Volume analysis and market breadth
The answer
Volume tells you how much trading happened behind a price move. Breadth tells you how many separate stocks took part in an index move.
Neither predicts anything. What they do is check a claim you hear every day and have no other way to test. When somebody says "the market is up", breadth is how you find out whether that is a statement about a market or a statement about 5 companies.
Why this costs you money
Cluster 01 article 11 made a point that most readers accepted and then forgot. Major indices are weighted by market value, so the largest few companies move the index far more than the rest. "The market is up 1%" can be true on a day when most listed companies fell.
Here is what that costs, and it is a specific, common and demoralising experience.
You hold 12 stocks. Most of them are mid-sized companies. The index is at a record high, the news says the market is strong, and your portfolio is down 18%. You conclude that you are bad at choosing stocks. You sell the losers, buy something that is going up, and repeat.
The conclusion was wrong and the correction made it worse. In a narrowing market, the index rises because a small number of very large companies rise, and the majority of stocks fall at the same time. Your results were normal for what you owned. You just had no instrument for seeing it.
The second cost runs the other way and it is larger. A market can rise for months on a shrinking group of stocks. The index looks healthy the entire time. Every month the advance depends on fewer companies. When those companies finally stop rising, there is nothing underneath, because everything else stopped rising long ago.
Breadth is the measurement that shows this while it is happening. It does not tell you when to sell. It tells you how much of the market's apparent health is real, and that is a question a price chart of an index physically cannot answer.
How it works
There are 2 separate ideas in this article, and they answer 2 different questions.
Volume: how much trading was behind this move
Volume is the number of shares traded in a period. It has no direction. Every trade has a buyer and a seller, so a large volume day is not a buying day or a selling day. It is a day on which many shares changed hands.
What volume measures is how much disagreement was resolved. A price move on heavy volume happened while many participants were willing to transact at those prices. A price move on light volume happened because few were.
Volume is only meaningful against the same instrument's own history. Compare today's volume to the average of the last 20 or 50 days on the same stock. Comparing volume between 2 different stocks tells you which is larger, not which is more active.
In India there is a second and better number, and it is described in the India section: delivery volume, the shares that actually moved between accounts rather than being squared off within the day.
Breadth: how many stocks took part
Breadth counts companies instead of weighting them by size. There are 4 measures worth knowing, and each answers a different question.
1. Advance-decline. Each day, count how many stocks rose and how many fell. The difference is the day's net advances. Add those differences up over time and you get the advance-decline line. It answers: is the typical stock rising?
2. New highs against new lows. Each day, count how many stocks made a new 52-week high and how many made a new 52-week low. It answers: is leadership expanding or is damage spreading? This measure reacts more slowly than advance-decline and it is harder to fake, because a 52-week high requires a year of work.
3. Percentage of stocks above their own moving average. For each stock in an index, check whether it is above its own 200-day moving average. Express the count as a percentage. It answers: what share of the market is in an uptrend on its own terms? This is the clearest of the 4 for a beginner, because the output is a single number between 0 and 100 with an obvious meaning.
4. Equal-weight against cap-weight. Compare an equal-weighted version of an index to the standard market-value-weighted version. If the cap-weighted index is rising faster, the largest companies are doing the work. It answers: how concentrated is this advance? This is the fastest check available and it needs no counting at all.
What it tells you, and what it does not
Breadth tells you the participation behind an index move. Volume tells you the activity behind a price move. Both are descriptions of what happened.
Seven limits, and 3 of them are severe.
Breadth divergences can last for years. An advance can narrow and keep going for a very long time. Anybody who sold the first time breadth weakened has usually been out for a large part of the remaining advance. Breadth is a condition, not a trigger.
Advance-decline counts treat all stocks equally, including the ones nobody trades. A stock that traded 400 shares counts exactly as much as one that traded 4 crore rupees. This matters enormously in India, where the listed universe has a very long tail of barely traded companies.
