The creator presents their first successful agentic AI trading strategy on Polymarket that uses AI-calculated fair value prices to place resting market-making orders, avoiding fees and slippage typical of taker bets, resulting in consistent modest profits. They emphasize the importance of extensive data collection, continuous model adjustment to mitigate risks like overfitting, and plan to expand their AI trading experiments to other markets while inviting viewers to follow their progress.
In this video, the creator shares their first successful agentic AI trading strategy implemented on Polymarket, focusing on a unique approach that leverages market making rather than the typical taker side betting. The strategy aims to avoid fees and slippage, common issues when placing taker bets on Polymarket, by placing resting orders on the maker side. The core idea is to use AI to calculate a “fair value price” for the market, which is then compared to the current market price to identify profitable opportunities where shares can be bought at a discount.
The strategy revolves around calculating the fair value price of a market, such as the probability of Bitcoin ending up or down in a five-minute window. Using AI models trained on extensive historical data, the creator determines a fair value price and sets resting orders at a price that is a few cents below this value to ensure a positive expected value. For example, if the market price is 55 cents but the AI model estimates the fair value at 51 cents, the resting order might be placed at 47 cents to capture value from impatient traders willing to sell quickly.
A significant challenge of this strategy is accurately calculating the fair value price, which requires collecting and analyzing large amounts of data. The creator has gathered over 144,000 fair value snapshots, 2,000 resolved markets, and 170 hours of live market data to refine their AI model. This extensive data collection helps ensure the model’s predictions are reliable enough to place profitable resting orders consistently. The strategy has shown promising results, with a steady profit of around $70 over several weeks of autonomous operation.
The creator also discusses the risks involved, such as overfitting the AI model to historical data, which could reduce its effectiveness in live trading. They emphasize the importance of continuous monitoring and adjusting parameters, like the price gap for resting orders, to maintain a stable Sharpe ratio and avoid significant drawdowns. The strategy is designed to be a long-running, low-cost AI agent that generates modest but consistent returns over time, rather than a get-rich-quick scheme.
Finally, the creator mentions plans to expand their AI trading experiments to other markets, such as options trading through Interactive Brokers, and invites viewers to follow their journey. They highlight that while the strategy is not financial advice, it demonstrates how combining AI with market making mechanics on platforms like Polymarket can create innovative trading approaches. The video concludes with a call to subscribe for future updates and deeper dives into similar AI-driven trading strategies.