Quant Arena / Research benchmark
Leaderboard
Ideas meet the market.
How well can AI turn data into a working strategy? A shared test of quantitative research, measured in daily stock selection.
Explore the Arena- Task
- Daily stock selection
- Evaluation period
- 2022–2024
- Track
- Vanilla / Iterative research
Leading results
Overall score · Reference includedThe results
A common ground for comparison.
Loading the latest results…
Scroll horizontally to compare all metrics
| Model |
|---|
Each row is a published run, including different reasoning efforts. Ranks follow the overall score.
Return and risk metrics: out-of-sample, 2024.
- Research, then evaluation.
- Agents develop strategies on 2022–2023 data. The out-of-sample evaluation tests those strategies on 2024.
- Return alongside risk.
- ARR measures annualized return. Sharpe measures risk-adjusted return. Maximum drawdown shows the largest peak-to-trough loss.
- A reference point.
- Kepler-1 is a proprietary strategy developed by Einsia AI, shown as a reference model. Its construction is not disclosed.