// analysis
FIFA and Predictstreet: who takes the trading risk?
An investigation of Predictstreet’s disputed activity, market-data methodology, liquidity and fees, with the limits of the public evidence made explicit.
A sports federation can sell visibility. A prediction market still needs participants willing to take different risks. That gap is at the heart of the questions surrounding ADI Predictstreet, which became the 2026 FIFA World Cup’s official prediction-market partner in April. FIFA’s partner page describes a form of collective intelligence: supporters’ views would become signals about future events. The value of that proposition depends on who actually controls the orders. 1, 13
On October 17, The Wall Street Journal reported that roughly $108 million of $167 million traded through late September came from wallets apparently linked to the platform. They reportedly traded almost entirely within the group and were funded by an address that also collected trading fees. For the period beginning in late August, the investigation attributed 99% of volume to 795 wallets, according to Legal Sports Report’s account. 2, 3
Gibraltar’s regulator offered a different assessment: a preliminary review reportedly found compliance with rules against self-trading and no consumer harm. That response belongs alongside the allegations. This documentary investigation did not obtain the complete transaction dataset or the detailed reasoning behind the review. The percentages therefore remain attributed reporting, not independently reproduced l0g measurements. 3
Public documents nevertheless make it possible to move beyond a clash of assertions. DefiLlama’s public collector, the surveillance commitments and Gibraltar’s rules describe three different things: recorded volume, an economically active market and a system of controls. Treating them as interchangeable gives marketing claims more authority than they deserve, and can make accusations against them equally overconfident.
Reported volumes cover different periods
The reported $108 million represents approximately 64.7% of the $167 million in cumulative transactions. Subtraction leaves $59 million outside the identified group. It tells us nothing further about the quality of that remainder: it is neither a customer count nor a certificate of independent demand. The 99% figure refers to the more recent window starting in late August. Applying it to the platform’s entire history would change the claim. 2, 3
Reconciling these figures with a blockchain dashboard would also require the authors’ exact volume convention. Their amounts are preserved as reported, without assuming equivalence with the DefiLlama aggregates examined below.
Figure 1. The amounts and recent share are attributed findings, not independently replicated here. The cumulative bar separates the flagged group from the unclassified remainder. The 99% uses a different denominator; the windows must not be merged. 2, 3
A wallet is a technical address. Several addresses can belong to one actor, while several customers may also use a common service infrastructure. The most flattering commercial denominator, a count of visible accounts, can therefore differ substantially from the number of independent decisions. Any serious measurement needs to identify its unit of observation before it describes a crowd.
The concern about wash trading is that transactions may be arranged to generate apparent activity without a corresponding economic transfer of risk between independent actors. Establishing that requires more than a long series of purchases and sales. A legitimate intermediary can buy, resell and finish with very little inventory too. Research on network-based detection explicitly recognises this difficulty. 8
The useful questions go beyond the size of a headline percentage. Who funds the positions? Who controls the orders? To which outside participants is risk transferred? Answers may reveal that a collection of addresses is one economic unit. Alternatively, they may show that accounts sharing technical infrastructure serve genuinely separate clients.
One dollar of payout can support several volume measures
Consider a fictional contract that pays one dollar if a team wins and zero otherwise. A buyer purchases 1,000 shares at $0.40 from a seller who already owns them. The buyer spends $400 before fees. The shares’ maximum conditional payout is $1,000. Both numbers describe the same transaction, from different perspectives.
