What is market breadth?
Market breadth measures how many stocks participate in a move, rather than how far an index travelled. An index can rise while breadth falls, which means a handful of large constituents are carrying it. QuantMedia publishes its own breadth measure — Signal Breadth — defined as the share of scored equities meeting a fixed 22-of-30 technical threshold, updated after every US close.
Why breadth and price can disagree
A capitalisation-weighted index reflects the largest members disproportionately. If five mega-caps rise 5% while 300 smaller members fall 1%, the index still prints a gain. Breadth measures catch that divergence, which is why they are used as a check on the durability of a move rather than as a directional signal.
Common breadth measures
| Measure | Definition | Notes |
|---|---|---|
| Advance/decline line | Cumulative advancers minus decliners | The classic; sensitive to universe composition |
| % above 200-day MA | Share of members above their long MA | Slow, widely quoted |
| New highs − new lows | 52-week highs minus lows | Sharp at extremes, quiet otherwise |
| QuantMedia Signal Breadth | Share of scored stocks at ≥ 22/30 signals | Multi-factor rather than single-condition; fixed universe of 180 |
The difference in the last row is that the underlying condition is a multi-factor score spanning trend, momentum, volume and volatility, not a single moving-average comparison. That makes it a stricter test of participation — a stock can sit above its 200-day average while failing most of the other checks.
Reading it honestly
- It is descriptive. QuantMedia publishes no evidence that its breadth reading predicts returns, and does not claim it does.
- Universe matters more than people expect. Breadth on 180 liquid large- and mid-caps behaves differently from breadth on 3,000 listed names. Readings from different universes are not comparable.
- Near-threshold effects. When the median score sits close to 22, breadth becomes volatile for reasons that have little to do with the market.
- It lags turns. Trend-weighted inputs mean breadth confirms moves rather than anticipating them.
What the first twenty observations showed
Signal Breadth started publishing on 17 August 2026 and is not backfilled. Over its first twenty sessions (to 14 September 2026) it ranged from 42.5% to 15.6%, with a median of 29.1%; behind those readings were 28 to 76 BUY names out of 179 scored. Two properties matter for reading it. The series is persistent: its one-day autocorrelation is 0.70, about three standard errors from white noise, because the underlying signals include slow-moving trend conditions. And it is almost the same information as the median score (correlation 0.84 across days), since both summarise one scan. A run of eleven down days against five up days over four weeks is therefore what one drawn-out decline looks like, not evidence of a regime. The full analysis, including how many observations a predictive test would need, is in Signal Breadth: what twenty observations can and cannot say.
Compute it yourself
The published JSON carries everything needed to recompute the figure, so it can be checked rather than trusted:
d = json.load(urllib.request.urlopen('https://quantmedia.io/data/signal_breadth.json'))
print(d['market_date'], d['buy_signals'], '/', d['scored_stocks'], '=',
round(100 * d['buy_signals'] / d['scored_stocks'], 1), '% (published:', d['breadth_pct'], ')')
The two numbers must agree to one decimal; the pipeline's validator
refuses to publish a day where they do not. Comparing the result with the
quoted threshold and signal_count fields also shows
whether the methodology version has changed since the figure was cited.
Current reading and data
The live figure, its distribution and the historical series are published on the Signal Breadth index page, with machine-readable copies at /data/signal_breadth.json and /data/breadth_history.json. Both are free to cite with attribution to QuantMedia and the market date.