — 4y

Done pre-litigation. A counterparty-capability check at signing — did P&ID have staff, plant, or any record of performance? — was a days-long exercise. It would have prevented the contract, the award, and a decade of sovereign risk.

Sources: Judgment — [2023] EWHC 2638 (Comm) (primary) · Spotlight on Corruption — Nigeria wins challenge to $11bn awards procured by fraud (public)

Stati v Kazakhstan

Fraud on the tribunal · Sweden / England / Belgium

SCC (Stockholm) award, 2013 — attacked as a fraud on the tribunal across three jurisdictions

The parties

The matter, in brief. In 2013 an SCC tribunal ordered Kazakhstan to pay the Stati parties damages in excess of US$500 million — around US$530m with interest — under the Energy Charter Treaty, most of it (US$199m) for an LPG plant valued on a single indicative bid. After the award, Kazakhstan showed that the plant's construction costs had been inflated through Perkwood, a supplier the Statis presented as an arm's-length third party but which was in fact their own related company. The state's route out was not to reopen the merits but to persuade national courts that the award itself was a fraud on the tribunal.

What needed to be done

1. Prove Perkwood was not independent — a related party concealed from the tribunal, not an arm's-length supplier.
2. Reconstruct the plant costs: equipment billed by Perkwood at ~US$115m (later increased) and a US$44m "management fee" with no contemporaneous record of any management performed.
3. Show the concealment made a difference — that the inflated costs fed the US$199m plant valuation the tribunal adopted.
4. Carry the fraud case through enforcement proceedings in England, Sweden and Belgium.

What the money bought. Years of forensic accounting and corporate-registry work to unwind the Perkwood relationship, then parallel enforcement litigation across three jurisdictions — the English Commercial Court (where Knowles J held there was a sufficient prima facie case of fraud to go to trial), the Swedish annulment courts, and the Belgian courts. The state carried its own costs while the award, and interest, ran against it.

Who paid — and when the money actually moved. No award money flowed to the Statis: enforcement was resisted jurisdiction by jurisdiction while a ~US$500m liability, compounding with interest, hung over the state for the better part of a decade. On 16 November 2021 the Brussels Court of Appeal refused exequatur, finding the award had been obtained by fraud — that the Statis deliberately presented Perkwood as a third party when it was affiliated. As with any fraud unwound at the enforcement stage, the real return was the payout that never had to be made, not a sum recovered.

No hindsight required — knowable at the time. Perkwood was a related company throughout; the Statis later conceded the relationship. Perkwood filed dormant accounts from 2006 to 2009 and was struck off in 2011 — facts sitting in the public register the whole time. The US$44m management fee had no supporting record of work done. Every strand the courts later relied on was in the contemporaneous documents; it was concealed from the tribunal, not created afterwards.

The clock — conduct to cash. Nothing paid — but ~$500m at risk for ~8 yrs

• Award issued, enforcement pursued (2013–15) — 2y
• English prima facie fraud finding (2015–17) — 2y
• Cross-border challenges to Brussels refusal (2017–21) — 4y

Done pre-litigation. A counterparty-independence and construction-cost review — registry-level entity mapping on the plant's key supplier, run before the investment costs were ever advanced to a tribunal — would have surfaced the Perkwood relationship and the unsupported billings at source. Commissioned by a general counsel, an audit committee or an acquirer, that assembly is the difference between defending an award and having one refused for fraud eight years later.

Sources: Stati v Kazakhstan [2017] EWHC 1348 (Comm) (primary) · Global Arbitration News (Baker McKenzie) (public)

Contax Partners Inc BVI v Kuwait Finance House

Fraud on the court · England & Wales

English Commercial Court — s.66 Arbitration Act 1996 enforcement set aside by Butcher J, [2024] EWHC 436 (Comm)

The parties

The matter, in brief. Contax obtained permission in August 2023 to enforce, as an English judgment, a purported Kuwaiti arbitral award said to arise from a roughly €53m dispute with Kuwait Finance House. On the strength of it, interim third-party debt orders issued against four London banks and a final order for £3,176,376.30 was made against Barclays, out of a total then standing at £70,634,614.04. The award was a fabrication: the court found there had been no arbitration agreement, no arbitration and no genuine award — passages had been copied, with modifications, from an entirely unrelated English judgment.

