Action Model quality Methods

Model quality

This page defends whether Forecast Alpha is trustworthy. It’s intentionally “review-ready”: clear metrics, rolling results, and regime-aware robustness — without exposing private code.
Series: Loading…
Overall confidence (published)
—
Rolling performance (example)
If you haven’t published history.json yet, this will show “missing” — that’s OK for now.

Primary inference tables (what your paper will cite)

Main table: pooled OLS with Newey–West standard errors + regime-interaction model (also Newey–West).
Robustness table: regime-conditional coefficients with wild-bootstrap CIs/p-values + effective sample size (n_eff).
Table Caption (paper-ready) What a consumer should take away
Table A Pooled & interaction regression (Newey–West). Estimates the relationship between model-implied inflation and market-implied inflation, including regime interactions. Standard errors are heteroskedasticity- and autocorrelation-consistent (HAC). “When the model differs from the market, it’s systematic — not random noise — and the relationship varies by regime.”
Table B Regime-conditional results (wild bootstrap). Regime-specific coefficients and inference using wild bootstrap to improve small-sample robustness; reports n_eff for each regime. “In some regimes, the signal is strong; in others it’s fragile — so we temper confidence accordingly.”