What market breadth tells you about Indian stocks
The tests behind every reading on the Indian Market Breadth page. Each measure had to work on two separate stretches of history, or it stayed off the page.
The question
Breadth counts how many stocks are taking part in a move: how many rose or fell today, how many are at a one-year high or low, how many are above their 50-day average. Traders watch it to judge whether a market is healthy. Does any of it tell you something useful about what comes next, and which readings are worth a daily look?
How it was tested
- Stocks: each day's 1,000 most-traded stocks on the NSE, by value traded over the previous 20 days, counted equally. So the results mostly describe mid-sized and small companies.
- History: NSE's daily files from April 2015 to September 2026, adjusted for splits and bonus issues.
- Two periods: every measure was tuned on 2016–2021, then checked unchanged on 2022–2026. A measure counts only if it worked in both.
- 41 measures: rising and falling stocks, 3–6% daily moves, 52-week and 26-week highs and lows, stocks above 10- to 200-day averages, moving-average crossovers, big monthly movers, volatility-scaled moves, and the Nifty against the broader market.
- Three things to predict, one for each decision a trader makes:
- the typical stock's return over the next 2–4 weeks (should I be in the market?);
- the next month's deepest fall in mid and small companies: the biggest drop over the next 20 trading days in an equal-weighted basket of the stocks ranked 251 to 1,000 (how much risk am I running?). On an average day it was about −3.8%;
- how often breakouts work. A breakout is a stock rising 4% on higher volume, or closing above its previous 50-day high (about 110 a day); it works if it reaches +10% before it falls 5%, within 20 trading days (should I press new entries?). About 30% work on an average day.
What it found
Breadth does not forecast direction. None of the 41 measures said whether the typical stock would rise or fall over the next few weeks. Results that looked good in one period reversed in the other.
It does tell you about risk. Weak breadth was followed by deeper falls in mid and small companies, and by fewer breakouts that worked, in both periods. Every trading day was grouped by its breadth reading, and what followed was averaged for each group. From 2022 to 2026, the years held back for checking:
Stocks above their 50-day average Deepest fall over the next month Over 76% −1.8% 63% to 76% −4.0% 50% to 63% −4.3% 30% to 50% −4.5% Under 30% −4.5% All days −3.8% Stocks at one-year highs minus lows (of 1,000) Breakouts that worked Over +62 33% +31 to +62 30% +9 to +31 30% −22 to +9 29% Under −22 26% All days 30% The two read differently. Falls were about as deep across most of the range and clearly milder only when breadth was broad, with three in four stocks above their average. Breakouts worked a little more often at each step up in highs over lows. These are averages over many days, not a forecast for any one day: a weak reading doesn't mean a fall is coming, only that falls have tended to be deeper and breakouts less reliable. It's a guide to how much to risk, not a signal to act on. The groups' edges were fixed on 2016–2021 data before these years were checked.
It often warns early at tops, not at bottoms. Before 7 of the 10 falls of 15% or more since 2016 (including late 2016, early 2018, both falls of 2022, December 2024 and July 2025), at least one of the page's main readings had peaked 2–6 months before prices did. There was no warning before the falls that began in March 2019 and January 2020 (COVID). At bottoms, breadth usually turned within a couple of days of prices.
Single days are noise. A big up or down day said almost nothing two days later. Ten-day averages and trend readings kept most of their value a week or two on. That's why the page shows each reading's history and its two-week average, not just today's figure.
Two readings add something the index doesn't. One-year highs minus lows, and the 10-day average of rising minus falling stocks, still forecast risk after allowing for where the index itself stood. The share of stocks above their 50-day average is the clearest picture of the market's state, but mostly repeats what the index's own trend says.
Exact thresholds don't matter. Up 3%, 4%, 5% or 6% in a day; 20%, 25% or 30% in a month; 26- or 52-week highs: nearby choices gave the same results. The page uses the familiar ones.
Washouts are only a weak hint. When fewer than 30% of stocks were above their 50-day average, the typical stock rose 2.0% over the next month in 2022–2026, but only 0.6% in 2016–2021. Worth watching, not a buy signal.
Why each reading is on the page
| Reading | What the tests found |
|---|---|
| Stocks at one-year highs minus lows | The best all-round reading: forecasts both breakout success and the depth of falls, in both periods, beyond what the index shows. Peaked early before several big falls. |
| Stocks above their 50-day average | The clearest split between calm and rough months ahead. 40-, 50- and 100-day averages behave alike; 50 is the familiar one. |
| Rising minus falling stocks, 10-day average | Forecasts risk beyond the index, and peaked 2–6 months early before five of the big falls. Covers many look-alike measures (4% moves, 4% ratios), which were dropped. |
| Stocks up or down 4% today, and up or down 25% or 50% in a month | Context: what happened, and how many big moves there are to trade. No forecasting value of their own. |
| Nifty 50, Midcap 150, Smallcap 250 | Show which size of company is leading. Their gap didn't forecast anything once the index's own move was allowed for, so it's context only. |
| Sectors | Sectors whose stocks led over the past 1–3 months tended to keep leading: the top 3 beat the bottom 3 by about 1.3% over the next month, in both periods (positive in 11 of 12 years). This shows up when a sector's stocks are counted equally; the size-weighted sector indices barely showed it. Breakouts in leading sectors didn't work much better, so it's a guide to where to look, not when. |
| India VIX against actual swings | Options priced in a bigger move than the Nifty then made on about 4 days in 5, and buying an at-the-money straddle every month lost money in most months. Nothing reliably said when buying options would pay. Shown so you know whether options are cheap or dear today. |
| Commodities | Judged on their own trend, because weak Indian breadth didn't time gold, silver, US or Hong Kong stocks, or bonds. Gold was the one steady diversifier: it rose in 6 of the 8 falls of 15% or more in small companies since 2016. |
Tested and left off
- A live breakout success rate: it moved with the readings already shown and added nothing.
- Breadth of large, mid and small companies separately: the three move together.
- Signals for short selling: breakdowns in F&O stocks worked no better than shorting at random since 2022. Shorts paid only during fast falls, which no reading flagged in advance.
- Timing option purchases with VIX or breadth, including out-of-the-money Nifty options with targets and stops: no rule made money in both periods.
- Switching to gold, silver, US or Hong Kong stocks, or bonds when breadth is weak: no reliable timing.
- A single "market state" label (uptrend, narrow, downtrend): it flipped about every two weeks and turned only after falls had begun.
- Weighing rising and falling stocks by money traded: it repeated the plain counts.
- Moving-average crossovers, 2-standard-deviation moves, and other averages: weaker, or duplicates of what's shown.
Limits
The effects are real but modest: tilts, not switches. Eleven years hold only about ten serious falls, so the warnings at tops are a pattern across a handful of cases, not a statistical certainty. Stocks that were later delisted drop out of the forward returns, and the sector results use today's sector list for the past. Use the page as one input to how much risk you take, not as a trading system.
Data
NSE's free daily price and index files; sectors from NSE's Nifty Total Market list; commodities from MCX futures through Zerodha's Kite Connect. The page's full daily history since April 2015 is a free download: breadth.csv. Study run September 2026.