What needed to be done

1. Compare the "award" text against the public record — it had been taken from Manoukian v Société Générale de Banque au Liban SAL [2022] EWHC 669 (QB), Picken J.
2. Obtain confirmationOne block built on top of the block containing a transaction. More confirmations, more settled. from the Kuwaiti forum and courts that no such proceedings ever existed.
3. Establish the absence of any arbitration agreement between the parties.
4. Reverse the enforcement machinery — the s.66 order and the third-party debt orders already biting on London bank accounts.

What the money bought. The defence bought a document-level forensic comparison exposing the plagiarised text, official letters from the Kuwait Chamber of Commerce arbitration secretariat and the Kuwait Ministry of Justice confirming no proceedings existed, and Commercial Court time to set aside both the enforcement order and the third-party debt orders — before the £3.18m final order against Barclays was paid out.

Who paid — and when the money actually moved. Barclays had been ordered to pay £3,176,376.30 into the enforcement, with a total of £70,634,614.04 sought behind it; the fabrication was caught before that money left the banks. Once the award was set aside on 29 February 2024, the third-party debt orders fell away and no sum was paid to the claimant. The exposure was the London bank balances the orders had frozen in the interim.

No hindsight required — knowable at the time. Every fact the court relied on existed when enforcement was sought: the award's text already matched a published 2022 English judgment word for word, and the Kuwaiti forum and courts had no record of any case between the parties. A comparison against the public judgment database and a confirmationOne block built on top of the block containing a transaction. More confirmations, more settled. request to the named arbitral forum — the checks any enforcing party could run — would have exposed the forgery at the door.

The clock — conduct to cash. Order set aside 2024 · nothing paid out

• Fabrication to enforcement order (2023) — under 1y
• Interim & final debt orders on London banks (Aug–Oct 2023) — under 1y
• Challenge to set-aside (Oct 2023 – Feb 2024) — under 1y

Done pre-litigation. The defence here was itself the run-early work, only forced by an enforcement application rather than commissioned ahead of one. For a general counsel or board facing an unfamiliar cross-border "award", a rapid independent authentication review — matching the award text against the public case-law record and confirming the arbitration with the named forum — is days of work. Run before payment, it turns a £70m enforcement into a set-aside with nothing paid.

Sources: Judgment — [2024] EWHC 436 (Comm) (primary) · Law Society Gazette — £70m arbitration award was fabricated, judge finds (public)

Inflated-supply dispute, defended in arbitration

Illustrative — anonymised composite.

Civil fraud · confidential commercial arbitration (ICC/LCIA seat)

The matter, in brief. A group faces a nine-figure claim in confidential arbitration: a counterparty says a supply or services contract was validly performed and demands payment plus lost profit. The suspicion is the opposite — the counterparty had little real capacity, the deliverables were thin or never provided, and the contract was procured through an intermediary. Arbitration is private, quick, and hard to reopen once an award is made, so the evidence has to be assembled and decision-ready before the hearing, not after.

What needed to be done

1. A counterparty-capability and performance review — did the claimant have the staff, plant and record to perform what it billed?
2. Payment-and-intermediary tracingFollowing value through transfers, swaps and exchanges to show where it ended up. across the relevant jurisdictions.
3. Reconciliation of what was invoiced against what was actually delivered.
4. A witness-testable chronology built from the contemporaneous record.

Run early — who commissions it, and what it changes. Commissioned by the general counsel or the instructing firm at the first demand, a focused independent review settles the factual spine before positions and costs harden. Where a fraud on the tribunal is in issue, that assembly is the difference between resisting an award and paying one — the public Nigeria v P&ID matter above is the same problem, run too late.

Basis. A recurring public pattern (cf. P&ID); illustrative composite, no confidential detail.

Conduct, ethics & employment

Where an investigation meets an individual's conduct and their right to hold a regulated role, the evidence must satisfy two audiences at once — an employment tribunal and a regulator — and keep the investigation separate from the employment decision.

James (Jes) Staley v FCA

Fitness & propriety · United Kingdom

Upper Tribunal (Tax and Chancery Chamber), London — [2025] UKUT 00203 (TCC); FCA prohibition upheld

The parties

The matter, in brief. A bank's CEO approved a letter to the regulator, sent on 8 October 2019, describing his relationship with Jeffrey Epstein as not close and their last contact as well before he joined Barclays in December 2015. His own emails said otherwise — Epstein was among his "deepest" and "most cherished" friends, and contact ran up to the days before his October 2015 appointment was announced. The FCA prohibited him from senior roles and fined him; he referred both to the Upper Tribunal, which upheld the ban.

