A new study by Stanford University and Singapore Management University found that Polymarket’s Bitcoin prediction markets were prone to millions of dollars’ worth of manipulation.
According to researchers, the platform’s five-minute price markets were easily influenced by big-money traders with large Bitcoin supplies. These individuals could move the price of Bitcoin through buying or selling their supply, allowing them to win prediction contracts.
Manipulation detected
Researchers analyzed approximately 16,000 executed Bitcoin contracts over two months at Polymarket, one of the largest prediction operators in the world.
Bitcoin prices were determined by several independent oracle providers at the exact expiration of the five-minute window chosen by users. Traders with large crypto supplies took advantage of that by allegedly making large trades in the final seconds of their five-minute window, helping them move Bitcoin’s price just before the market closed. This ultimately preserved the prediction market prices and still allowed them to profit.
The study turned up 821 suspected manipulators, who won around $8.2 million — an average of just about $9,987.82 per person.
A report detailing the findings also estimated that about $1.3 million was effectively transferred from everyday users to big-money whales who were single-handedly changing prices. Assuming this applies to all 821 suspects, that’s an average of $1,559.08 per person.
Flying under the radar
As part of the study, researchers found that trading volume at Binance, a leading cryptocurrency exchange, increased 3.9 times its usual rate during settlement windows. Prices often returned to their previous amounts just after windows closed, suggesting that owners bought back in as soon as they won their prediction contracts.
Many of these instances occurred during late-night and early-morning hours and weekends, when there were fewer active traders and markets were easier to manipulate. Polymarket’s settlement price matched Binance’s 85% of the time during the study period.
Researchers noted they could not prove that Polymarket’s prediction markets and Binance traders were influenced by the same individuals.
The answer to the problem
In the interest of preserving markets from manipulation by a small group of traders, researchers came up with several solutions.
First, they noted that the observed issue largely disappeared when prediction markets were extended from five to 15 minutes. They also suggested using a time-weighted average price (TWAP), which calculates the average price of an asset over a given time window.
As an example, Bitcoin costs about $63,000 at the time of writing. A surge to $63,050 in the final 30 seconds of a 15-minute window would render a TWAP of only $63,001.67.
Although Polymarket responded to the study by saying it did not believe that manipulation interfered with its prediction markets, it said it plans to add average-price settlement to its platform within the upcoming year.
Cutting down on nefarious interference is an obvious concern for prediction operators such as Polymarket, Kalshi, and others. Polymarket itself reached $3.5 billion in U.S. trading volume during June, and although a hefty chunk of that was driven by the FIFA World Cup, Bitcoin markets are also among the most popular offerings on the platform.
The Trade Handle Prediction Markets Take
As noted by researchers, the primary concern is that potential manipulation could extend beyond crypto prices to contracts tied to the S&P 500 and Nasdaq. Sudden movements in these prices would greatly affect ordinary investors and the overall economy, which could be used as a source of profit at prediction markets.