By Alexandra End, Marlene Haas, and Greg Leonard[1]
Introduction
Market abuse litigation often centres on allegations by private plaintiffs and government entities that market participants manipulated key reference rates. The economic significance of these actions stems from the foundational nature of financial benchmarks. As one of the core pillars of the global financial markets architecture, financial benchmarks serve as the reference for the pricing of trillions of dollars in commercial contracts, provide the settlement mechanism for derivative markets, and supply the critical inputs required for the daily valuation of complex asset portfolios. Crucially, benchmark manipulation cases are rarely single, linear track. Instead, globally, these actions typically advance through two, often parallel, channels: (i) private antitrust litigation, and (ii) non-antitrust proceedings, which primarily consist of a mix of regulatory enforcement actions and fraud-based claims.
Within the private antitrust trajectory, United States (US) litigants predominantly rely on Section 1 of the Sherman Act to allege horizontal conspiracies among market competitors. Litigants in the European Union (EU) and United Kingdom (UK) look to Article 101 of the Treaty on the Functioning of the European Union (TFEU)[2] and the Chapter I Prohibition (Competition Act 1998),[3] respectively, to challenge anti-competitive agreements and concerted practices among market competitors.
The second track of non-antitrust proceedings is a complex, multi-jurisdictional web. The same underlying misconduct is simultaneously pursued by a host of regulatory bodies under a variety of statutes. In the US, this includes, among other pathways, the Department of Justice (DOJ) prosecuting wire fraud, the Commodity Futures Trading Commission (CFTC) leveraging the broad anti-manipulation provisions of the Commodity Exchange Act (CEA), and the Securities and Exchange Commission (SEC) enforcing securities laws.
In the UK, the Financial Conduct Authority (FCA), the Serious Fraud Office (SFO), and the Office of Gas and Electricity Markets (Ofgem) have served as the primary enforcers, armed with tools like the UK Market Abuse Regulation (UK MAR)[4] under the Financial Services and Markets Act. In the EU, the European Commission has pursued benchmark-fixing as an illegal cartel, while the EU’s Benchmarks Regulation (BMR)[5] and Market Abuse Regulation (EU MAR)[6] impose a forward-looking regulatory regime. Across the EU, national regulators may bring enforcement cases as well. For participants in these markets, overlapping jurisdictions mean that a single instance of alleged misconduct may trigger parallel investigations across the globe, each with a potentially different legal standard.
What complicates matters further is that each benchmark is unique in its calculation. Careful consideration therefore needs to be taken when evaluating benchmark manipulation and antitrust claims. The central challenge in these evaluations is distinguishing allegedly disruptive intent from legitimate economic activity. Many valid trading strategies, when viewed in isolation, can mimic the patterns of purportedly manipulative schemes. For instance, a firm with a large derivatives position tied to a month-end benchmark may engage in significant trading activity in the underlying market near the fixing period. While this could be construed as an attempt to artificially skew the benchmark price, this could also be a textbook case of hedging — a prudent risk management practice to protect against adverse price movements. As such, when evaluating claims related to market manipulation and trading conduct around benchmark-setting periods, consideration must be given to a trader’s or trading desk’s entire strategy including their risk profile and position, instead of focusing on individual trades.
In conjunction with evaluating trading strategies, risk profiles, and positions, trading patterns must be evaluated within the broader macroeconomic environment. A trader aggressively selling a currency futures contract might not be attempting to depress the price, but rather reacting rationally to a sudden central bank announcement or an economic data release. This context is necessary when evaluating trading strategies and trading intent.
The remainder of this article first provides an overview of financial benchmarks and their design characteristics, then discusses in more detail the factors that are important to consider when analysing market manipulation claims related to benchmarks. We consider these factors under three general categories: (i) trading requirements, (ii) characteristics of benchmarks, and (iii) characteristics of market(s) relevant to benchmarks. This is not meant to be an exhaustive list of factors nor are any of these factors to be viewed in isolation; in assessing market manipulation claims there may be layers of factors that need to be considered.
Overview of Benchmarks
In financial and commodities markets, benchmarks provide a yardstick against which value can be measured, performance assessed, risk managed, and contracts settled. From the interest rate on a home mortgage to the price of a barrel of oil, benchmarks are the architecture designed to allow for stable and transparent price references for trillions of dollars in transactions daily.
In financial markets, benchmarks serve as the primary tool for performance measurement and price setting by serving as reference prices. For investors, equity indices like the US’s Standard and Poor’s 500 (S&P 500), a stock market index tracking the stock performance of 500 leading companies listed on stock exchanges in the US, or the UK’s Financial Times Stock Exchange 100 Index (FTSE 100), representing the 100 most highly capitalised blue chip stocks listed on the London Stock Exchange, serve as critical benchmarks. These indices distil the performance of hundreds of individual stocks into a single number, allowing fund managers to gauge their own success and providing the basis for passive investment vehicles like index funds. In the debt markets, the role of benchmarks is even more direct. Interest rate benchmarks, such as the Secured Overnight Financing Rate (SOFR), serve as the reference for pricing countless financial products, including corporate loans, bonds, and derivatives. Without these benchmarks, lending and borrowing would face significantly higher transaction costs and reduced transparency, with no common, independently verifiable basis for setting rates.
