Agentic cryptocurrency trading on POLUSDT 2025-01-05 to 2026-01-05 15-minute bars (test)
52Number of TradesWebCryptoAgent (gemini-3-flash-preview)
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| WebCryptoAgent (gemini-3-flash-preview)Provider=gemini, Model=gemini-3-flash-preview, Memory=Off2026.01 | 52 | 26.92 | -48.1 | -48.19 | 59.51 | -0.8395 | -6.527 | -6.772 | 5,190.21 | |
| WebCryptoAgent (qwen-max)Provider=qwen, Model=qwen-max, Memory=Off2026.01 | 39 | 38.46 | -29.12 | -29.18 | 41.91 | -0.8093 | -1.801 | -1.538 | 7,088.23 | |
| WebCryptoAgent (gemini-3-flash-preview)Provider=gemini, Model=gemini-3-flash-preview, Memory=On2026.01 | 34 | 29.41 | -28.68 | -28.75 | 32.01 | -1.3348 | -5.613 | -6.332 | 7,131.94 | |
| WebCryptoAgent (qwen-max)Provider=qwen, Model=qwen-max, Memory=On2026.01 | 30 | 53.33 | -0.6 | -0.6 | 25.25 | 0.0949 | -1.545 | 0.916 | 9,940.49 | |
| WebCryptoAgent (gpt-5.2)Provider=openai, Model=gpt-5.2, Memory=Off2026.01 | 21 | 47.62 | -15.71 | -15.75 | 23.89 | -0.7855 | -1.778 | -0.765 | 8,429.4 | |
| WebCryptoAgent (gpt-5.2)Provider=openai, Model=gpt-5.2, Memory=On2026.01 | 13 | 38.46 | -6.41 | -6.43 | 14.37 | -0.6121 | -3.511 | -2.342 | 9,358.58 | |
| WebCryptoAgent (deepseek-chat)Provider=deepseek, Model=deepseek-chat, Memory=Off2026.01 | 10 | 60 | -0.16 | -0.16 | 17.9 | 0.0693 | 0.829 | 0.951 | 9,983.74 | |
| WebCryptoAgent (deepseek-chat)Provider=deepseek, Model=deepseek-chat, Memory=On2026.01 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 10,000 |