Market Efficiency Trend
Weekly R-squared of the one-parameter Prelec fit applied to active retail trades.
As of 2026-04-20, the fit R-squared was 0.895, with a 52-week rolling z-score of -1.72. 13-week moving average: 0.936.
Weekly history
R-squared by wallet class (latest week)
| Wallet class | R-squared |
|---|---|
| Algorithmic | 0.887 |
| Active Retail | 0.895 |
| Sophisticated | 0.922 |
| Casual | n/a |
| One-Shot | n/a |
Methodology
Each week, for each wallet class, we fit a one-parameter Prelec weighting function to that class's pooled price-vs-realized-frequency scatter. The R-squared captures how well a single behavioral parameter describes that week's calibration gap. Higher values indicate a smooth, predictable distortion curve; lower values indicate that the week's mispricing is dominated by idiosyncratic shocks rather than a stable behavioral signature.
The headline index uses the active-retail R-squared because active retail is the largest human wallet cohort and most closely represents how well human beliefs price the market. The algorithmic R-squared is shown for comparison but can be noisy because algorithmic trading is more strategic than belief-driven.
Data
efficiency_history.csv · efficiency_latest.json · efficiency_timeseries.json