In a stunning development that threatens to alter the landscape of online prediction markets, one recently-released study has uncovered an alarming pattern: automated “bot-like” betters have taken millions of dollars from ordinary users on the famed platform, Polymarket, by exploiting pricing inefficiencies and effectively creating a riskless profit to the detriment of others.
The study was based on analysis of over 86 million bets and was provided by researchers at IMDEA Networks Institute, concluded that automated traders exploited an oversight in the design of the prediction market. Initially, the study was aimed at analyzing normal (non-automated) trades on Polymarket, but scrutinization revealed in them high-frequency trading with automation (up to 400 trades per hour of 2 seconds per trade), resulting in approximately $40 million profit for a handful of users in a one-year period.

Understanding the ‘Arbitrage Advantage’

To grasp what’s happening, you need to understand arbitrage. It’s a term often used in finance to describe a strategy where a trader exploits a price difference for the same asset in different markets. On Polymarket, this isn’t about different platforms, but rather about mispriced shares within a single market. The platform allows market forces to continually set the odds, and sometimes, the probabilities of all possible outcomes don’t add up to a perfect 100%. This short misalignment presents an opportunity for informed traders — or more accurately their automated programs — to buy or sell share with a small, guaranteed profit.

Traditional betting sites, which set their own odds, work against this type of behavior. But as Polymarket has a decentralized, market-driven structure, they have found a unique loophole that a few have leveraged very well.

The Scale of the Bot-like Behavior

The research provides significant insight regarding this form of activity. The study indicated the top three wallets made over 10,200 bets, collectively, in a year, with a profit of an astonishing $4.2 million. The number of transactions, and the speed that the bets were coming in at, seemed to suggest the use of automated programs. Humans simply cannot place bets at such a fast rate using original methods.
Most arbitrage trades only resulted in small profits, mostly less than 1% to 5% in value for each trade, but the researchers also found cases with extreme market inefficiency. For instance, one user profited almost $59,000 from one significantly out of priced market.

Politics and Profits: A Surprising Connection

The study provides some interesting results with regard to markets with a political context, particularly the 2024 U.S. presidential election, as these markets were the most profitable, likely due to the high volume of trades and often quick, irrational changes in sentiment which can result in mispricing. Although there were more arbitrage opportunities in sporting markets, the potential profits were less in general.
Because of the high volatility of political events, along with the tremendous capital flow associated, these strategies can flourish.

Looking Ahead: The Future of Prediction Markets

These results have implications for the sustainability and fairness of the long-term health of prediction markets such as Polymarket. The research suggests that as these markets continue to grow in popularity, so will the prevalence of sophisticated, automated trading strategies. This could push casual bettors into a position of constant disadvantage relative to capitalized algorithmic players.
The researchers pointed out that prediction market arbitrage is still a largely unstudied field, with a greater amount of specialized strategies to come and more sophisticated forms of exploitations likely to follow. The ultimate problem for Polymarket and platforms like it will be to find a balance between their open market model and the very real though un-observable profit taking programs they will need to protect their user base from.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *