Advance Decline Ratio Explained: The A/D Line and How to Read It

The advance decline ratio is one of the simplest numbers in market data: take all the stocks in a chosen universe, count how many closed higher than the previous day, count how many closed lower, and divide the first by the second. A ratio above one says more names went up than down. A ratio below one says the opposite.

Its whole purpose is to answer a question the index level cannot: when the market moved, how many stocks actually came along for the ride? An index can rise on a handful of heavyweights while most of its constituents fall. The advance decline data is how you see that.

What the data actually is

Every trading session, an exchange publishes a simple tally of its listed stocks:

• Advances: stocks that closed above the previous day’s close.
• Declines: stocks that closed below it.
• Unchanged: stocks that closed at exactly the same price.

That is the raw material. Everything built on top of it is arithmetic on those three counts.

Two definitional choices matter more than most people realise, and they are the first thing to check before comparing one breadth figure to another.

The first is the universe. Breadth for all stocks listed on an exchange is a different measurement from breadth within the Nifty 50 or the Nifty 500. The full exchange universe includes thousands of names, many of them small, thinly traded, and prone to hitting price bands. A narrow index universe is cleaner but tells you only about large companies. Neither is wrong, but a ratio computed on one is not comparable to a ratio computed on the other.

The second is what counts as a stock. Some tallies include every traded scrip, others exclude suspended names, illiquid counters, or non-equity instruments. A universe padded with securities that barely trade will produce breadth numbers driven by noise rather than by the decisions of real buyers and sellers.

The ratio, the difference, and the line

Three related measures come out of the same counts, and they are routinely confused.

The advance decline ratio is advances divided by declines. It is a single session snapshot. A ratio of 2 means twice as many stocks rose as fell. A ratio of 0.5 means twice as many fell as rose.

Advance decline breadth, sometimes called net advances, is advances minus declines. It is the same idea expressed as a count rather than a proportion. Practitioners use it when the raw scale matters, since a ratio can look extreme on a day when very few stocks moved at all.

The advance decline line is the cumulative version. Start from an arbitrary base number, then each day add advances minus declines to the running total. Plot that total over months or years and you get a line that rises when participation has been broadly positive and falls when it has been broadly negative. The A/D line is the form most analysts actually look at, because a single day’s breadth is mostly noise while the accumulated path is a description of a regime.

One important property of the A/D line: its absolute level is meaningless. It depends entirely on where you started the running total and on how many stocks are in the universe. Only its shape and direction carry information.

How it is read

The standard use is comparative. You put the A/D line next to the index over the same period and look at whether the two agree.

When an index makes a new high and the A/D line makes a new high alongside it, the move is described as broad: a large share of the universe is contributing. When the index makes a new high but the A/D line does not, the move is described as narrow: the index is being carried by fewer names, often the largest ones, while the typical stock is flat or falling. Practitioners call this a breadth divergence.

A second common reading is at extremes. Sessions where the ratio is very lopsided, in either direction, are often described as washout or thrust days, and analysts note them because they tend to cluster around turning points in sentiment. Note the word “cluster”. Clustering near turning points is not the same as marking them.

A third use is simply as context for other work. If you are studying sector rotation or looking at relative strength, knowing whether the broad market is participating changes how you interpret a single sector’s move. A sector that is rising while breadth is broadly negative is doing something different from a sector rising while everything rises.

What it does not tell you

Breadth is one of the most over-read numbers in market data. It is intuitive, it is free, it updates daily, and it lends itself to confident narration. That combination is exactly why it deserves a hard list of limits.

It is a count, not a magnitude. A stock that rose 0.05 percent and a stock that rose 12 percent both count as one advance. A session in which 300 names crept up and 200 names collapsed will show a ratio above one while enormous value was destroyed. Breadth deliberately throws away the size of moves. That is its design, not a bug, but it means breadth alone can point the opposite way to the actual change in market value.

It ignores size. Every stock gets one vote. In a market where a small number of very large companies account for a large share of total capitalisation, breadth and index returns can disagree for long periods without either being wrong. They are measuring different things: one measures the typical stock, the other measures capital weighted value.

It is not a forecast. This is the central error. A narrowing A/D line is a statement about what has already happened to participation. Turning that into “the rally is about to end” adds a prediction that the data does not contain. Breadth has narrowed and stayed narrow for extended stretches. It has also narrowed before declines. The indicator gives you no way to distinguish the two in advance, and any claim otherwise is being read into the number rather than out of it.

It is sensitive to universe choice. Change the universe and the divergence can appear or vanish. If you compute breadth on the full exchange list, thousands of micro caps dominate the count. If you compute it on a large cap index, they are absent entirely. Any breadth conclusion should carry its universe in the same sentence.

Its history is easy to corrupt. Building an honest long A/D line requires knowing which stocks were actually listed and traded on each past date, including names that have since been delisted or merged away. Rebuilding it from today’s constituent list quietly drops the failures. That is the same survivorship bias problem that afflicts backtests, and it applies to breadth history too.

It says nothing about a company. Breadth is a market aggregate. It carries no information about any individual business, its earnings, or its valuation. It is a description of a market’s internal texture and belongs in the context section of a research note, not the conclusion.

Practical habits

• Always state the universe alongside the number. “Breadth was positive” is incomplete.
• Prefer the line to the ratio for anything beyond a daily comment, since one session is mostly noise.
• Treat divergences as questions to investigate, not as conclusions.
• Check whether the underlying constituent history is point in time before drawing any long-run inference, since point-in-time data determines whether the line describes the past or a reconstruction of it.
• Never let a breadth reading stand in for company level work.

Related reading

• Portfolio metrics explained: the hub for how portfolio and market measures are defined and read.
• Market breadth indicators: the wider family of participation measures this one belongs to.
• FII and DII flows explained: another widely watched aggregate, with its own over-reading problem.
• Sector rotation strategy in India: where breadth data is often used as background context.
• How Indian indices are constructed: why index moves and typical stock moves can diverge.

This article is educational. Altys Labs is not a registered research analyst or investment adviser, and nothing here is investment advice or a recommendation to buy, sell, or hold any security.