What needed to be done

1. Establish the true nature and timing of the relationship from the contemporaneous emails, not later accounts.
2. Set that record against the two representations in the 8 October 2019 letter, word by word.
3. Prove the state of mind — that the approval of a factually inaccurate letter was reckless and lacked integrity.
4. Sustain all of it through a contested reference against a former CEO's own leading counsel.

What the money bought. The FCA's case, and Staley's defence of it, turned on a small, fixed set of documents: emails between the two men from 2015–17 and the single letter of 8 October 2019. The Tribunal found breaches of the requirements to act with integrity (ICR 1), to be open with the regulator (ICR 3) and to disclose what the FCA would expect (SMCR 4), holding that Staley knew the letter's contents were inaccurate and recklessly misled the Authority — with no remorse.

Who paid — and when the money actually moved. The FCA proposed £1.8m. The Tribunal reduced the penalty to £1.1m — not through mitigation of conduct, but because after the Decision Notice Barclays declined to release deferred shares Staley could otherwise have received, so that value fell out of the income figure the penalty is calculated on. This sits atop the £642,430 the FCA and PRA had jointly fined him in 2018 over his response to the underlying whistleblower letter. The prohibition — the thing that ends a career in regulated finance — cost nothing to impose and is not measured in the penalty at all.

No hindsight required — knowable at the time. Nothing in the decisive evidence surfaced later. The emails existed from 2015–17; the letter was written in 2019. Every fact the Tribunal relied on was already in a document held by the bank or its former CEO when the letter was approved. The gap was not discovery — it was that the record was never read against the representation before the representation was made.

The clock — conduct to cash. Ban upheld 2025 · penalty fixed

• Relationship & emails (2015–17) — 2y
• Letter to the FCA → Decision Notice (2019–23) — 4y
• Reference → Tribunal upholds ban (2023–25) — 2y

Done pre-litigation. The whole matter was decided by reading two sets of the institution's own documents against each other before one was sent to a regulator. An independent review of a senior individual's contemporaneous record — commissioned by the general counsel or the board committee that signs off regulatory correspondence, before the letter goes out — is a days-long exercise. Run then, it either corrects the representation or stops it, and the prohibition, the £1.1m and six years of referral never begin.

Sources: FCA — Upper Tribunal upholds Jes Staley ban; [2025] UKUT 00203 (TCC) (primary) · FCA — Final Notice, James Edward Staley (2025) (public)

HBOS Reading & the Masterton disclosure

Whistleblower detriment on a real fraud · England & Wales

Southwark Crown Court (convictions 2017); FCA Final Notice, Bank of Scotland (2019); Lloyds/Masterton settlement (2018); Dobbs Review (ongoing)

The parties

The matter, in brief. Between 2003 and 2007, HBOS's Reading Impaired Assets team forced distressed small businesses to appoint a consultancy that then loaded them with debt and siphoned off the proceeds — a £245m scheme run through the bank's own lending. Sally Masterton, a senior HBOS risk officer, authored an internal report ("Project Lord Turnbull") alleging that the fraud had been concealed around the 2009 Lloyds takeover. She was subjected to detriment and left the bank; the fraud was later proven at Southwark Crown Court and the bank fined by the FCA.

What needed to be done

1. An independent test of Masterton's disclosure against the contemporaneous lending and credit records.
2. Protection of the reporter and the process, kept separate from any employment decision.
3. A finding stated to a standard that would satisfy a regulator and a criminal court — not just an internal committee.
4. A clear line between the conduct question (was there a fraud, was it reported) and the treatment of the person who raised it.

What the money bought. Operation Hornet ran roughly six years to convictions; the FCA fine on Bank of Scotland was £45.5m (£65m before a 30% early-settlement discount) for failing to report its suspicions; Lloyds estimated up to £115m in customer compensation; and the Dame Linda Dobbs Review — commissioned in 2017 to examine what was reported and when — is still open years later.

Who paid — and when the money actually moved. The fine flowed to the public purse in 2019, twelve years after the bank first identified suspicious conduct in 2007. Customer compensation followed the same slow curve. Masterton settled with Lloyds in two agreements; in 2018 the bank acknowledged she had "acted with integrity and in good faith" in raising her concerns. The money moved a decade after the disclosure did.