Commodities markets are equally reliant on benchmarks for price discovery. A barrel of oil from West Texas is not identical to one from the North Sea, just as aluminium from one smelter may differ slightly from that of another. Benchmarks like West Texas Intermediate (WTI) and Brent Crude for oil, or the official prices from the London Metal Exchange (LME) for industrial metals, account for these differences and provide reference prices for a standard grade and delivery location. Producers, consumers, and traders then price their specific products at a premium or discount to this common benchmark (e.g., “Brent plus $0.50”). This system allows a global, decentralised market to price goods efficiently and provides the point of reference for futures contracts, which businesses use to hedge against price volatility.
Furthermore, benchmarks can help mitigate information asymmetries and act as vehicles to provide transparency to markets. This is particularly true for markets that are potentially fragmented such as foreign exchange (FX) markets which are largely over-the-counter (OTC) markets. OTC markets can be less transparent than exchange-traded markets as there is no centralised exchange through which market participants can readily observe the majority of trading activity in the market. In the case of FX markets, there are several benchmarks across different currencies that support greater price transparency. These include the WM/Reuters (WMR) benchmarks which are calculated for a range of currencies and products in FX markets. In this same vein, benchmarks also provide protection against information asymmetries between informed and uninformed market participants such as retail investors who tend to have less access to information (such as real-time data feeds) compared to more sophisticated participants such as institutional investors or brokers.[7]
A number of design choices underlie financial benchmarks. As discussed below, understanding benchmark design is important as these choices will significantly influence market activity around the benchmark. For example, some benchmarks are based on assessments or other information provided directly by market participants (submission-based benchmarks) while many other benchmarks are based on observed market activity such as transactions and orders (market-based benchmarks).[8]
Submission-based benchmarks are calculated based on certain inputs and assessments from market participants such as quotes, estimates, or internal assessments rather than based directly on observed market activity such as actual transactions. Within the broader category of submission-based benchmarks, there have historically been two sub-categories. The first category captures benchmarks that are based on estimates and quotes provided by market participants in the form of surveys. For example, the London Inter-Bank Offered Rate (LIBOR) was a survey-based benchmark. Part of the reason it relied on survey inputs was due to the limited number of transactions that could be used to inform the benchmark. Rather than being based on a small number of interbank transactions, it was instead based on a survey of large banks which asked them the rate at which they estimated that they could hypothetically borrow short-term from other banks.[9] However, due to methodological and manipulation concerns, LIBOR was phased out and has been replaced by the Sterling Overnight Index Average (SONIA) in the UK and by SOFR in the US, both market-based benchmarks. More generally, however, given these concerns, there has been movement away from survey-based benchmarks to those based on confirmed trades, bids, and offers, but where the index reporting agency uses an internal assessment approach (usually based on a published methodology) to determine the benchmark value. This is the second category of benchmarks within the submission-based benchmarks. For example, the index reporting agency might consider all observed transactions and quotes but then prioritise the most liquid and relevant market activity and/or filter out data they consider to be irrelevant or anomalous. Examples of this type of benchmark include benchmarks published by agencies like S&P Global Platts (e.g., Platts Dated Brent or Platts Japan-Korea Marker) and Argus Media (e.g., gasoline, diesel, jet fuel).[10]
Market-based benchmarks on the other hand rely on activity in the market without any internal assessment by a price reporting agency. Typically, the body — such as an exchange — calculating the benchmark will determine a specific window of time in which they will evaluate market activity for the purposes of the benchmark. Other choices may include determining which transactions or orders to consider and whether outliers will be included. A benchmark provider may also consider whether activity is weighted in a certain way for the purposes of calculating the benchmark. As noted, SONIA, the LIBOR replacement, is a market-based benchmark. To determine SONIA, the Bank of England applies a calculation methodology based on eligible pound-denominated deposit transactions meeting certain requirements including, for example, being executed between 0:00 and 18:00 London time and settled on the same day.[11]
A useful framework for evaluating benchmark design — and for interpreting trading patterns around benchmark-setting periods — comprises three quality dimensions: representativeness (does the benchmark accurately reflect the underlying market it purports to measure?), attainability (can participants actually transact at or near the published benchmark price, or does the benchmark deviate systematically from executable prices?), and robustness (is the benchmark resistant to market shocks and manipulation by individual participants?). Benchmark designs have different trade-offs between these dimensions. A wider fixing window, for example, can improve robustness by making it more costly to sustain price pressure across a longer period, but may reduce attainability in that a single trade’s execution price may deviate from the published benchmark (e.g., during a period of sustained volatility in the benchmark period). This tension shapes how legitimate trading patterns should be interpreted; apparent deviations between a trader’s execution price and the published benchmark can be a consequence of the benchmark’s design, not evidence of distortion or misconduct. This framework and relevant examples are discussed in more detail throughout the remainder of this article.
