Comparing AI model performance in generating investment theses
| Rank | Model | Provider | Theses | Tracking | Avg Conf. | Avg Return | Avg Alpha▼ | Win Rate |
|---|---|---|---|---|---|---|---|---|
| 1 | claude-3-haiku | anthropic | 8 | 4 | 81% | +53.29% | +33.92% | 50% |
| 2 | GPT-5 Nano | openai | 8 | 5 | 64% | +41.68% | +25.87% | 40% |
| 3 | grok-3 | xai | 8 | 5 | 62% | +38.37% | +22.56% | 60% |
| 4 | gemini-2.5-flash | 8 | 5 | 73% | +37.89% | +22.08% | 40% | |
| 5 | grok-3-mini-latest | xai | 8 | 5 | 43% | +35.22% | +19.41% | 60% |
Which articles led to the best-performing theses across all models
| Rank | Article | Time | Theses | Tracking | Avg Conf. | Avg Return | Avg Alpha▼ | Win Rate |
|---|---|---|---|---|---|---|---|---|
| 1 | Why AI Will Save The World | 2y 8m | 27 | 27 | 46% | +90.18% | +52.41% | 96% |
| 2 | Jeff Bezos & John Elkann - Italian Tech … | 4 months | 27 | 26 | 55% | +7.28% | +5.73% | 73% |
| 3 | Why "AI Coming for Your Job" is Not a Ba… | 6 months | 27 | 27 | 51% | +13.04% | +5.14% | 74% |
| 4 | Elon Musk: A Different Conversation w/ N… | 3 months | 27 | 27 | 51% | +1.25% | +1.35% | 44% |
| 5 | Trump 2.0 Tariff Tracker | 1 month | 27 | 0 | 68% | — | — | — |
Alpha = Portfolio return minus benchmark (SPY) return. Win Rate = Percentage of theses that outperformed the benchmark. Avg Conf. = Average confidence assigned to theses. Click any column header to sort.