No hindsight required — knowable at the time. The fraud sat in the bank's own lending files, and a risk officer had written it up internally by 2013. Bank of Scotland had identified suspicious conduct in early 2007. Every element the criminal court and the FCA later relied on was contemporaneous; what was missing was not evidence but an independent test of the disclosure, taken seriously when it landed.

The clock — conduct to cash. Fine 2019 · Dobbs Review still open

• Conduct (2003–07) — 4y
• Suspicions unreported (2007–13) — 6y
• Convictions & fine (2013–19) — 6y
• Review & redress (2017– ) — 9y

Done pre-litigation. This is the whistleblower-detriment case built on a real fraud. Commissioned the moment the Turnbull report landed — by the general counsel, the board or the audit committee — an independent review tests the disclosure against the record, protects the reporter, and discharges the reporting duty on day one. The cost of not doing it: a £45.5m fine, up to £115m in redress, a constructive-dismissal claim, and a review still running years later.

Sources: FCA — Final Notice, Bank of Scotland (2019) (primary) · Statement from Sally Masterton on her Lloyds settlement (public)

Senior-executive misconduct & a regulated-function referral

Illustrative — anonymised composite.

Conduct investigation · regulated function (UK SM&CR context)

The matter, in brief. A financial institution receives a credible concern about a senior individual holding a regulated function — conduct that, if made out, bears on fitness and propriety and on the right to hold the role. The institution must investigate fairly (a process the individual can challenge at an employment tribunal) and, in the UK, may have to make a regulatory reference before that person can take up another regulated role elsewhere. The evidence has to answer to both audiences.

What needed to be done

1. A scoped, independent investigation with a documented methodology.
2. Secure, proportionate collection of the relevant communications and records under a defensible data-protection basis.
3. Findings stated to a standard that supports both the internal decision and any regulatory reference.
4. A clear line held between conduct findings and legal characterisation.

Run early — who commissions it, and what it changes. Commissioned by the general counsel, HR/legal, or a board committee, an independent specialist runs this faster, more cheaply and more focally than a general instruction, and keeps the conduct investigation separate from the employment decision — so the outcome is defensible whichever way it is later tested.

Basis. The intersection of investigations and employment/regulatory duty; illustrative composite, no specific matter.

A whistleblower's disclosure, independently tested

Illustrative — anonymised composite.

Ethics & reporting · protected disclosure

The matter, in brief. A protected disclosure — internal or to a regulator — alleges misconduct implicating people senior enough that an in-house investigation would not be seen as independent. The organisation's duty is to clarify what happened, act on it, and be able to show it did so properly. Doing that credibly usually means someone outside the reporting lines.

What needed to be done

1. An independent, documented review of the disclosure against the contemporaneous record.
2. Protection of the reporter and the process from conflict.
3. Findings that discharge the clarification duty whether or not the allegation is made out.

Run early — who commissions it, and what it changes. Commissioned the moment the disclosure lands, independent testing turns a whistleblower — or a headline — from a crisis into a documented, closed matter: the clarification duty discharged and evidenced.

Basis. Whistleblowing/reporting-obligation practice across the EU, UK and CH; illustrative composite, no specific matter.

Concealment & subterfuge

The hardest matters hide the conduct in structure — front entities, intermediaries, related parties. The evidence work is tracing and entity-mapping, and it can be run before a red flag becomes an insolvency or a prosecution.

Glencore — the SFO conviction and Swiss Art. 102

Corporate bribery · England & Switzerland

Southwark Crown Court (guilty plea, 2022); Swiss OAG resolution (2024)

The parties

The matter, in brief. The commodity trader paid bribes through intermediaries — including cash flown to West Africa — to secure oil cargoes. It pleaded guilty in London in 2022, settled with the US, and in 2024 the Swiss OAG resolved against it under Art. 102 SCC for failing to prevent bribery: the rare, live application of Swiss corporate criminal liability.

What needed to be done

1. Reconstruct agent payments and cash withdrawals across years of trading.
2. Connect cargo allocations to the payments that bought them.
3. Establish organisational failure for Art. 102 — what controls existed and why they did not bite.
4. Coordinate three prosecutors to a coherent global resolution.

What the money bought. Multi-year investigations on three tracks; the UK resolution alone cost £281m; the Swiss order was a CHF 2m fine plus a ~US$150m compensation claim, dominated by disgorged profit; ex-traders' individual trials continue — plus a group-wide compliance rebuild.