In the United States the count is contaminated by things that are not companies. A large share of NYSE-listed issues are closed-end funds, exchange traded funds, preferred shares and interest-rate-sensitive instruments. On a day when interest rates fall, those issues rise together, and the advance-decline line reports broad strength that has nothing to do with companies.
Volume has no direction. This bears repeating because it is the most common error in the whole subject. Heavy volume on a down day is not "selling pressure". It is a day on which a lot of shares changed hands and the price fell.
Volume comparisons break on event days. Expiry days, index rebalance dates, results days and block deals produce volume that is mechanical. A volume spike tells you an event happened. It does not tell you an opinion changed.
Breadth says nothing about your specific holdings. Wide breadth does not mean your stock will rise. It tells you the environment, and nothing more.
Narrow breadth is not automatically unhealthy. Sometimes a small number of companies genuinely are where the economic change is happening. The concentration is a description of risk, not a verdict on value.
The decision rule
Use breadth to translate the sentence "the market is up" into a fact about your own portfolio.
If the index is rising and the percentage of stocks above their 200-day average is rising too, the advance is broad. Your mid-sized and smaller holdings should be participating. If they are not, the problem is in your selection.
If the index is making new highs while that percentage falls, the advance is narrowing. Expect a portfolio of ordinary companies to lag the index. This is not a reason to sell, and it is not a reason to abandon your holdings for whatever is leading. It is a reason to stop measuring yourself against the index and to reduce your total risk.
If new lows are expanding while the index is near a high, damage is spreading underneath a calm surface. Treat every breakout with more suspicion and every stop with more respect.
Use volume only to confirm a price event you already identified. Did the breakout happen on volume well above this instrument's own recent average, or not? That is a real question with a real answer, and it is nearly the whole honest use of volume.
Unless the day is an expiry, an index rebalance or a results day. On those days both volume and breadth carry mechanical flow, and reading them as opinion is reading a settlement procedure.
Try this now
Five minutes, and it will permanently change what you hear when somebody says the market went up.
- Pick a recent day on which your main index rose. Any day in the last month where the NIFTY 50, or the S&P 500, or whichever index you follow, closed higher. Note how much it rose, as a percentage.
- Open the constituent list for that index. Every exchange website and most broker apps have it. In India, the NSE website lists NIFTY 50 constituents with the day's change. In the United States, any index page lists S&P 500 members with their daily change.
- Count how many of the constituents actually rose that day, and how many fell. Write down both numbers.
- Now find the 5 largest companies in the index by weight. Note what each of them did that day.
- Do the same for a second day, chosen because the index fell.
What you should see. On many "up" days the count is far less impressive than the index move suggested. It is common to find an index up 0.8% on a day when roughly half the constituents fell. It is not rare to find an index up while a clear majority of its own members were down.
In step 4 you will usually see why. On the narrow days, 3 or 4 of the largest companies moved sharply in the index's direction, and their weight carried the whole number.
Now take the 1 extra step that makes this yours. Count how many of your own holdings rose on that day. If the index rose, half its members fell, and most of your portfolio fell, you have just measured the exact experience described at the top of this article. Your portfolio did not underperform because you chose badly. It underperformed because you own companies and the index moved on 4 of them.
Do this once a week for a month and you will develop something more useful than any indicator on your chart, which is an accurate sense of how much of the daily market number is real.
Three real cases
1. India in 2018 — the index rose and the market fell The NIFTY 50 finished calendar year 2018 modestly higher, while the Nifty Midcap 100 and the Nifty Smallcap 100 fell heavily over the same year. A person reading only the headline index number would have concluded that 2018 was a positive year for Indian equities. A person holding a typical portfolio of mid-sized and small Indian companies had a badly negative year. Both descriptions were accurate at the same time. The gap between them is exactly what breadth measures, and any of the 4 measures in this article would have shown it clearly by the middle of the year.
