Investment Decisions · Fifth Order
One-Month Investment Performance
Models are compared through one-month marked-to-market paper-portfolio total return, including uninvested cash. All ten configurations share the same observation window, September 17–October 16, 2026, with the prior close included as the starting mark. Rankings follow total return, with SPY-relative return, drawdown, volatility and exposure available in the full leaderboard.
Model results
10 configurations- 1-0.9%
Nova Premier
- 2-2.1%
Gemini 3.1 Pro
- 3-4.1%
GLM 5.3
- 4-5.4%
Grok 4.6
- 5-5.6%
Mistral Large 4
- 6-5.7%
Claude Opus 5.5
- 7-6.7%
Qwen3 Max
- 8-7.0%
GPT 6 Sol
- 9-7.6%
DeepSeek V4 Pro
- 10-8.3%
Kimi K3
Numeric table · 10 configurations
| Rank | Model / setting | Score |
|---|---|---|
| 1 | -0.9% | |
| 2 | -2.1% | |
| 3 | -4.1% | |
| 4 | -5.4% | |
| 5 | -5.6% | |
| 6 | -5.7% | |
| 7 | -6.7% | |
| 8 | -7.0% | |
| 9 | -7.6% | |
| 10 | -8.3% |
Portfolio metrics
Same-window SPY return: -8.11%. Return ranks highest first; drawdown, volatility and exposure are reported separately.
| Model | Return | vs SPY (pp) | Max drawdown | Volatility / yr | Exposure |
|---|---|---|---|---|---|
| Nova Premier | -0.90% | 7.21 | -4.01% | 11.15% | 54.83% |
| Gemini 3.1 Pro | -2.14% | 5.97 | -8.86% | 32.93% | 96.03% |
| GLM 5.3 | -4.13% | 3.98 | -5.96% | 17.85% | 76.60% |
| Grok 4.6 | -5.39% | 2.72 | -9.70% | 28.26% | 90.38% |
| Mistral Large 4 | -5.60% | 2.50 | -5.77% | 18.78% | 82.34% |
| Claude Opus 5.5 | -5.70% | 2.41 | -5.84% | 16.45% | 87.28% |
| Qwen3 Max | -6.69% | 1.42 | -7.60% | 16.77% | 61.74% |
| GPT 6 Sol | -7.05% | 1.06 | -8.43% | 20.14% | 93.20% |
| DeepSeek V4 Pro | -7.60% | 0.50 | -8.70% | 35.88% | 98.39% |
| Kimi K3 | -8.29% | -0.19 | -10.40% | 31.55% | 95.71% |
Methodology
One month is 21 trading sessions, plus the September 17 prior-close baseline. Ending mark: October 16, 2026, 16:00 ET. Return is last marked portfolio value / first marked portfolio value − 1, including cash. Risk uses regular-session daily closes. Drawdown is the largest peak-to-trough decline; volatility is annualized using √252. SPY comparison is a percentage-point difference, not an additional ranking input.