The first is its actual traded value. The second is notional volume, expressed here using the shares’ possible payout. That difference is not evidence of manipulation. It becomes misleading when venues are compared without checking their definitions, or when notional volume is presented as new money deposited. Price and quantity need to accompany the counter. 6, 7
Another trap lies in the transaction records. One execution can produce several events: a record for one side of the trade and another for the other side. Adding their values indiscriminately can count the same economic transfer twice. The version of DefiLlama’s collector examined for this article groups events by emitting contract, transaction and batch position, then treats ordinary trades differently from the creation or redemption of complementary shares. 6
Figure 2. Fictional purchase of existing shares, before fees. The reviewed collector distinguishes traded value, notional volume and technical events. Complementary-share creation and redemption use different counting rules. 6
The code also exposes an important detail for anyone investigating counterparties: the exchange contract itself appears as the taker in some events. A network built on the assumption that every taker field identifies a customer would be misleading. An analyst must first reconstruct the economic meaning of the execution. This observation does not establish that the journalists’ researchers made that mistake; their full processing pipeline has not been reproduced here. 6
The public collector addresses a counting problem. The reviewed file does not classify beneficial owners or filter trading between accounts under common control. A correctly deduplicated series may still contain economically circular transactions. Conversely, mirrored technical events are not sufficient evidence of an intention to inflate activity. 6
The market maker and the closed loop
A new market needs someone willing to quote buying and selling prices before a large customer base exists. That is the market maker’s role. It may receive compensation for providing liquidity, automate its orders and seek to reduce its exposure quickly. From a distance, its work can look like a repetitive sequence of transactions in opposite directions.
Economic relationships help distinguish the cases. In a simple example, an intermediary buys 100 shares from Alice and sells 100 shares to Bernard. Its final inventory returns to its starting level, but Alice has reduced her exposure and Bernard has increased his. The intermediary has enabled a transfer between two separate decisions. In a closed circuit where A, B and C resell the same shares and all return to their original positions, the transaction counter rises without that sequence, by itself, demonstrating outside demand.
Figure 3. Arrows represent shares, not payments. Actors and relationships are fictional. A closed-loop pattern needs to be tested against control relationships and outside trading; it is not a finding of unlawful activity. 8
Research by Allen Sirolly, Hongyao Ma, Yash Kanoria and Rajiv Sethi proposes examining groups that trade predominantly with one another. Their study concerns Polymarket; its numerical findings cannot be transferred to Predictstreet. Its relevance here is methodological: it directs attention to how closed a trading network is, rather than relying on a turnover measure alone. 8
Even that test requires care. Under one possible explanation, a custody or pooled-funding service supplies several independent customers. Under another, common funding connects accounts to one decision-maker. These explanations need to be tested against identification records, mandates and order logs. A public graph produces useful clues and specific questions, but establishing the legal identity of whoever controls the trades may require non-public information.
The strongest defence of a market-making operation would therefore demonstrate its relationship with independent customers: executable quotes, trades with outsiders, exposure actually borne and verifiable execution conditions. Showing only that software exchanged tokens or paid fees would leave those issues unresolved.
Liquidity matters when the customer wants out
Volume adds up past transactions. Liquidity describes what a buyer or seller can obtain now, for a specified quantity. An active trading history can coexist with an almost empty order book when a customer needs to close a position. Conversely, an infrequently traded market can offer reasonable execution capacity if an intermediary genuinely commits to buying.
Consider another fictional example, unrelated to observed Predictstreet trading. A customer owns 1,000 shares. The best displayed bid is $0.50, but it covers only 100 shares. The next 200 shares can be sold at $0.45, and the remaining 700 at $0.30. Selling the entire position immediately into those orders produces $350 before fees, an average of $0.35 per share. Marking the whole position at the best bid would have suggested $500.
Figure 4. Step widths represent executed quantities and heights represent prices. Total proceeds are $350 before fees, assuming all orders remain available. No actual Predictstreet order book or customer loss is depicted.
The $150 difference, or 30% of that reference valuation, arises entirely from order-book depth in this model. It is not a measured Predictstreet loss or the result of a sporting wager. The example isolates execution cost: different quantities need not receive the price shown at the top of the screen.
Slippage measures the difference from an explicit reference price, excluding fees in this example. The price of getting out examines the same order-book mechanism in government bonds, using different units and a separate scenario.
A useful market-quality audit could publish exit costs for several position sizes, with timestamps and an explicit statement about fees. It should also test whether quotes remain available, because an offer withdrawn before a seller reaches it provides little practical protection. Volume would then have a companion measure directly relevant to customers, instead of carrying the entire burden of demonstrating success.