Who paid — and when the money actually moved. Penalties flowed to the UK, US and Swiss treasuries between 2022 and 2024; the states where the bribery occurred received comparatively little. The Swiss share was mostly disgorged profit — Art. 70/71 confiscation, not fine.

No hindsight required — knowable at the time. The agent commissions and cash withdrawals sat in the group's own records throughout. Payment-pattern analytics on intermediaries would have surfaced the routes internally years before three prosecutors did it externally.

The clock — conduct to cash. Probes 2019 · UK plea 2022 · Swiss order 2024

• Conduct (2007–18) — 11y
• Investigations (2019–22) — 3y
• Resolutions (2022–24) — 2y

Done pre-litigation. A standard third-party payment review — run internally, any year — finds the cash routes at a fraction of $1.5bn; Airbus showed the discount for finding it yourself and self-reporting.

Sources: SFO — Glencore case (gov.uk) (primary) · Swiss OAG — Glencore penalty order (2024) (primary)

1MDB / PetroSaudi — the Swiss limb

Criminal ML · Switzerland — the EDD case

Swiss Federal Criminal Court, Bellinzona; FINMA enforcement v BSI, Falcon, Coutts

The parties

The matter, in brief. Malaysia's sovereign wealth fund moved about US$1bn into a joint venture with PetroSaudi in 2009, most of it immediately diverted; in all, some US$1.8bn was misappropriated through the PetroSaudi limb between 2009 and 2015. The Swiss chapter: the money ran through Geneva accounts at small private banks that onboarded it over the objections of their own compliance staff. Switzerland prosecuted the PetroSaudi executives and FINMA dismantled the worst-offending bank.

What needed to be done

1. Reconstruct fund flows across dozens of banks and shell companies.
2. Prove the stated source-of-wealth was false against contemporaneous records.
3. Establish the banks' onboarding failures from their own files.
4. Sustain criminal proceedings across nine years and multiple jurisdictions.

What the money bought. Nine years of federal prosecutorial resources and mutual legal assistance across dozens of jurisdictions, plus FINMA enforcement teams working through the banks' own onboarding records — a public bill borne by the Swiss state. The banks paid in confiscated profits; BSI paid with its existence.

Who paid — and when the money actually moved. FINMA confiscated unlawful profits from BSI, Falcon and Coutts (2016–17). Goldman Sachs paid Malaysia US$2.5bn plus an asset guarantee in 2020; US DOJ recoveries returned further billions. The money came back a decade after it left — and the heaviest corporate price was BSI's: FINMA ordered it absorbed into EFG International and dissolved within a year, and Singapore's regulator shut its local unit outright.

No hindsight required — knowable at the time. The banks' own compliance functions raised concerns as these relationships ran — Coutts's 1MDB account opened in 2009 over what became documented internal objections, and a BSI manager warned in 2012 that staff were "implementing these transactions without really knowing what we are doing." The JV's absence of commercial substance was visible on the face of the structure. Every red flag FINMA later cited carries a contemporaneous date; none required hindsight, only authority.

The clock — conduct to cash. Main recoveries 2020–24 · 11+ yrs after the flows

• Onboarding → exposure (2009–15) — 6y
• FINMA + OAG build (2015–19) — 4y
• Proceedings → verdict (2019–24) — 5y
• Restitution (2020–24) — 4y

Done pre-litigation. This is the prolonged-EDD case: a genuine substance-and-source review of the JV at onboarding — the work compliance asked for and was overridden on — ends the Swiss chapter on day one. The cost of not doing it: a bank.

Sources: Federal Criminal Court — decisions (primary) · FINMA — BSI in serious breach of money-laundering rules (1MDB), 2016 (primary)

Parmalat — the Bonlat forged bank confirmation

Concealment fraud · Italy / United States

US District Court SDNY (SEC civil action & securities class action); Italian extraordinary administration (Bondi) with bank clawback litigation

The parties

The matter, in brief. A Milan-listed European food group reported €3.95bn (about $4.9bn) of cash and securities held by its Cayman subsidiary Bonlat in a Bank of America account in New York. The balance rested on a single bank confirmationOne block built on top of the block containing a transaction. More confirmations, more settled. — which the bank disavowed as forged when it was finally tested. The account and the assets did not exist. Parmalat disclosed the hole on 19 December 2003, filed for bankruptcy on 24 December 2003, and was declared insolvent days later, with more than €14bn of debt in default: Europe's largest corporate collapse.

