Lead-lag Detection

Lead-lag Detection measures whether the chart symbol systematically moves before or after a related instrument, and by how many bars. The cross-correlation sweep behind lead-lag detection reruns on both series' returns every bar, from -20 to +20 lags, and the peak value plots with approximate significance bounds at 2/sqrt(n). A relationship is only called while the peak sits at a nonzero lag beyond the bounds; the dashboard names the leader and its lag in plain language.

How to Trade the Lead-lag Detection?

• Peak at a positive lag: the chart symbol leads the comparison by that many bars; at a negative lag, the comparison leads the chart.
• Significance first: inside the bounds the honest read is no significant lead - alerts mark the transitions in and out.
• Compare against lag 0: the Lag-0 Correlation plot shows the plain contemporaneous correlation - the value added by shifting one series in time, at a glance.

Peak Lag Shift and Leader Flip alerts round out the set - a relationship is only worth trusting while its lag and sign persist.

Lead-lag Detection Settings

• Comparison Symbol (default SP:SPX): the second instrument of the pair.
• Source (default close) and Returns (default Log): the correlated series, sampled only on fresh comparison bars so session mismatches cannot fabricate lags.
• Correlation Window (default 100): overlapping observations n at every lag, and the bounds.
• Max Lag (default 20): the sweep spans -K to +K bars; wider sweeps multiply the odds of spurious peaks.
• Show Profile (on) with Profile Offset (default 5): the per-lag correlogram beside the last bar.
• Show Dashboard (on, Top Right, Small); style: Lag-0 Correlation (on), Normalized Peak Lag (off), Significance Bounds (on).

Frequently Asked Questions

How does this relate to Lead-lag Instrument Triggers?

Why correlate returns instead of prices?

Should I trust a peak that barely clears the bounds?