Factors to Consider When Analysing Market Manipulation Claims Related to Benchmarks
Factors Related to Trading Requirements
Risk Management: Hedging is a risk management strategy designed to offset potential losses in a portfolio. A classic example involves a large energy producer, like an oil exploration and production company, with a significant volume of WTI crude oil scheduled for production and sale over the next year. If this company’s management anticipates a downturn in oil prices — perhaps due to forecasts of rising global inventories or slowing economic demand — they face the risk of a significant revenue shortfall on their physical sales. To mitigate this risk, the producer might sell futures contracts on the WTI index. If the price of WTI does indeed fall, the lower revenue the company receives from selling its physical barrels of oil will be at least partially offset by the gains on its short futures position. For example, Southwest Airlines established a reputation for this type of hedging through its fuel hedging strategy, most notably during the mid-2000s. While competitors suffered under volatile energy markets, Southwest’s internal trading team locked in long-term contracts years in advance, essentially treating fuel procurement as a primary competitive advantage rather than a simple operational cost. The most famous manifestation of this strategy occurred during the 2008 oil price spike. As crude oil climbed toward a record $147 per barrel, Southwest had already secured approximately 70% of its fuel at a price equivalent to roughly $51 per barrel. This hedge saved the company billions of dollars and accounted for the vast majority of its profitability during that time, allowing it to maintain lower fares and a stronger balance sheet than its struggling rivals.[12]
In both examples, the futures trades are not intended to profit from the large transactions in the underlying “physical” assets and any impact that these transactions might have on the benchmarks that the futures contracts are based on. Instead, the futures transactions are a necessary measure to effectively hedge the substantial dollar values of the firms’ underlying portfolios.
Thus, in market manipulation investigations that are centred on allegations of large amounts of trading during benchmark periods, it is important to differentiate legitimate hedging from illicit manipulation. A legitimate hedger has a genuine and potentially significant risk exposure to the price movements reflected in the benchmark that they are trying to mitigate. This type of assessment is all the more important in volatile markets, in which the need to adjust hedges and balance portfolios arises more frequently (an aspect that is magnified for options trading and companies implementing delta or other hedging goals), and can therefore lead to more frequent trading close to or during benchmark-setting periods.
A structurally distinct but equally important hedging dynamic arises in markets where benchmarks are determined by auction rather than by averaging transactions over a defined window. The London Bullion Market Association (LBMA) Gold Price, for example, is set twice daily through an iterative electronic auction. Successive rounds of the electronic auction run until buying and selling interest balance within a defined tolerance, at which point a single clearing price is published.[13] Market participants such as gold exchange-traded funds, mining companies managing production revenues, central banks, and derivatives counterparties routinely commit to transact “at the fix” — that is, to buy or sell gold at whatever price the auction will ultimately set. A dealer receiving such an order carries genuine price risk from the moment the order is received until the auction clears; in order to manage this price risk and be able to transact at the fix price, these dealers join the auction. However, common risk management practice — in particular for very large orders, to ensure full and best execution — is to begin hedging this exposure in the spot market, against internal positions, or elsewhere before the auction opens.
These pre-auction hedging trades can move the market price in the same direction as the underlying client order. This potential impact of pre-hedging trades has been observed more generally for different markets and benchmarks. For example, holding all else constant, a dealer pre-hedging a large client buy order purchases gold in the spot market ahead of the auction, which may lead to prices moving upward in the period immediately before the benchmark is set. Critically, this is precisely the pattern — directional trading in the period immediately before a benchmark-setting event — that a straightforward or surface-level forensic reading may treat as the hallmark of attempted manipulation. However, in such circumstances, this directional pattern can reflect the mechanism by which dealers manage price risk when accepting ‘at the fix’ orders — activity that, depending on the facts and market circumstances, may benefit clients through tighter pricing spreads. A robust assessment of trading conduct around an auction-based benchmark fixing should generally begin with an analysis of the dealer’s fix-order book and risk management strategies before characterising pre-auction directional activity as suspicious.
Optimal Trade Execution Strategies: A second area of regulatory focus is the timing of transactions, particularly large orders executed precisely at or moments before the benchmark’s determination. This activity is often cited as evidence of an intent to influence the price-setting mechanism. This perspective, however, does not always adequately account for the contractual obligations and fiduciary responsibilities of a significant segment of market participants, namely asset and portfolio managers.
Many investment vehicles (e.g. index funds, exchange-traded funds (ETFs)) are contractually mandated to replicate the performance of a specified benchmark. Put differently, the primary objective of such an investment vehicle is not beating a benchmark, but the precise replication of its return. For these entities, a central contractual and operational objective — and in many cases a fiduciary responsibility — is the minimisation of ‘tracking error’, the delta between the fund’s return and the benchmark’s return. To fulfil this mandate, managers must periodically rebalance their portfolios. This can, for example, be necessitated by variable cash flows (i.e. new investor capital inflows or redemptions) or by periodic changes in the index’s underlying constituents (i.e. index re-constitution). To closely replicate a designated benchmark’s return, the economic question becomes one of optimal trade execution.[14]
As an illustrative example: the S&P 500 index’s official value is determined by the closing prices of its 500 constituent stocks at the end of the trading day (e.g. 4:00 PM ET). If an investment vehicle such as an ETF on the S&P 500 received a large inflow of new investor cash during the day, then the manager must use this cash to buy stocks in the exact proportion of the S&P 500. To be able to precisely replicate a designated benchmark’s return, the manager of the fund will want to trade as close to the publishing of the closing value of the S&P 500 index as possible for the following reasons:
- Trading significantly prior to the determination of the benchmark (in this case significantly prior to the publishing of the constituent stocks’ closing value) exposes the fund to ‘timing risk’, whereby subsequent market volatility or news events between the trade’s execution and the benchmark’s publication can introduce tracking error.
- Trading after the determination of the benchmark means the fund executes at prices that may have diverged from the official benchmark price, introducing tracking error. For example, imagine a significant corporate news release just following market close. If the fund trades after the news event, the trade prices could significantly differ from the closing price.