What needed to be done

1. Confirm the Bonlat balance directly with Bank of America — past the forged document.
2. Unwind the offshore structure to show what genuine assets existed, and where the money had gone.
3. Prove the banks and auditors enabled or overlooked the concealment, to fund recovery.
4. Sustain parallel proceedings across two jurisdictions — SDNY and the Italian administration.

What the money bought. An SEC civil action and a US securities class action in New York; an Italian extraordinary administration reconstructing a decade of offshore entries; and clawback litigation the commissioner pressed against the banks and the auditor. The professional and estate costs ran for years across two continents; the hole stood above €14bn.

Who paid — and when the money actually moved. The recoveries came from the enablers, not the fraud's architects. As commissioner, Bondi settled with Deloitte's Italian arm for $149m (2007), Credit Suisse for €172.5m and UBS for €185m (€150m claim plus €35m damages, both 2008), and Bank of America for $100m (2009) — flowing between roughly 2007 and 2009, years after the collapse and a fraction of the loss. Founder Calisto Tanzi was convicted of fraudulent bankruptcy in 2010, sentenced to 18 years.

No hindsight required — knowable at the time. The €3.95bn balance was checkable with one letter to Bank of America at any moment; when the confirmationOne block built on top of the block containing a transaction. More confirmations, more settled. was finally tested, the bank disavowed it and the assets proved fictitious. Nothing was discovered later that was not addressable at any prior audit — only the direct confirmation was skipped.

The clock — conduct to cash. Collapse 2003 · recoveries 2007–09 · conviction 2010

• Falsification unquestioned (1997–2003) — 6y
• Collapse → first recoveries (2003–07) — 4y
• Bank & auditor settlements (2007–09) — 2y

Done pre-litigation. An independent third-party confirmationOne block built on top of the block containing a transaction. More confirmations, more settled. of the Bonlat balance — a bank letter, days of work — ends the fraud on the day it is sent, as it eventually did. Commissioned by an audit committee, a lender or an acquirer testing the offshore cash, it turns a €14bn insolvency into a control finding years earlier. This is the direct ancestor of the Wirecard escrow: the same forged-confirmation trick, run a generation before.

Sources: SEC Litigation Release LR-18803 (SEC v. Parmalat) (primary) · SWI swissinfo — Banks settle in Parmalat dispute (public)

Hin Leong Trading / Lim Oon Kuin ("OK Lim")

Trade-finance concealment · Singapore (cross-border lenders)

Singapore High Court — liquidators' civil consent judgment (Sept 2024); criminal conviction & appeal

The parties

The matter, in brief. For roughly a decade one of Asia's largest oil traders buried its derivatives losses — about US$808m — by overstating trading gains by as much as US$2.1bn, and financed the gap with fabricated paper: forged bills of lading, sales invoices and swap confirmationsOne block built on top of the block containing a transaction. More confirmations, more settled., fictitious sale-and-repurchase trades over its own storage terminal, and cargo that did not exist or was pledged to several banks at once. When crude prices collapsed in early 2020 the scheme unwound; the trader owed around US$3.5bn to some 23 lenders and went into judicial management under PwC. The liquidators sued the family; the state prosecuted the founder.

What needed to be done

1. Reconstruct the true derivatives book — the ~US$808m of losses masked by up to US$2.1bn of overstated gains.
2. Prove the financing documents were forged, and that cargo was non-existent, overstated, or pledged to multiple banks.
3. Reconcile inventory and receivables against what genuinely existed to size the ~US$3.5bn hole.
4. Carry a civil recovery to judgment while sustaining a criminal trial on sample charges.

What the money bought. A PwC judicial-management and liquidation reconstruction across dozens of banks, tracingFollowing value through transfers, swaps and exchanges to show where it ended up. fabricated bills of lading, sales invoices, swap confirmationsOne block built on top of the block containing a transaction. More confirmations, more settled. and inter-tank transfer certificates; years of civil claims by the liquidators and by HSBC; and a criminal trial on three representative charges drawn from 130. The professional and prosecutorial bill ran for years against a group whose assets were already gone.

