> ## Documentation Index
> Fetch the complete documentation index at: https://docs.worthcall.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Model Stats

> Per-model performance comparison — accuracy, agreement, and calibration.

# Model Stats (`/model-stats`)

Compares Claude, GPT-4o, and Gemini performance across all evaluations.

## Metrics per model

* **Total evaluations** scored
* **Average score** given
* **Win rate** when that model recommended trading
* **Accuracy** — how often the model's recommendation matched the outcome
* **Agreement rate** — how often models agree with each other

## Model comparison

Side-by-side table showing which model is most accurate, most conservative, and most aggressive.