A recent AI model trading competition just concluded, and the results were unexpected.
**AI Lineup**: Qwen3 Max, DeepSeek, Claude, Grok, Gemini, and GPT-5—six major models, each allocated $10,000, only allowed to trade BTC, ETH, BNB, SOL, XRP, and DOGE futures.
**Key Findings**: Only 2 models made money; the rest lost. Qwen3 won by using high leverage and a concentrated strategy, DeepSeek won with low-frequency trading and risk control. Gemini and GPT-5's high-frequency operations and full short positions led straight to liquidation.
---
**The Human Team Joined In Too**: 10 traders played a round under the same rules.
The result? Even worse. Fewer than half made a profit, with the highest at just +21.84%, and the lowest wiped out completely. Common reasons for failure: - Too greedy, going all-in - No stop-loss, hoping for a rebound but losing more - Emotional trading, buying tops and selling bottoms - Holding positions overnight without monitoring, resulting in unexpected liquidation
The guy who named himself "all-in take a gamble" really did go all-in—and lost it all. Dropped out immediately.
---
**Three Major Lessons**:
1. **Low Frequency > High Frequency**: DeepSeek’s trend-following strategy crushed high-frequency chasing in terms of stability. Efficiency was much higher than frequent trading.
2. **Risk Control > Returns**: Qwen3 earned the most, but used 25x leverage and concentrated positions—one wrong move and it's game over. DeepSeek’s conservative strategy had lower risk but was more stable.
3. **AI ≠ God**: AI only looks at technicals (RSI, MACD), ignoring macro events and news. Humans are no different. In the end, it comes down to luck and mindset.
**The most ironic part**: The contestant who just held BTC and did nothing took third place.
AI may be more cold-blooded, but human greed, fear, and regret are often even more fatal.
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
# AI vs. Human Futures Trading Showdown: Who Won?
A recent AI model trading competition just concluded, and the results were unexpected.
**AI Lineup**: Qwen3 Max, DeepSeek, Claude, Grok, Gemini, and GPT-5—six major models, each allocated $10,000, only allowed to trade BTC, ETH, BNB, SOL, XRP, and DOGE futures.
**Final Results**:
- 🥇 Qwen3 Max: +22.3% ($12,231)
- 🥈 DeepSeek: +7.66% ($10,489)
- ❌ Claude: -30.81%
- ❌ Grok: -45.38%
- ❌ Gemini: -56.71%
- ❌ GPT-5: -62.66%
**Key Findings**: Only 2 models made money; the rest lost. Qwen3 won by using high leverage and a concentrated strategy, DeepSeek won with low-frequency trading and risk control. Gemini and GPT-5's high-frequency operations and full short positions led straight to liquidation.
---
**The Human Team Joined In Too**: 10 traders played a round under the same rules.
The result? Even worse. Fewer than half made a profit, with the highest at just +21.84%, and the lowest wiped out completely. Common reasons for failure:
- Too greedy, going all-in
- No stop-loss, hoping for a rebound but losing more
- Emotional trading, buying tops and selling bottoms
- Holding positions overnight without monitoring, resulting in unexpected liquidation
The guy who named himself "all-in take a gamble" really did go all-in—and lost it all. Dropped out immediately.
---
**Three Major Lessons**:
1. **Low Frequency > High Frequency**: DeepSeek’s trend-following strategy crushed high-frequency chasing in terms of stability. Efficiency was much higher than frequent trading.
2. **Risk Control > Returns**: Qwen3 earned the most, but used 25x leverage and concentrated positions—one wrong move and it's game over. DeepSeek’s conservative strategy had lower risk but was more stable.
3. **AI ≠ God**: AI only looks at technicals (RSI, MACD), ignoring macro events and news. Humans are no different. In the end, it comes down to luck and mindset.
**The most ironic part**: The contestant who just held BTC and did nothing took third place.
AI may be more cold-blooded, but human greed, fear, and regret are often even more fatal.