The best-known version is Martin Zweig's, which measures a ten day average of advancers as a share of all issues moving from below 40 percent to above 61.5 percent within a short window. Variants use volume or the mcclellan-oscillator.
The rationale is that a sudden, broad surge in participation reflects a genuine shift in demand rather than a bounce led by a few names, and historical instances have a notably good record of being followed by strong returns.
The sample is small. Breadth thrusts are rare by construction, which means the impressive hit rate rests on a handful of observations spanning decades, and definitions have been adjusted over time in ways that flatter the record. Treat it as suggestive context, not a system.
Related: market-breadth, mcclellan-oscillator, percent-above-moving-average, sample-size, capitulation