Has the prediction market become a 'market-making' tool? Suspected large-scale manipulation emerges in Polymarket's 5-minute Bitcoin contract

Article author: @Steve_4P

Source: DeepFlow Tech

Peter Drucker, the father of modern management, once said: 'The best way to predict the future is to create it.' If one can turn an imagined future into reality through their own efforts, that may be the most certain form of prediction. However, in prediction markets, this statement takes on a radically different meaning. When participants can alter outcomes through their own actions, a market designed to allocate capital toward possible futures may instead evolve into a mechanism that actively creates those futures.

Recently, prediction markets appear to have moved beyond mere information aggregation, with cases emerging where participants attempt to directly influence settlement outcomes. This article will examine the structural vulnerabilities of prediction markets and the dual nature of their underlying technology through concrete examples.

To understand this phenomenon, one must first grasp how Polymarket’s 5-minute Bitcoin contract works. The contract asks: at the end of a 5-minute window, is the Bitcoin price higher than at the start of the window? Participants can trade positions on two outcomes: 'up' or 'down.' A position on 'up' pays $1 if the settlement price is higher than the opening price, and $0 otherwise. Essentially, this contract is a binary bet, with its value entirely dependent on the direction of Bitcoin’s price movement over a fixed, short time interval.

This design creates a sharp settlement threshold. Rather than primarily capturing Bitcoin’s long-term trend, the outcome hinges on a single reference point at a specific moment. When Bitcoin’s price nears the opening price as settlement approaches, even minor fluctuations can determine whether the entire contract settles in favor of 'up' or 'down.'

Polymarket settles these contracts using external oracles rather than its own order book. This means the final outcome depends on the price reported by the oracle at the exact moment of settlement. When traders hold positions in this prediction market while simultaneously participating in the spot Bitcoin market, they may have an incentive to influence the price reflected by the oracle—thereby affecting settlement conditions—rather than merely making predictions.

A recent study identified a distinctive pattern following Polymarket’s launch of the 5-minute BTC contract: in the final seconds before settlement, order flow on Binance’s spot market surged dramatically, and prices often reversed sharply immediately after contract expiration. While brief price movements alone do not prove manipulation, this behavior is inconsistent with trading based on new information—which typically produces lasting effects—and instead resembles trading activity tightly clustered around the settlement clock.

The study proposes a straightforward mechanism: a trader could first buy an 'up' contract on Polymarket, then place orders in the spot market toward the end of the settlement window. If the resulting price movement is captured by the oracle’s reported price, the 'up' contract could settle in their favor. The reverse strategy would apply for a 'down' position. Crucially, the profit from the prediction market position could far exceed the cost of creating a temporary price move in the underlying market.

The study estimates that a small group of accounts earned approximately $8.2 million in profits during periods classified as potentially manipulative, while ordinary participants incurred losses of about $7.6 million. These figures should be interpreted cautiously—the research did not directly observe individual intent nor establish legal proof of manipulation. Instead, it relied on empirical patterns such as trade timing, post-settlement price reversals, and profit concentration. Collectively, these indicators strongly support the hypothesis of settlement manipulation.

Prediction markets are typically valued for their ability to aggregate dispersed information and form collective views about the future. However, this function breaks down when participants can influence the very outcomes they are trading on. Similar risks exist for events like elections, sports matches, public opinion polls, official statements, and even weather readings. When traders can directly or indirectly affect the outcome, a market designed to answer 'what might happen?' may instead reward those who ask, 'how can I make this happen?'

Mitigating this risk requires more prudent market design.

First, caution should be exercised with markets settled over extremely short timeframes. Five-minute Bitcoin contracts are susceptible to manipulation—even if the underlying price information itself is useful. When settlement occurs within a narrow window, minor temporary fluctuations near the contract’s end can be enough to determine the outcome. Longer-duration contracts are not entirely immune, but because their outcomes reflect a broader price discovery process, they are generally less exposed to the risk of short-term manipulation. Therefore, prediction market platforms should establish clear minimum contract durations based on the characteristics of the market and the liquidity of the underlying asset.

Second, platforms need to reconsider what types of information merit being turned into markets. Polymarket and Kalshi have drawn significant attention for reflecting social and economic information faster than traditional media or opinion polls. In principle, markets on topics such as inflation, elections, or policy decisions can provide useful signals by aggregating insights from diverse participants.

However, some markets appear to be driven more by trading volume and attention than by informational value. For example, Polymarket once launched a market on whether Federal Reserve Chair Jerome Powell would say 'Good Morning' in his Jackson Hole speech. That contract generated approximately $80,000 in trading volume but offered limited insight into monetary policy or broader economic prospects.

If prediction markets are to justify their existence by generating socially useful information, platforms will need clearer listing criteria. Not every event or piece of information needs to become a tradable product. While blockchain and tokenization enable a wide range of assets and outcomes to be traded, technical feasibility alone does not create social value.

Finally, oracle design must be regarded as central to market integrity. The five-minute BTC contract uses Chainlink price data for settlement. Chainlink states that its price feeds aggregate data from multiple exchanges and data providers. However, publicly available information makes it difficult to easily verify exactly which exchanges are included and how much weight each source carries.

The close relationship between Binance prices and Chainlink’s settlement price raises important questions. This does not prove that Binance directly determines the settlement outcome, but it does make it difficult to assess the extent to which a single exchange influences results and how diversified the oracle’s inputs truly are. In prediction markets, oracles are not merely technical data sources—they determine which outcome wins and how funds are allocated. If users cannot evaluate an oracle’s methodology, source composition, and resistance to manipulation, the market’s transparency and fairness may themselves be called into question.

Peter Drucker’s idea that 'the best way to predict the future is to create it' captures the value of innovation and execution. Yet in prediction markets, this same notion can become problematic—when participants are rewarded for influencing outcomes rather than predicting them.

The issue is not whether prediction markets should exist. The more critical questions are: Who can influence the outcome? How is settlement determined? And does the market design provide sufficient safeguards against manipulation?

Prediction markets should not be viewed through either blind optimism or blanket skepticism. Like any emerging technology, they must be evaluated through the lens of both their potential and their limitations. Their long-term value will depend on whether platforms can uphold markets’ role as tools for information discovery, rather than allowing them to devolve into mechanisms for strategic intervention.

1496 Views