Can AI truly replace human traders? Real trading competition data: humans currently hold the upper hand, with performance fluctuations 20 times less variable

Aster brought together 70 human traders and 30 AI models for a trading competition. Each participant started with $10,000, and the total prize pool is $200,000. The competition runs from December 9 to December 23, with live streaming on the Aster event page throughout. Currently, the top spot is held by renowned trader Ao Ying Capital, who tripled his position within a day. Statistics show that human traders consistently rank in the top 20 but also occupy the bottom 20. AI models are concentrated in the middle and lower tiers.

Regarding AI traders, the most conservative version, Qwen ( under Alibaba’s Qwen ), leads with a profit of $618. The most poorly performing models are Ernie 4.5 aggressive, Gemini 3 aggressive, and Claude Sonnet 4.5 in various versions. In terms of trading volume and number of positions, AI adopts a small-amount, diversified long strategy, while humans tend to close positions or concentrate on a single position. Human PNL standard deviation: 4875.0. AI PNL standard deviation: 224.8. This indicates that human performance fluctuations are over 20 times greater than AI.

Data shows: Humans have higher upper limits and lower lower limits

The current leader is the well-known trader Ao Ying. In fact, on the first day of the event, he was at the bottom of the leaderboard, seemingly catching a major trend. Starting with $10,000, he tripled his position, with an unrealized profit of $23,220.

(From $3,000 to $40 million, the highs and lows of legendary crypto trader Ao Ying Capital)

Second place goes to renowned Texas Hold’em player Wesley, who has already made a profit of $20,000. Third is @nextfckingthing@ representing the English-speaking community, currently with a profit of $17,870. Notably, his trading volume is $6.36 million, the second-highest on the leaderboard. Well-known figures in the Chinese community, such as Feng Wuxiang and veteran crypto traders, are also in the top ten. However, beyond rank 81, all are human traders, with even the Alert clubhouse traders having zeroed out. Overall, human traders tend to have higher upper limits and lower lower limits, while AI models are concentrated in the middle and lower tiers.

Image: Chain News

For AI traders, the most conservative version, Qwen ( under Alibaba, leads with a profit of $618, with Ernie 4.5 conservative and balanced versions in second and third place. The subsequent rankings include DeepSeek 3.1 balanced, ChatGPT 5 balanced, DeepSeek 3.1 aggressive, ChatGPT 4o conservative, and ChatGPT 4o radical. The worst performers are Ernie 4.5 aggressive, Gemini 3 aggressive, and Claude Sonnet 4.5 versions.

Human trading performance fluctuates over 20 times more than AI

Statistics show that human traders have an average trading volume of approximately $623,000 and an average of 0.83 positions. AI traders average about $99,000 in trading volume with 5.30 positions. While AI trades smaller amounts, they hold significantly more positions, indicating a small-amount, diversified long strategy, whereas humans tend to close positions or concentrate on a single position.

If we further categorize profits over $1,000 as profit group and losses over $1,000 as loss group, the loss group has an average of 187 trades, significantly higher than the profit group’s 132 trades. This may imply risks of overtrading or “dancing with the trend.” Traders with large gains or losses tend to have significantly higher trading counts than average. This reinforces previous observations: high activity correlates with high volatility. When further analyzing standard deviation:

Human PNL standard deviation: 4875.0 )extremely volatile(

AI PNL standard deviation: 224.8 )very stable(

This shows that human performance fluctuation exceeds AI by more than 20 times. This aligns with the tendency of humans to take on high risks for high returns (or suffer high losses), while AI strictly manages risk, pursuing stable, low-volatility strategies. As of the deadline, human traders have earned over $60,000, far ahead of the AI team’s $741.

Is AI really capable of replacing human traders? The live competition data suggests that humans are currently outperforming AI, with performance volatility 20 times greater. Originally published in Chain News ABMedia.