Who paid — and when the money actually moved. In September 2024 the Singapore High Court approved consent judgments under which the Lim family agreed to pay the liquidators US$3.5bn — plus interest and costs, and a separate US$85.3m to HSBC — without admitting liability; the family was declared bankrupt in December 2024. Against a bankrupt estate the paper judgment is unlikely to be recovered in full. The lenders wrote down their exposure — HSBC's ~US$600m the largest — from 2020. The founder's liberty came later: convicted in 2024 of cheating HSBC of US$111.7m and abetting forgery, sentenced to 17.5 years, reduced to 13.5 on appeal in 2026.

No hindsight required — knowable at the time. The concealment lived in the same documents the banks financed against. A bill of lading pledged to two lenders, a sale-and-repurchase over the trader's own terminal, receivables "paid" by fabricated remittance advices — each is checkable against the counterparty or the registry at the moment of financing. PwC rebuilt the whole scheme from records that existed throughout; nothing was discovered that could not have been confirmed at drawdown.

The clock — conduct to cash. Consent judgment 2024 · recovery thin · sentence to 2026

• Concealment years (2010–19) — 9y
• Collapse → judicial management (2020) — 1y
• Civil claims → consent judgment (2020–24) — 4y
• Criminal trial → appeal (2020–26) — 6y

Done pre-litigation. This is a documentary-verification case: a lender or syndicate agent confirming bills of lading, inventory and receivables directly with counterparties and terminals — days of work per facility — exposes the double-pledging and phantom cargo before the next drawdown. Commissioned by a credit committee or an incoming risk head on the first anomaly, that check ends the vicious cycle while it is still a lending problem, not a US$3.5bn insolvency and a criminal trial.

Sources: SCMP — US$3.6bn payout ends OK Lim's civil cases (2024) (public) · Global Trade Review — Hin Leong's "vicious cycle" of trade-finance fraud (public)

Diversion through undisclosed related parties

Illustrative — anonymised composite.

Subterfuge & concealment · private-company scale

The matter, in brief. Value leaves a group through counterparties that look independent but are secretly connected — inflated purchases, "consulting" with no deliverable, sales to related buyers. It is the same structural trick the public Steinhoff and Glencore matters turned on, run at private-company scale. The conduct is hidden in the structure, so the evidence work is entity-mapping and flow-tracingFollowing value through transfers, swaps and exchanges to show where it ended up., not document review alone.

What needed to be done

1. Registry- and ownership-level mapping to expose the connections.
2. Payment-flow tracingFollowing value through transfers, swaps and exchanges to show where it ended up. across the related entities.
3. Reconciliation of what was paid against what was delivered.
4. A picture the board — or an acquirer — can act on.

Run early — who commissions it, and what it changes. Commissioned by an acquirer in diligence, a board on a red flag, or an incoming CFO, entity-relationship mapping breaks the circle while it is still a control problem — not an insolvency or a prosecution.

Basis. The recurring related-party concealment pattern (cf. Steinhoff, Glencore); illustrative composite, no confidential detail.

Money laundering, bribery & tax fraud

Cross-border schemes whose arithmetic fails on the face of the record — reconstructed years too late by prosecutors, when a single recurring check would have caught them at the start.

Mozambique v Privinvest (the "tuna bond" loans)

Civil fraud & bribery · England & Wales

English High Court — multi-month trial, judgment 2024

The parties

The matter, in brief. Three Mozambican state companies borrowed US$2bn for a tuna fleet and coastal-security projects, arranged through Credit Suisse's London desk and supplied by Privinvest — with secret state guarantees and about US$200m flowing back in bribes and kickbacks (roughly US$150m to Mozambican officials and US$50m to the arranging bankers). When the hidden debt surfaced in 2016, the country defaulted. Mozambique sued in London to escape the guarantees and recover from those who profited.

What needed to be done

1. Show the supply contracts were inflated — the vessels were worth far less than invoiced.
2. Trace the kickbacks to named bankers and officials.
3. Prove the guarantees were procured by bribery, and therefore unenforceable.
4. Marshal millions of documents of disclosure against a well-resourced defendant.

What the money bought. Years of e-disclosure across banks, brokers and shipyards; marine and project-economics experts to value what was actually delivered against what was invoiced; cross-border payment tracingFollowing value through transfers, swaps and exchanges to show where it ended up.; and a multi-month High Court trial — funded by one of the world's poorest states while it lived with the consequences of the debt.

Who paid — and when the money actually moved. Credit Suisse paid $475m to US and UK authorities in 2021 and forgave $200m of Mozambique's debt — money that went to regulators, not the victim state. Mozambique's own award against Privinvest (2024) remains in appeal and enforcement. Meanwhile the real cost — default, currency collapse, suspended IMF support — was paid by the Mozambican public from 2016 onward.