2. The United States in 2023 — concentration measured in a single ratio In calendar year 2023 the S&P 500 rose substantially, while the equal-weighted version of the same 500 companies rose by a much smaller amount. The difference is the concentration. A small group of very large technology companies carried most of the cap-weighted index's gain, and the average member of the same index did far less. This case is worth remembering because it shows the simplest breadth check working with no counting at all. Two index returns, subtracted, answered the question.
3. The United States, 1999 into 2000 — breadth deteriorating under a rising index Through 1999 the Nasdaq Composite rose dramatically, while the NYSE advance-decline line weakened, and the number of stocks making new 52-week lows was elevated for extended periods even as headline indices climbed. The advance was carried by a shrinking group. This is the most cited breadth case in market history and it carries an honest warning inside it. The divergence ran for many months before it mattered. Anybody who sold at the first sign of narrowing gave up a very large advance first. Breadth described the structure correctly and told nobody when.
The question that resolves it
A novice sees the index at a record high and asks: is the market strong?
An expert asks: how many companies is this record made of?
The first question has a 1-word answer that is often wrong. The second has a number, the number is available in about 60 seconds, and it changes what you do next. A record built from 400 rising companies and a record built from 40 rising companies are 2 different markets wearing the same headline.
There is a second question underneath, and it is the one that protects your portfolio. Is my benchmark still a fair description of what I own? In a narrow market it is not, and comparing yourself to it will make you sell ordinary companies at the wrong moment to buy the leaders at their most expensive.
What would make this wrong
The claim is that breadth measures participation, that participation and index level often diverge, and that knowing the difference changes how a reader interprets their own results.
The first part is easy to falsify and you should try. Take 20 days on which your index rose. Count constituents up and down on each. If the count was overwhelmingly positive on nearly every one of those days, then your index is not narrow, and the concern in this article does not currently apply to your market.
The second part is harder and more important. Nothing here shows that breadth predicts. If you collect every occasion on which an index made a new high while the percentage of stocks above their 200-day average was falling, and check what happened over the following 3 months, you will find a mixture of outcomes. Some narrow advances ended badly. Many continued for a long time. If you find that the divergences were consistently followed by declines within a defined period, believe your own count, but expect not to find that.
Four honest limits.
First, breadth is slow and vague as a timing tool. The 1999 case is the warning. A correct observation about structure told nobody when to act.
Second, the measures disagree with each other. Advance-decline can be positive while new lows expand. That is not a contradiction, because they measure different things, but it means there is no single breadth number.
Third, the composition of the counted universe matters more than anybody admits. Different data providers count different sets of stocks and produce different lines from the same market.
Fourth, volume confirmation is weak evidence. A breakout on heavy volume is somewhat more likely to hold than one on light volume, but the effect is not strong enough to trade on its own.
In India
Indian breadth data is available, less widely published than in the United States, and distorted in 2 specific ways that a reader must know about.
The advance-decline count is available daily. NSE publishes the number of advancing and declining stocks each session, and most Indian broker apps display it. NSE also publishes the number of stocks that hit their upper and lower circuit limits, which is an India-specific breadth measure with no direct American equivalent. A day with 300 stocks at lower circuit is a day of widespread forced selling, and no index number expresses that.
Distortion 1: the long tail of illiquid listings. A large number of listed Indian companies trade very little. In a raw advance-decline count, a company whose entire day's turnover was a few thousand rupees counts the same as a company that traded hundreds of crores. Indian raw breadth counts are therefore noisier than they look.
Distortion 2: index concentration in financial services. The NIFTY 50 carries a very large share of its weight in banks and other financial companies. This has 2 consequences and both matter. First, the index can move sharply on a banking development while the rest of the economy's listed companies do nothing. Second, and less obvious, the financial sector's own stocks are numerous and tend to move together, so a banking move affects both the weighted index and the unweighted count at the same time. Indian breadth and Indian index level are less independent of each other than American breadth and American index level.
