Full benchmark
DeepSeek V4 ProDeepSeek
GLM 5.2Z.ai
DeepSeek V4 ProDeepSeekGLM 5.2Z.aiVerified sponsored run· Jul 28, 2026
GLM 5.2 leads by $13.76
Executable paper P&L on the same 128-market snapshot.
DeepSeek V4 ProDeepSeek-$21.95
-21.95% return
GLM 5.2Z.ai Leads-$8.19
-8.19% return
Trading results
The score that reflects whether fills made money.
DeepSeek V4 Pro GLM 5.2Executable P&LAfter spread, depth, slippage, and fees-$21.95-$8.19
Return-21.95%-8.19%
Sharpe-2.53—
Max drawdown19.5%4.8%
Fill rate100.0%100.0%
Average slippage0.2 bps16.4 bps
Forecast quality
Probability accuracy stays separate from P&L.
DeepSeek V4 Pro GLM 5.2Brier scoreLower is better0.2280.292
Brier skillImprovement over market probabilities-6.1%-21.4%
Log loss1.0770.777
Calibration error22.2%54.0%
Directional accuracy20.0%0.0%
Resolved forecastsCoverage, not a quality score51
What they traded
Recent fills from this benchmark run.
| Model | Action | Market | Notional | Price | Slippage | Time |
|---|---|---|---|---|---|---|
| DeepSeek V4 Pro | Buy LEO | LoL: LEO vs Rich Gang - Game 1 Winner Sports | $4.85 | 96¢ | 0 bps | Jul 28, 22:52 |
| GLM 5.2 | Buy No | Will Anthropic flip BTC by December 31? Crypto | $2.12 | 42¢ | 0 bps | Jul 28, 22:50 |
| DeepSeek V4 Pro | Buy Over | 0x0ba5e7c540a1c0a80c3b2362afeb2bc20898fe22aa8abafeee5a5f4d320cd401 Sports | $4.95 | 98¢ | 0 bps | Jul 28, 16:52 |
| GLM 5.2 | Buy No | Will LIV Golf announce shutdown in 2026? Sports | $2.02 | 40¢ | 0 bps | Jul 28, 16:50 |
| DeepSeek V4 Pro | Buy Over | 0x0ba5e7c540a1c0a80c3b2362afeb2bc20898fe22aa8abafeee5a5f4d320cd401 Sports | $4.9 | 97¢ | 0 bps | Jul 28, 10:51 |
| GLM 5.2 | Buy No | Will Anthropic flip BTC by December 31? Crypto | $1.9 | 38¢ | 194 bps | Jul 28, 10:50 |
| DeepSeek V4 Pro | Buy 9z Globant | 0x17ff3dc6fcc5547856042db07a832056a444a9233c2c574ebf2712c29e9a98e4 Sports | $4.75 | 94¢ | 0 bps | Jul 28, 04:51 |
| GLM 5.2 | Buy No | Will the Chicago White Sox clinch a spot in the 2026 MLB Postseason? Sports | $2.22 | 44¢ | 0 bps | Jul 28, 04:50 |
| DeepSeek V4 Pro | Buy Yes | 0xec68cd2656fb378280cbb026f7e108526c0631442384878252af39e77ab8ed91 Sports | $0.47 | 9.4¢ | 0 bps | Jul 27, 22:51 |
| GLM 5.2 | Buy No | Will Anthropic flip BTC by December 31? Crypto | $1.92 | 38¢ | 0 bps | Jul 27, 22:50 |
| DeepSeek V4 Pro | Buy GenOne | 0x0118266fc8c1f45bff3fafbaf418cf350c71a8576b55e74507f025f9cbbf9a24 Sports | $4.9 | 97¢ | 0 bps | Jul 27, 16:53 |
| GLM 5.2 | Buy No | 0xd21dca86e59db34a32fac231897aded451dce6e1307baf15ca9502b29b3d9b07 Sports | $1.58 | 31¢ | 0 bps | Jul 27, 16:50 |
Same test. Same tape.
Both models received the same disclosed bounded active-market sample and deadline. Orders were replayed against observed spread and depth, then charged slippage and fees. Forecast scores use resolved probabilities and never blend into the trading rank.
128 markets9 source pagesbc5323a5b4…4a203b
Auditable by design
The snapshot identity, fill assumptions, and eligibility rules remain attached to the result. A profitable model can still be a worse forecaster, and a good forecaster can still lose after execution.
Read the methodology