Therefore, to fulfil their benchmark-tracking fiduciary duty, managers of investment vehicles typically execute rebalancing trades in the constituent stocks as close as operationally feasible to the timing of the determination of the constituent stocks’ closing prices (often in the stocks’ closing auctions) and in turn the benchmark’s determination. A forensic analysis of alleged benchmark manipulation needs to consider these contractual obligations and fiduciary responsibilities as part of understanding the overall trading pattern in a benchmark manipulation matter.
Best Execution Obligations: The considerations discussed above of market participants’ optimal trade execution strategies are also relevant in relation to best execution obligations for customer orders. In the US, best execution refers generally to a broker’s obligation to “seek the most favourable terms reasonably available under the circumstances for a customer’s transaction”.[15] For example, the SEC lists the following factors that brokers must consider when executing customer orders: “the opportunity to get a better price than what is currently quoted, the speed of execution, and the likelihood that the trade will be executed”.[16] Similar factors are listed by the National Futures Association, which is an industry-wide, self-regulatory organisation for the US derivatives industry.[17]
In both the UK and EU, regulations similarly stipulate an obligation to execute orders on the most favourable terms for the client as codified in Article 27 of MiFID II (EU) and COBS 11.2A of the FCA Handbook (UK).[18] Best execution may be measured across several factors including price, speed at which an order is executed, and whether the order is executed for the full size.
As discussed in more detail below with regards to characteristics of benchmark windows, the period around the benchmark-setting time may, in many markets and under typical conditions, represent a period of greater liquidity relative to other times of the trading day. Executing a customer order during a period of greater liquidity can minimise costs and optimise order execution for clients along the dimensions referred to above (price, speed, execution likelihood).
Clients may also prefer to trade at the benchmark (and may even do so directly in the form of Market-on-Close (MOC) trades as discussed below) or they may use a benchmark as a measure by which to judge the execution of their order. For example, S&P Global Platts publishes benchmarks (referred to as price assessments) for many commodity products, including both physical and derivatives, based on activity during a defined period of time each day during which designated market participants trade respective products (e.g., Brent and WTI crude futures, fuel oil derivatives, gasoline derivatives). This assessment window is often characterised by a period of greater market activity and therefore liquidity. Normal market activity that occurs directly before the end of the defined assessment window will typically have the most direct impact on the value published by Platts on the day.[19] As such, market participants, under either formal or informal best execution obligations, that trade on behalf of a customer may focus their trading just prior to the end of a benchmark window to benefit from the heightened liquidity as well as to execute at a price as close to the published benchmark value as possible. A forensic analysis should take these institutional details and requirements into account when assessing claims of allegedly distortive trading.
Factors Related to Benchmark Characteristics
Characteristics of Benchmark Calculations: Another factor to consider is the way in which benchmarks are calculated. A common misconception is that an index should match the price of the last reported trade or the simple average of a set of trades over a defined window. When an official benchmark price then deviates from individual trades reported in the market, claims of manipulation may arise. However, this perceived discrepancy is often rooted in the legitimate, complex characteristics of the benchmark’s calculation methodology itself.
For example, the ICE Brent Index is an average value calculated across an entire trading day to represent the broader market sentiment, volume, and depth. The index is derived from prices sampled at five fixed points throughout the trading day. It incorporates weighted values from several sources of Brent price information: (i) the highly liquid ICE Brent Futures contract, (ii) physical forward trades in the BFOETM (Brent-Forties-Oseberg-Ekofisk-Troll-Midland) market, and (iii) related instruments like Exchange for Physical (EFP) transactions.[20] If individual trades occur outside these sampling windows or represent a minor part of the total volume, the prices can naturally deviate from the final index, which is designed to smooth out intraday volatility. Even more stark deviations can occur if prices move significantly over the course of the day. For example, if new market information is released during the day causing a 10% increase in prices in the middle of the day, the overall index value may only rise 5% (due to the averaging effect over the course of the day), but at the end of the day individual trades will be 10% higher than at the start of the day and roughly 5% higher than the final index value. A forensic analysis should take these technical details into consideration when assessing the extent to which trades show hallmarks of manipulation.
A different structural complexity arises in benchmarks determined not by averaging transactions over a window, but by an iterative auction converging on a single clearing price. The LBMA Gold Price operates on this model, where successive auctions run until the imbalance between buying and selling interest falls within a defined tolerance. The published benchmark is therefore not an average of trades occurring during the session, but rather the single price at which the final round clears.[21] The mechanics of participation across rounds — entering orders, modifying quantities, or withdrawing interest as the price adjusts — are the normal instruments of price discovery in any iterative auction. As such, client orders that are entered, modified, or withdrawn can potentially cause large swings in bids between auction rounds. Forensic analysis of alleged distortion in an auction-based benchmark must be grounded in a detailed understanding of how the auction mechanism works and must distinguish between conduct that is legitimate participation in the auction process and conduct that genuinely falls outside those bounds.
Characteristics of Benchmark Windows: Trading during and close to benchmark periods can also depend on the unique characteristics of benchmark periods relative to other times of the trading day. One crucial aspect of benchmark-fixing periods is the endogenous concentration of liquidity that makes settlement windows the focal point of trading. Put differently, the concentration of trading volume and large trades during settlement windows is often not an anomaly but rather an endogenous and predictable feature of the microstructure of financial markets.[22] For example, S&P Global Platts’s assessment window discussed above is such a time period that is typically characterised by greater market activity and therefore liquidity.