This distinction also matters when prices are interpreted as probabilities. The fictional contract bought at $0.40 earns $0.60 if the event occurs and loses $0.40 otherwise. Ignoring fees and funding costs, and reasoning in terms of expected monetary value, the break-even probability is 40%. That arithmetic does not make the price a representative forecast automatically. Enough independent views must be able to challenge it through executable orders.
Our analysis of Polymarket’s European strategy examines collateral, executable prices and the conditions governing access to these contracts.
Fees can circulate inside the same economic group
Trading fees may seem to provide firmer evidence of a commercial business. They look more tangible than a count of addresses: someone gave up value to complete a transaction. Yet the destination of that value, and the economic identity of whoever paid it, remain crucial.
In a teaching model, suppose there are 100 trades worth $100 each, with a hypothetical 1% fee charged to each side. The fee collector receives $200. If both trading accounts and the collector belong to one economic group, that $200 is also a cost borne by the group. Before network and operating costs, the internal transfer creates no external revenue for that perimeter. The interpretation changes if independent customers pay the fees.
Figure 5. Each side pays $100 in the model. The collector receives $200, offset by the same amount of expenses within the assumed common perimeter. This diagram attributes no tariff or account ownership to Predictstreet.
The example is not a description of Predictstreet’s tariff or wallet ownership. It explains why revenue analysis needs to trace the origin of payments rather than stop at their destination. Any separate compensation or subsidies supporting trading would also need to be considered where they exist: gross protocol receipts and customer-acquisition spending may travel through different channels.
The DefiLlama page consulted on October 20 displayed approximately $9.41 million in cumulative fees. Its collector treats trading fees as protocol revenue. That is a technical aggregate, not the audited profit of Predict Street Limited, the operator identified in the platform’s terms. It cannot establish how much was paid by economically independent customers. 6, 7, 11
It would be equally misleading to compare cumulative trading volume with collateral measured on one date and declare the platform insolvent. The same funds can support successive transactions. A solvency assessment requires outstanding obligations, the assets covering them and withdrawal rights. An inquiry into circularity requires relationships between counterparties. Combining the two would sacrifice the precision this case needs.
Announced controls and unresolved responsibilities
The launch announcement names Sumsub for identity and anti-money-laundering checks, Global Ledger for wallet analysis and Modulus for transaction surveillance. It also identifies sports-data suppliers used to settle contracts. That architecture is described by the company. Having named providers does not measure the effectiveness of the rules actually configured. 4
Each function answers a particular question. A correct sports result allows the right winning contract to be paid. Verified identification links an account to a customer file. Transaction surveillance looks for prohibited or abnormal behaviour. These checks can complement one another, but none removes the need to examine economic relationships between accounts. Without access to service mandates and alerts, attributing a failure to a named supplier would be premature.
Gibraltar’s regulations, effective July 13, expressly address wash trading, self-trading and conflicts associated with market making. They also preserve operator responsibility when functions are outsourced. Their scope and entity-specific authorisation conditions need to be checked in each case; the legislation is not a ruling on Predictstreet. 9
The public register consulted on October 20 lists Predict Street Limited under authorisation PMA01. It publishes no grant date, individual conditions or decision on the disputed trading. 14
Figure 6. An assessment chain proposed by l0g. Public code documents the first layer; the others require additional evidence. FIFA’s commitments and Gibraltar’s framework are not a ruling on the alleged transactions. 1, 6, 8, 9
Differences between the journalists’ analysis and the regulator’s preliminary position might turn on the definition of linked accounts, the treatment of particular transactions or private evidence reviewed. None of these explanations can be attributed to the regulator without documentation. A review becomes assessable when readers know its time window, applicable rules, ownership relationships and treatment of internal trades. Confidential customer identities need not necessarily prevent publication of a methodology and aggregated findings.
The absence of identified harm also addresses a different question from the quality of an activity indicator. A customer might have received the correct contractual payment while published volume led them to overestimate the market’s independent activity. That scenario establishes no actual damage in this case. It explains why settlement, trading integrity and fair disclosure warrant separate tests.