No hindsight required — knowable at the time. The Kroll independent audit (2017) reconstructed much of the scheme from the 2013–14 loan documents — and recorded how much had been withheld from it. Debt-service arithmetic against tuna-fleet economics failed on the face of the loan papers, and the guarantees breached IMF programme limits knowable at signature. Everything the court later relied on existed at credit approval.

The clock — conduct to cash. Judgment 2024 · enforcement still running

• Hidden loans (2013–16) — 3y
• Exposure → claim (2016–19) — 3y
• Proceedings → judgment (2019–24) — 5y
• Recovery (open) — 2y

Done pre-litigation. An independent use-of-proceeds and project-economics review at credit approval — standard work, done early — would have shown debt service was impossible and forced the guarantees into the open before a dollar moved.

Sources: Judgment — [2024] EWHC 1957 (Comm) (primary) · SEC — Credit Suisse to pay ~$475m over Mozambique bond offerings (2021) (public)

Cum-Ex — Berger, Shah and the dividend machine

Criminal tax fraud · Germany & Denmark

Bonn & Wiesbaden Regional Courts; Danish courts (Shah conviction, Dec 2024)

The parties

The matter, in brief. Trading rings circulated shares around dividend dates so that multiple parties could each reclaim the same withholding tax. In aggregate the arithmetic was impossible — more tax was reclaimed than ever withheld. It ran for years across European markets — Germany's loss alone is put in the tens of billions — before prosecutors caught up.

What needed to be done

1. Reconstruct who owned which share at the dividend instant, across chains built to obscure it.
2. Prove design and intent from structuring advice and marketing documents.
3. Extradite the architects — Berger from Switzerland.
4. Run rolling trials against banks and individuals, jurisdiction by jurisdiction.

What the money bought. A dedicated prosecution unit for over a decade in Germany; Danish civil and criminal proceedings across three continents against Shah's network; journalist consortia (the CumEx Files) reconstructing from leaked data what supervision never assembled.

Who paid — and when the money actually moved. Recovery is partial and rolling: Denmark has clawed back through civil judgments and confiscations; German courts ordered confiscation from banks (one, Maple Bank, collapsed under it). A decade on, most of it has not come back.

No hindsight required — knowable at the time. Custodians could see reclaim volumes exceeding withheld tax from the mid-2000s; Germany legislated against the domestic variant in 2012 after years of warnings. The CumEx Files, published in 2018, laid out what the data had always shown.

The clock — conduct to cash. Conduct 2006–12 · convictions from 2020 · recovery ongoing

• Scheme era (2006–12) — 6y
• Exposure → first trials (2012–19) — 7y
• Convictions rolling (2020– ) — 6y

Done pre-litigation. A reclaim-versus-withholding reconciliation — one recurring query across custodian data — flags the aggregate impossibility instantly. It was run years too late.

Sources: BGH — cum-ex is criminal tax evasion (1 StR 519/20, 2021) (primary) · CORRECTIV — The CumEx Files (2018) (public)

The process, anatomised

Where the years and the money go in a civil fraud claim.

The same broad machinery ran in every one of the litigated matters. Most of the elapsed time and cost sits in two places: assembling the evidence, and waiting for the money after judgment. Both are addressable before proceedings ever start.

93 court days, in real terms

The bill is not only in pounds.

Closer to the money, faster. Stages one and two — triage and evidence assembly — run in weeks when commissioned deliberately, before a dispute: as pre-transaction verification, enhanced due diligence at onboarding, or a rapid independent review on the first red flag. Everything after them exists only because they were skipped. Litigation funding and ATE insurance can carry a claim; nothing insures the years.

What running it early changes

The court did not find new facts. It funded the assembly of old ones.

In every matter above, the decisive material existed years before trial or award — in ledgers, escrow accounts, deal files, onboarding records and registries. What the parties paid for, at litigation rates, was its collection, organisation and interrogation.

A rapid independent review does that assembly when it is cheapest: before positions harden, before disclosure deadlines, before the meter runs at litigation speed — commissioned by the people who carry the risk: general counsel, heads of compliance, boards, acquirers, and the firms that instruct us.

Figures from public sources · verify before reliance. Composites are illustrative and anonymised — they describe recurring patterns, not specific matters.