This concentrated liquidity arises from several sources:[23]
- Indexation and Tracking Error: Large amounts of capital are managed by index funds (ETFs, mutual funds) whose primary mandate is to minimise tracking error relative to a benchmark. As discussed above, this necessitates transacting at or near the official benchmark price, creating a massive, non-discretionary source of liquidity.
- MOC/TAS Mechanisms: Exchanges have institutionalised this demand through trade at settlement (TAS) and market on close (MOC) order types. TAS order types allow market participants to trade at the yet-to-be-determined daily settlement price. For MOC orders, the exchange algorithmically aggregates and executes trades at the closing price, which often serves as an official reference price for portfolio valuation and derivative products.
- Derivative Expiration and Hedging: The settlement of options, futures, and swaps against these benchmarks necessitates hedging and positioning activity that is naturally concentrated around the fix.
For a large institutional investor with a significant order to execute (e.g. a portfolio rebalancing), this ‘liquidity pool’ is attractive. Large orders may more readily execute at times of elevated liquidity, and in many circumstances can be expected to face lower execution costs and reduced market impact compared to trading during less liquid periods. A reliable assessment of alleged disruptive trading patterns will account for the unique characteristics of benchmark periods relative to other times of the trading day by establishing the differences in liquidity and trading activity, and how these features play into the facts and circumstances of the matter at issue.
Characteristics of Markets Related to Benchmarks
Market Making Activity: Other potentially relevant factors to consider in any benchmark litigation matter are the characteristics of the market that underlie the benchmark. As an example, in many markets, the exchange that facilitates the market has a set of designated market makers. In general, market makers provide liquidity to markets by placing buy and sell orders, profiting from the small difference between the two prices (the ‘spread’). Designated market makers operate under obligations for liquidity provision. These obligations can include quoting requirements for a defined number of products or amount of time. Moreover, liquidity providers on exchanges such as Eurex can receive different incentives from the exchange by meeting certain requirements. This includes rebates in response to order book coverage of at least 85% or a larger rebate for coverage of greater than 90%.[24] These requirements and incentives, in turn, may have an impact on market activity that underlies the calculation of benchmarks.
An example of where designated market maker requirements and incentives may come into play are options on the EURO STOXX 50 index, which comprises the 50 largest and most actively traded companies in Europe.[25] Options on this index can be traded on Eurex and allow market participants to trade the EURO STOXX 50 index by entering a contract providing the option to buy or sell a specified amount based on the value of the EURO STOXX 50 on a future date. EURO STOXX 50 index options also underpin the calculation of the VSTOXX index, a forward-looking measure of European stock market volatility.[26] It is calculated based on the prices of options on the EURO STOXX 50 index, including analysing options with a range of strike prices and two specific expiry dates (around 30 days out), calculating an ‘implied volatility’ based on each option’s price, and combining these implied volatilities into a mathematically weighted value. Due to filtering mechanisms involved in selecting the strikes that will be included in the calculation, quoting activity by liquidity providers across EURO STOXX 50 index options strikes can affect the VSTOXX calculation.[27] Designated market maker obligations, while established for liquidity provision reasons for one particular market, can therefore indirectly impact other indices.
Analysing manipulation allegations for a benchmark such as the VSTOXX requires an understanding of both the methodological complexities of the benchmark and the intricacies of the market(s) related to the benchmark — in this case the EURO STOXX 50 market — to adequately assess activity in the market and the potential impact on a benchmark. Due to the relationship between quoting activity and the index calculation, increased activity could be perceived as manipulation, but it could instead simply relate to liquidity provision or quoting obligations. A forensic analysis needs to take all of these angles into account.
Interconnected Markets Underlying Benchmarks: In assessing benchmark manipulation claims, it is also important to consider interconnected markets and how trading in one product may impact activity in another connected product including around a benchmark. Energy derivative contracts are one example of where this may occur and, in particular, those that settle based on prices over a period of time, rather than a fixed point of time. Specifically, there are energy derivative contracts that settle against the average of the benchmark (often referred to as the price assessment) published by a Price Reporting Agency (e.g. Platts or Argus) for the underlying physical product over a period of time (e.g. a ship loading period, a week, or a month). In other words, if an energy derivative contract settles in the month of March, the final settlement price will be the average of published benchmark prices for the relevant underlying product for each business day in March. This means that on each day of the settlement period, that day’s published benchmark price is locked into the running average and can no longer be influenced by future price moves. As a result, a market participant’s exposure to price risk in that contract diminishes every day, since fewer days remain to affect the final average. Contracts such as the ICE Futures Dated Brent Futures, ICE Rotterdam Fuel Oil futures, and ICE Singapore Fuel Oil futures settle in this way.
Notably, this may have an impact on positions in calendar spread contracts. A futures calendar spread involves simultaneously buying and selling two futures contracts on the same underlying asset with different expiration dates to profit from changes in the price relationship (premium or discount) between them. When a spread contract comes up on its maturity, it is forced to settle, which may leave one of its ‘legs’ unhedged if that leg has yet to mature. Market participants that do not want to hold an unhedged exposure in a spread contract leg will have to trade out of this leg or perhaps buy spot/physical contracts. This latter activity can impact the price assessments of the leg that is expiring because the price assessment calculation typically takes into account the physical transactions.
