Overview

Metrics for the Common × Continuous cell — one time-series factor broadcast across n_assets assets (a market-wide signal: VIX, USD index, oil, sentiment). Aggregation is time-series first: per-asset ordinary least squares (OLS) \(\beta\) over all dates, then a cross-asset \(t\) on the mean of the per-asset betas.

Metric	Page
Cross-asset \(t\) on per-asset \(\beta_i\) (BJS aggregation)	common_beta
Positive / negative / neutral profile of per-asset \(\beta_i\)	common_beta
Spread between top- and bottom-bucket assets sorted by \(\beta_i\)	common_quantile
Sign-asymmetric slopes (positive vs negative regimes)	common_asymmetry

common_beta carries the cross-asset significance test; common_quantile and common_asymmetry are descriptive profile diagnostics.

These metrics test the cross-asset distribution of per-asset \(\beta_i\), so they need n_assets >= 2. At n_assets == 1 there is no cross-section: the cell is PANEL, so evaluate raises IncompatibleAxisError (or returns NaN + structure_mismatch under strict=False) rather than running. For single-asset time-series questions use predictive_beta for the direct dense predictive-regression slope, the panel-input directional_hit_rate on (date, asset_id, factor, forward_return) for directional skill, or the two-column series diagnostics (positive_rate, oos_decay, ic_trend) on an explicit (date, value) series.

The Common × Sparse cell swaps this continuous regressor for a {0, R} broadcast event dummy (R unrestricted; {0, 1} for a pure event flag is the simplest form). It does not reuse this time-series-first OLS-\(\beta\) flow — a broadcast sparse dummy is evaluated through the event-time metrics under Common sparse.