The better Indian breadth check is index against index. Because of the illiquid tail, the most reliable and fastest Indian participation check is to compare the NIFTY 50 with the Nifty Midcap 100 and the Nifty Smallcap 100 over the same period. Those indices have liquidity requirements for membership, so they exclude the tail automatically. When the NIFTY 50 rises and the Smallcap 100 falls, participation is narrow, and you have established it with 3 numbers.
Delivery percentage is the Indian volume advantage. NSE and BSE publish the deliverable quantity for each stock: the shares that actually moved between demat accounts rather than being squared off intraday. A breakout on heavy total volume but very low delivery percentage was an intraday event. The same breakout on heavy volume with delivery percentage above the stock's own average is a different thing. No United States equivalent of this figure is published, and almost no Indian retail traders use it.
In the United States
The United States has the deepest published breadth data in the world, by a wide margin, and a long history of people arguing about how to read it.
The series go back decades. The NYSE advance-decline line has been computed for a very long time, and new-high and new-low counts have been published for generations. This means American breadth arguments can be tested across many complete market cycles. Indian equivalents cover a far shorter history.
Named tools built on breadth are standard. Sherman and Marian McClellan developed the McClellan Oscillator in 1969, which smooths daily net advances into a single fluctuating line, and the McClellan Summation Index which accumulates it . The percentage of S&P 500 stocks above their 50-day and 200-day moving averages is published daily by several data providers and is probably the most useful single breadth number available to a beginner.
The equal-weight comparison is easy in the United States. An equal-weighted version of the S&P 500 exists as a published index and as widely traded funds, so comparing the 2 return series takes a few seconds and requires no data work at all.
The contamination problem is real. As noted above, a large share of NYSE-listed issues are not operating company common stock. Many practitioners therefore prefer a "common stock only" advance-decline line, and the 2 versions can disagree, particularly during large moves in interest rates. If you use a United States advance-decline line, find out which universe your provider counts.
Where they differ, and what that tells you
Three real differences, each with a practical instruction attached.
1. The United States publishes more breadth data, over a far longer history. Decades of advance-decline and new-high and new-low series exist and are freely available. Indian series are shorter and less consistently published.
What that tells you is where to be careful with borrowed rules. Breadth rules developed on American data have been tested across many cycles. The same rules applied to Indian data have not been. A threshold that has meaning on the NYSE series is a guess on the NSE series until somebody tests it there.
2. Indian breadth is distorted by index concentration; American breadth is distorted by non-company listings. The NIFTY 50's heavy weight in financial services means an index move and a breadth move often share a single cause. The NYSE's large population of funds and preferred issues means the American count includes instruments that respond to interest rates rather than to business.
The instruction differs accordingly. In India, check the midcap and smallcap indices rather than the raw advance-decline count, because they exclude the illiquid tail and are less dominated by one sector. In the United States, check which universe the count includes, and prefer the percentage of index members above their own 200-day moving average, which counts only real index members.
3. India publishes delivery volume; the United States does not. This runs in India's favour and it is the single most underused piece of data available to an Indian retail trader. On any breakout, an Indian trader can ask whether the volume represented shares that stayed in somebody's account. An American trader cannot ask that question at all.
The translation rule for this cluster's final article is therefore simple. Take the American concept, which is participation, and measure it with the Indian instrument, which is the liquid index comparison plus delivery percentage. Do not import the American thresholds. Import the question.
Carry this
- Volume measures activity and has no direction. Compare it only to the same instrument's own recent average.
- Breadth counts companies. The index weights them. When the 2 disagree, the index is describing a small group.
- The fastest check needs no counting: compare a cap-weighted index to an equal-weighted one, or in India compare the NIFTY 50 to the Smallcap 100.
- Narrow breadth is a description of concentration, not a sell signal. It can persist for a very long time.
- In India, check delivery percentage before believing a volume-confirmed breakout. Nobody else in the world can.