Breadth Calculations: Difference, Ratio and Percentage
One session can produce several valid readings. Start with the question each one answers.
In Breadth Analysis 101, we introduced different ways of expressing market breadth. I want to take one session and show why the numbers can look different without contradicting one another.
Are we asking who won, how strongly one side outnumbered the other, or what share of the market advanced? Those are three questions, so they deserve three clearly labelled answers.
Start with the same raw observations
Imagine a hypothetical NSE basket with 100 eligible stocks: 55 advanced, 35 declined and 10 were unchanged. Every stock has comparable closing prices for both sessions. These are teaching numbers, not a historical market reading.
We will write advances as A, declines as D and unchanged as U. Our eligible total E is A + D + U = 100. If these terms are new, start with the stock-count lesson.
Difference: how many more stocks advanced?
Net advances = A − D = 55 − 35 = +20 stocks.
This is a headcount margin. I find it useful as the most direct description of the balance. It is not a percentage and says nothing about the average price change.
If a second basket has 550 advances and 350 declines, its difference is +200. The margin is larger, but so is the basket. Raw differences alone do not establish stronger proportional participation.
Ratio: how many advances per decline?
A/D ratio = A ÷ D = 55 ÷ 35 ≈ 1.57.
There were about 1.57 advancing stocks for each declining stock. A ratio of one means equal counts; it does not mean every stock was unchanged. The ratio also excludes unchanged stocks from both terms.
When D is zero, ordinary division is undefined. I would report the underlying counts rather than quietly substitute an arbitrary large number. A ratio can become very large when only a few stocks decline, so smoothing it does not automatically make it well behaved.
Percentage: which denominator did we choose?
Advancing share of all eligible stocks = 100 × A/E = 55%.
Advancing share among movers = 100 × A/(A + D) ≈ 61.11%.
Both are correct. The second deliberately leaves out the ten unchanged stocks. Calling it the percentage of the entire basket that advanced would be incorrect.
A different public measure is net-advance percentage = 100 × (A − D)/E = +20%. StockCharts documents the net-advance percentage convention. It measures the margin as a share of the basket, not the advancing share. Its range is −100% to +100%, not zero to 100%.
The comparison I would actually make
| Question | Reading from our example |
|---|---|
| How many more advances than declines? | +20 stocks |
| How many advances per decline? | 1.57 |
| What share of eligible stocks advanced? | 55% |
| What was the net margin as a share? | +20% |
I think the useful habit is to finish the sentence after the number. “Breadth was 55” is incomplete. “Fifty-five percent of eligible stocks advanced” tells the reader exactly what happened.
Limits before interpretation
Changing constituents, incomplete coverage and inconsistent observation times can spoil any of these comparisons. The eligible-stock denominator lesson explains why. None of the four readings measures the size of gains or losses, and none on its own establishes a profitable entry.
When using IndexBreadth, read the displayed label before comparing values across charts. These are standard public calculation relationships; the purpose is to understand the question, not to assume a platform’s internal formula.