The Nifty 50 is free-float market-capitalisation weighted. A handful of the largest constituents can carry the index higher while most of the other names fall, and the headline level will not tell you that happened. Breadth is the check on that.
Two families of measure are in common use. Advance-decline statistics count how many constituents rose, fell and were unchanged in a session. Participation ratios count what share of constituents satisfies a condition, most often trading above a moving average.
Common measures
| Measure | Computation | Question it answers |
|---|---|---|
| Advances and declines | Count of constituents up versus down on the day | Was today’s move shared? |
| Advance-decline ratio | Advances divided by declines | How lopsided was it? |
| Advance-decline line | Running cumulative sum of advances minus declines | Is participation trending with the index or away from it? |
| Percent above 200-day moving average | Share of constituents whose close exceeds their own 200-day average | How much of the universe is in a longer-term uptrend? |
| Percent above 50-day moving average | The same on a shorter window | The medium-term version, noisier and faster |
| New highs minus new lows | Count at 52-week extremes | Where the tails of the distribution sit |
The percent-above-moving-average measures are the ones most easily reproduced, because they need only closing prices and a constituent list.
Getting the constituent list right
Breadth is defined relative to a universe, and the universe changes. The Nifty 500 of 2015 is not the Nifty 500 of today, and computing a fifteen-year breadth series against today’s membership list means every constituent survived to be in it. That is survivorship bias in its purest form, and it inflates the historical share of constituents in uptrends because the names that collapsed and were removed never enter the count.
A correct series needs point-in-time index membership, so that each date’s breadth is computed over the constituents as they stood on that date.
Caveats
Breadth measures are not comparable across universes. Percent above the 200-day average in the Nifty 50 and in the full NSE cash universe are different statistics, and a threshold calibrated on one does not transfer.
Counts are equal-weighted by construction. RELIANCE and a ₹400 crore small cap contribute the same one vote, which is the point of the measure and also its limitation.
No level is inherently meaningful. Thresholds like “below 30 percent above the 200-day average” circulate widely without published Indian evidence behind them. Treat any such rule as a hypothesis and hold out data you have not looked at before believing it.
Breadth describes the present. It is computed from prices that have already printed, so a strategy reading it acts from the next session at the earliest.