Using a European Fuel Oil Futures August contract as an illustrative example: each day in August, as a portion of the settlement price for the contract is set, the exposure faced by market participants holding the contract goes down. However, a market participant that entered the August contract as part of a calendar spread will continue to have exposure to the second leg of the spread (e.g. a September contract). As a result, the market participant may wish to remove (or reduce) this second leg exposure by trading out of portions of its position in the September contract over the course of August. Alternatively, it may do so by buying the physical fuel oil to maintain a balanced position for risk management purposes. However, these transactions in the physical fuel oil might affect the August benchmark if trading occurs during the benchmark-setting period (e.g. because of the increased liquidity during this period).[28] Even so, trading activity observed in one contract may be an economically rational response to changes in activity or exposure in a related contract or related market. This is important to consider in assessing claims of market manipulation.
The Forensic Toolkit
To effectively substantiate these economic realities in a regulatory or litigation setting, defence teams must employ a highly sophisticated forensic toolkit. The necessary tools required may vary based on factors such as the alleged conduct or the markets involved, and typically include high-level statistics and data to understand patterns across time and markets as well as a granular dissection of the alleged incident. However, relying solely on aggregated exchange data or end-of-day pricing summaries may show only a partial picture of the market at the time of the incident under investigation. These aggregated data tend to obscure rapid, complex interactions that define how modern markets function. For example, sudden sharp price movements might be attributed to the timing of the trading activity under investigation. A granular analysis of the activity of all market participants at the relevant time, however, may reveal multiple unrelated parties trading in the same direction, which may be the ultimate cause of an observed price movement.
Economic experts and legal counsel should therefore consider analysing ‘depth of market data’, which consists of granular tick-by-tick order book reconstructions, including submitted, modified, and cancelled orders at the microsecond level. Examining the full depth of the market in the context of the benchmark architecture is an important tool for shedding light on the trader’s conduct and putting it in the proper context. Absent other sources of evidence that allow one to establish trading intent, this type of data and granular analysis is typically the most precise way to distinguish whether a specific trading pattern shows hallmarks of manipulative attempt to distort a benchmark or, instead, a legitimate, rational response to shifting liquidity and overarching risk management mandates. Tools that can facilitate this granular form of analysis include order book reconstruction and event study controls.
Order book reconstruction requires granular tick-by-tick data, tracking every submitted, modified, and cancelled order at the microsecond level across the full depth of the order book as well as the trader’s corresponding actions. These data capture the sequence and timing of order submissions in relation to the benchmark window — information that is invisible in aggregated daily or hourly data. The FCA’s study of the WM/Reuters 4pm fix provides an example of this methodology applied to a major benchmark, demonstrating how order-level data can be informative in the context of assessing intent and market impact that cannot be addressed from transaction records alone.[29]
Event study controls may be important to isolate benchmark-specific effects from market-wide movements that would have occurred regardless of any individual participant’s conduct. For example, a difference-in-differences approach — comparing price behaviour around the benchmark window to behaviour in control periods or control instruments — can provide the counterfactual needed to attribute price movements to specific participants and actions rather than to macroeconomic news, index-rebalancing flows, changes in liquidity provision, or other events.
Conclusion: The Necessity of Contextual Analysis
As benchmarks continue to serve as the foundation for pricing trillions of dollars in global assets, the regulatory and legal scrutiny surrounding them will remain intense. However, the path to distinguishing illicit manipulation from rational economic behaviour is rarely linear. The surface-level factors that are often raised as evidence of manipulation, such as high-volume trading during a fixing window or aggressive activity in related derivatives, are often indistinguishable from the footprints of legitimate trading strategies, hedging, and prudent risk management without a detailed quantitative, forensic analysis.
Whether it is a pension fund manager concentrating trades at the close to minimize tracking error, an airline hedging fuel costs against long-term volatility, or a trader balancing exposure across calendar spreads, market participants frequently execute legitimate strategies that could be mistaken for a manipulative scheme. Viewed in isolation, these trades can appear market distorting; viewed through the lens of a trader’s entire risk profile and broader market obligations, they may reveal themselves as legitimate responses to genuine risk management mandates or fiduciary obligations — rational participants in the price discovery process rather than disruptive forces.
Therefore, assessing market manipulation claims requires more than a forensic review of trade timing. It demands a holistic economic analysis that accounts for the specific design of the benchmark, the interconnected nature of financial markets, and the legitimate trading requirements of the participants involved. Failing to integrate these layers of context risks penalising the very liquidity and transparency that benchmarks were designed to foster. In the complex ecosystem of modern financial markets, context is not merely a mitigating factor — in many cases it can be a critical distinction between market abuse and prudent conduct.
References
[1] Alexandra End is a Senior Manager at Cornerstone Research. Marlene Haas is a Principal at Cornerstone Research. Greg Leonard is a Senior Vice President at Cornerstone Research. The views expressed in this article are solely those of the authors and do not necessarily represent the views of Cornerstone Research or any of their clients. The authors wish to thank Eléonore Viatte and Tim Wie of Cornerstone Research for their valuable research and comments.
[2] Consolidated Version of the Treaty on the Functioning of the European Union, 2008, OJ C115/47, Article 101.
[3] Competition Act 1998 (UK), Chapter 1.
[4] Regulation (EU) No 596/2014 of the European Parliament and of the Council of 16 April 2014 on market abuse (Market Abuse Regulation) as it forms part of domestic law by virtue of the European Union (Withdrawal) Act 2018, Articles 12 and 15. The benchmark-specific prohibition is Article 15 (prohibition on market manipulation) read with Article 12(1)(d), which expressly prohibits any conduct that manipulates the calculation of a benchmark.