The commercial risk reaches the partner
For FIFA, the issue concerns the scope of its own public commitment. Its April announcement promised surveillance of suspicious trading and structured information sharing. The full commercial contract is absent from the documents reviewed, leaving audit rights, escalation duties and possible exit provisions unknown. Nothing in this dossier establishes that FIFA ordered, financed or knew about the disputed transactions. 1
The potential reputational mechanism is nevertheless straightforward. Association with a familiar institution can reduce a new user’s initial suspicion. If market quality is later challenged, questions return to the institution that lent its name: what did it verify, and what explanation can it now obtain? Buying visibility creates a relationship the public may interpret more broadly than the contract’s confidential clauses.
That relationship extends beyond one tournament. The announced project with DAZN would bring sports viewing and market participation closer together, subject to the required approvals. The more a transaction becomes part of a familiar experience, the more important it is to identify its execution venue and counterparties. Shared branding, however, does not establish that every partner service uses the same order book. 5
Territorial authorisation also needs to be distinguished from international visibility. On July 3, ESMA reiterated that event-contract classification depends on product characteristics and may trigger European binary-options restrictions or national gambling rules. The platform’s chosen label does not settle the question. That general statement is not an individual decision against Predictstreet. 10
Evidence needed to assess the disputed trading
The most useful evidence would reconcile recorded executions with economic counterparties in a reproducible way. It would begin with a dated set of contracts and blocks, explain how events are grouped, separate traded value from notional volume, and document the control relationships used to classify accounts. Results could then distinguish internal transactions, market making with outsiders and unclassified activity, rather than turning every uncertainty into an “organic” residual.
A second layer would test the service: what quantities can an independent customer actually buy or sell, at what cost and with what withdrawal options? Finally, a fee reconciliation would show how much money enters from outside the economic perimeter supplying liquidity. These are complementary tests of the commercial proposition, none of which requires publishing every participant’s personal information.
As of October 20, the documents establish the partnership, public surveillance commitments and the methodology behind one market counter. Complete account attribution and detailed reasoning for the regulatory review remain unresolved. The newspaper’s figures provide a specific reason to seek that evidence. For a platform selling collective intelligence, the decisive demonstration would show how independent decisions meet, take risk and exit on the advertised terms.
Scope of the investigation
Documentary investigation as of 20 October 2026. The disputed shares are attributed to the WSJ and LSR’s account, without independent replication by l0g. The regulator’s preliminary response is reported through those sources.
The DefiLlama collector was read at a pinned revision, without a security audit or full event reconstruction. The three numerical models and the network are teaching examples; they do not measure Predictstreet’s fees, losses or ownership links.
No interviews, contact with the parties or access to private contracts took place. External originals are linked, not archived in full. Wash trading alone does not establish money laundering or manipulation of sporting results.
The opening image is an AI-generated conceptual illustration; it does not depict an actual FIFA or Predictstreet device or installation.
Sources and documents
- ADI Predictstreet named official prediction market partner of the FIFA World Cup 2026
- Nearly All Volume on FIFA-Linked Prediction Market Is Circular Trading
- FIFA Prediction Market Partner Tied To $108M In Circular Trades: WSJ
- ADI Predictstreet Goes Live as the Countdown to the FIFA World Cup 2026 Begins
- DAZN and ADI Predictstreet Announce Exclusive Global Partnership to Launch Next-Generation Regulated Sports Prediction Markets
- Predictstreet collector: code at a pinned revision
- PredictStreet TVL, Fees, Revenue & Volume
- Network-Based Detection of Wash Trading
- Prediction Market Regulations 2026 · LN 2026/176
- ESMA reminds firms of existing rules and obligations under binary option measures amid growing popularity of prediction markets globally
- Terms & Conditions
- ADI Predictstreet Secures Gibraltar Regulatory Approval to Scale Its Data-Driven Forecasting Platform Globally
- ADI Predictstreet · partner page
- Prediction Market Authorisations
This analysis is not investment advice.
// cite this analysis
l0g, “FIFA and Predictstreet: who takes the trading risk?”, l0g.fr, published October 20, 2026, updated October 20, 2026, https://l0g.fr/en/analysis/fifa-predictstreet-trading-volume-risk/
$ cd ../analysis