AI Trading Tournament Reveals Winning Strategy
The author set up a 4-strategy paper trading tournament to determine the most effective algorithmic trading approach. The strategies included RSI mean reversion, MACD crossover, Bollinger Band squeeze, and an AI-based confluence model.
Why it matters
This experiment demonstrates the value of empirical testing to determine the most effective algorithmic trading strategies, rather than relying on opinions or anecdotal evidence.
Key Points
- 1The author ran a 30-day paper trading tournament with 4 different algorithmic trading strategies
- 2Strategies included classic technical indicators like RSI and MACD, as well as a custom AI-based approach
- 3The tournament used a shared paper trading engine to evaluate the strategies on the same stock universe and starting capital
- 4The AI Confluence strategy, which required alignment of multiple signals, outperformed the other approaches
Details
The author, an algo trader, wanted to put common trading strategies to the test in a controlled experiment. They set up a paper trading tournament using the TradeSight platform, with four strategies competing over 30 days on a universe of 10 stocks. The strategies included classic technical indicators like RSI mean reversion and MACD crossover, as well as a Bollinger Band squeeze approach and the author's own AI Confluence model. The AI Confluence strategy required alignment of at least two out of three signals (RSI, MACD, 20-day SMA) to generate a buy or sell signal, aiming for higher win rate and lower trade count. By running the strategies in parallel with the same starting capital and universe, the author was able to objectively compare their performance without any look-ahead bias or curve-fitting.
No comments yet
Be the first to comment