Skip to content
StockMarketAgent
Methodology · Breadth

Breadth health score

A composite breadth indicator that rolls up four sub-components into a 0-100 score classifying the breadth environment as healthy, caution, or unhealthy.

Specification

Breadth health score — operational spec

Narrow breadth is one of the most reliable late-cycle markers in the historical record. The score classifies the breadth environment so the reader can size new positions accordingly.

Inputs
  • Daily advance and decline counts across the broad market
  • Daily new-52-week-high and new-52-week-low counts
  • Daily percent of constituents trading above their 200-day moving average
  • Daily McClellan-style smoothed advance-decline oscillator
Computation
  • Each sub-component is converted to a percentile rank against its own multi-year history.
  • Equal weights are applied. Recency-weighted variants were tested and rejected for back-fitting.
  • The composite is mapped to a three-band classification: healthy (>= 60), caution (30-60), unhealthy (< 30).
  • Cumulative advance-decline slope is the highest-information sub-component and its delta is published prominently.
Outputs
  • Composite breadth score 0-100 with band classification.
  • Four sub-component scores with month-over-month deltas.
  • A narrow-leadership flag when the cap-weighted index is rising while equal-weighted breadth deteriorates.
Limitations
  • Breadth signals can deteriorate for months before a market top materializes; the score is a prior, not a trigger.
  • Index membership churn can produce step-function changes in new-high counts that are not regime signals.
  • The McClellan oscillator inherits the smoothing parameters of its construction; the score uses the conventional 19/39 EMA spread but is sensitive to that choice.

Frequently asked

Does narrow breadth always precede a drawdown?
No. Narrow breadth is associated with elevated drawdown risk but is not deterministic — the index can drift sideways or melt up further while breadth deteriorates. The score is a prior that should widen the margin of safety on new positions, not a market-exit signal.
Why include both percent-above-200-DMA and the McClellan oscillator?
They measure different time horizons. Percent-above-200-DMA is a slow trend-participation read; the McClellan oscillator is a faster smoothed-momentum read on advance-decline. Combining them captures both the regime backdrop and the near-term inflection.
Is the score predictive of next-month returns?
Marginally, and not enough to use as a standalone signal. Its predictive power is conditional — it is most informative when the regime composite is also flagging late-cycle conditions, which is why the two are published together.
See the full methodology hub for the rest of the model registry, or open the glossary entry for the headline definition.