[5] Regulation (EU) 2016/1011 of the European Parliament and of the Council of 8 June 2016 on indices used as benchmarks in financial instruments and financial contracts or to measure the performance of investment funds, Articles 11 and 14. The BMR contains no standalone manipulation prohibition; Articles 11 and 14 impose controls and reporting obligations on benchmark administrators and contributors, with the substantive prohibition on manipulation contained in EU MAR (see footnote 4).
[6] Regulation (EU) No 596/2014 of the European Parliament and of the Council of 16 April 2014 on market abuse (market abuse regulation), Articles 12 and 15. Article 12(1)(d) expressly defines manipulation of the calculation of a benchmark as a form of market manipulation; Article 15 contains the operative prohibition. The equivalent UK MAR provisions (see footnote 4) are identical in substance.
[7] Darrell Duffie, Piotr Dworczak, and Haoxiang Zhu, “Benchmarks in Search Markets”, Journal of Finance, 2017, 72(5), 1983–2044 (“Among other roles, benchmarks mitigate search frictions by lowering the informational asymmetry between dealers and their ‘buy-side’ customers.”).
[8] International Organization of Securities Commissions, “Principles for Financial Benchmarks”, July 2013, available at https://www.iosco.org/library/pubdocs/pdf/IOSCOPD415.pdf.
[9] The precise question put to each panel bank by the British Bankers’ Association was: “At what rate could you borrow funds, were you to do so by asking for and then accepting inter-bank offers in a reasonable market size just prior to 11 am?” The answer was explicitly a judgment-based estimate, not a report of an actual transaction. As Duffie and Stein (2015) explain, this design was in part an acknowledgment of reality: there were surprisingly few actual interbank loan transactions at the 3- and 6-month maturities most widely used as reference rates, making a purely transaction-based fixing impractical. See Darrell Duffie and Jeremy C Stein, “Reforming LIBOR and Other Financial Market Benchmarks”, 2015, Journal of Economic Perspectives, 191–212 (“Duffie and Stein (2015)”) (“However, there are surprisingly few actual loan transactions between banks that could be used to fix most of the IBORs, including those for the 3- and 6-month maturities that are so widely used as benchmark rates.”); HM Treasury, “The Wheatley Review of LIBOR: Initial Discussion Paper”, August 2012, available at https://assets.publishing.service.gov.uk/media/5a7b44f5e5274a319e77e2b5/condoc_wheatley_review.pdf.
[10] Platts, S&P Global Commodity Insights, “Platts Assessments Methodology Guide”, October 2025, available at https://www.spglobal.com/content/dam/spglobal/ci/en/documents/platts/en/our-methodology/methodology-specifications/platts-assessments-methodology-guide.pdf; Argus, “Argus Jet Fuel – Methodology and Specifications Guide”, April 2026, available at https://www.argusmedia.com/-/media/project/argusmedia/mainsite/english/documents-and-files/methodology/argus-jet-fuel.pdf.
[11] Bank of England, “SONIA key features and policies,” available at https://www.bankofengland.co.uk/markets/sonia-benchmark/sonia-key-features-and-policies.
[12] Note, the Southwest Airlines case is an illustrative example of how companies may hedge based on commercial needs. The hedging was not necessarily done during a benchmark window nor are the authors aware of any manipulation allegations that were brought in relation to Southwest’s hedging activity during this period. Deseret News, “Oil sets new trading record above $147 a barrel”, 11 July 2008, available at https://www.deseret.com/2008/7/11/20263305/oil-sets-new-trading-record-above-147-a-barrel/; Southwest, “The Southwest Jet Fuel Hedge Strategy”, available at https://southwest50.com/our-stories/the-southwest-jet-fuel-hedge-strategy/.
[13] LBMA, “LBMA Gold Price FAQs”, available at https://www.lbma.org.uk/prices-and-data/lbma-gold-price/lbma-gold-price (“LBMA Gold Price”).
[14] Almgren, Robert and Neil Chriss, “Optimal Execution of Portfolio Transactions”, Journal of Risk, (2001).
[15] SEC Release No. 34-37619A, “Order Execution Obligations”, September 12, 1996, available at https://www.sec.gov/files/rules/final/37619a.txt (“This duty of best execution requires a broker-dealer to seek the most favorable terms reasonably available under the circumstances for a customer’s transaction.”).
[16] SEC, “Best Execution”, May 9, 2011, available at https://www.sec.gov/answers/bestex.htm (“Some of the factors a broker must consider when seeking best execution of customers’ orders include: the opportunity to get a better price than what is currently quoted, the speed of execution, and the likelihood that the trade will be executed.”).
[17] National Futures Association, “9048 – NFA Compliance Rule 2-4: The best execution obligation of NFA members registered as broker-dealers under Section 15(b)(11) of the Securities Exchange Act of 1934”, July 31, 2002, available at https://www.nfa.futures.org/rulebooksql/rules.aspx?RuleID=9048&Section=9#.
[18] Specifically, COBS 11.2A of the FCA Handbook stipulates that “A firm must take all sufficient steps to obtain, when executing orders, the best possible results for its clients…”, and Article 27 of MiFID II notes that “Member States shall require that investment firms take all sufficient steps to obtain, when executing orders, the best possible result for their clients taking into account price, costs, speed, likelihood of execution and settlement, size, nature or any other consideration relevant to the execution of the order”. See Financial Conduct Authority, “COBS 11.2A Best execution – MiFID provisions”, October 23, 2025, available at https://handbook.fca.org.uk/handbook/cobs11/cobs11s9?timeline=true; European Securities and Markets Authority, “Article 27 Obligation to execute orders on terms most favourable to the client”, available at https://www.esma.europa.eu/publications-and-data/interactive-single-rulebook/mifid-ii/article-27-obligation-execute-orders.
[19] In the case of its futures assessments, Platts notes the following regarding its assessment process: “Platts examines traded levels, bid and offer levels prior to the close of regional MOCs, and employs the same methodological principles used in its physical assessments – repeatability and incrementability – when assessing the prevailing value of futures at the close in each region. Platts tracks the movements in the bids and the offers, the spread between the bids and the offers, and the execution of those trades. Furthermore, Platts analyzes the price trends leading up to the close, and considers only normal market activity in the assessment process. This is to ensure that the Platts assessment reflects a prevailing and representative value at the close, rather than an unusual trade occurring at that time, earlier or later”. Platts applies a similar requirement of “repeatability” in its assessments in the physical market. See S&P Global Commodity Insights, “Specifications Guide: Global Platts Forward Curve Products”, October 2024, available at https://www.spglobal.com/commodityinsights/plattscontent/_assets/_files/en/our-methodology/methodology-specifications/platts-forward-curve-oil.pdf
[20] Intercontinental Exchange, “The ICE Brent Index”, August 2025, available at https://www.ice.com/publicdocs/futures/ICE_Futures_Europe_Brent_Index.pdf; Intercontinental Exchange, “What are the Differences Between ICE Brent and NYMEX WTI Futures?”, available at https://www.ice.com/insights/energy/what-are-the-differences-between-ice-brent-and-nymex-wti-futures (“Until mid-2023, ‘Brent’ referred to a basket comprised of five different North Sea crudes (Brent, Forties, Oseberg, Ekofisk, and Troll, commonly referred to as BFOET). In June 2023, following June 2023 cargo deliveries, WTI Midland crude was added to the Brent basket and became part of the Brent complex.”).
[21] See LBMA Gold Price.
[22] The theoretical foundation for this result is provided by Admati and Pfleiderer (1988), who show that when traders can choose when to trade, each trader prefers to trade when others are already trading, creating a self-reinforcing concentration of liquidity in a single period. Admati and Pfleiderer (1988) formally characterize this outcome as the unique robust Nash equilibrium of the trading-time choice game: given that all other discretionary liquidity traders concentrate their activity in a single period, any individual trader’s best response is to do the same. Anat R Admati and Paul Pfleiderer, “A Theory of Intraday Patterns: Volume and Price Variability”, Review of Financial Studies, 1988, pp 3–40 (“Admati and Pfleiderer (1988)”) at p. 3 (“This article develops a theory in which concentrated-trading patterns arise endogenously as a result of the strategic behavior of liquidity traders and informed traders.”). Almgren and Chriss (2001) formalise the cost-minimisation problem facing large traders, demonstrating that execution cost is minimised by trading when market depth is greatest, which, at benchmark fixing windows, is precisely the period of peak liquidity. Robert Almgren and Neil Chriss, “Optimal Execution of Portfolio Transactions”, Journal of Risk, 2001.
[23] These sources will generally have varying degrees of relevance depending on the particular market of the benchmark.
[24] Eurex, “Eurex Circular 020/25 Attachment 1 — Product Specific Supplement for Equity Options and Selected Equity Index Options”, 2025, available at https://www.eurex.com/resource/blob/4318050/c60fdcf2357fc3caa2162928eb8ae11a/data/Eurex_Circular_020_25_en_Attach1.pdf.
[25] Stoxx, “EURO STOXX 50”, available at https://stoxx.com/index/sx5e/?factsheet=true.
[26] Stoxx, “EURO STOXX 50 Volatility (VSTOXX)”, available at https://stoxx.com/index/v2tx/.
[27] STOXX, “VSTOXX 101: Understanding Europe’s Volatility Benchmark “, October 2024, available at https://web.stoxx.com/hubfs/Whitepapers/2024%20Whitepapers/STOXX_WP_VSTOXX101_UnderstandingEuropesVolatilityBenchmark_102024.pdf (“First, a filter is applied that ignores any trade price, mid-quote or daily settlement price below 0.5 points, so as to exclude all negligible values.”).
[28] The case of a trader with a spot – futures hedge provides a similarly interesting example. In this case, as the futures contract enters its month of expiry and begins to price out, the trader will have to trade out of its spot position, if they do not want to hold unhedged spot. In this case, because the futures will price out each day based on the spot price benchmark, the trader will want to trade out of their spot position as close to the spot price benchmark as possible. They will want to trade as close as possible to the benchmark determination not simply because of the typical increased liquidity in this period, but also to mitigate risk of price divergences in their hedge.
[29] Fixing the Fix (2018), pp. 3 (“[D]ataset that allows us to identify the actions of individual traders.”), 8 (“We use proprietary order-book data from Thomson Reuters Matching (TRM) in our analysis, which contains all order-book events from the venue’s matching engine (new orders, cancellations, executions — and subsets therein: hidden orders, non-resting orders, etc.).”).