What Is Delta Divergence?
Delta divergence occurs when cumulative delta - buying volume minus selling volume - moves in the opposite direction to price. Measuring it requires trade-by-trade data and a rule for classifying each trade as a buy or a sell, and that classification is inferred rather than reported.
Covered on this page: TradingView.
Delta divergence is a disagreement between price and a measure of aggressive buying. Before it can mean anything, it is worth being precise about what that measure actually contains.
How it works
Each trade is classified as a buy or a sell. A trade printing at the ask is treated as buyer-initiated; one printing at the bid as seller-initiated.
Delta is the running difference. Buy-classified volume minus sell-classified volume, accumulated across a bar or a session.
Divergence is when the two point different ways. Price makes a higher high while cumulative delta makes a lower one, or the reverse - the reading that gives the concept its name.
Every trade has both sides
Volume is never one-sided. A hundred shares trade because somebody bought a hundred and somebody sold a hundred, so “more buyers than sellers” is not a thing that can happen.
What delta measures is aggression. Who crossed the spread to get filled immediately, rather than waiting patiently in the book - a genuine distinction with real meaning.
But it is inferred, not reported. No feed states who initiated; the classification is deduced from where the trade printed relative to the prevailing quote, and that deduction can be wrong.
A worked example
Price makes a new high while cumulative delta does not. The reading is that the advance was made without matching aggressive buying, and that it may therefore be weak.
That is a coherent story and it has alternatives. A large seller resting patiently in the book absorbs aggressive buying and produces the same pattern, with price still rising.
So one reading says weakness and the other says absorption. The delta chart is identical in both cases, and nothing in the data selects between them.
Which is the honest description of the tool. It narrows what happened, it does not identify why, and the interpretation added afterwards is the part doing the work.
The data you actually need
Tick-by-tick trades, with quotes alongside. Without the bid and ask at the moment of each print, no classification is possible at all.
Most chart packages do not supply this. The data is sold separately, costs more than end-of-day prices, and a standard candle feed cannot produce delta at any setting.
Futures markets are where it is most available. Centralised venues with a single order book make the data coherent in a way that fragmented equity markets do not.
And fragmentation is the quiet problem in equities. One stock trades across many venues, so the national quote you compare a print against may not be the quote that print executed against.
Footprint and delta charts are the same data drawn differently. A footprint shows the buy and sell volume at each price within a bar; cumulative delta collapses that into one running total. Neither adds information to the other - one is the detail and the other is the summary, and both depend entirely on the same classification step being right.
The original data
On this site’s shared series: median bar range 0.493, ninetieth percentile 1.101, largest bar 2.338. Direction runs average 2.01 bars with a longest of 11. A round trip costs 0.0098, about 2% of the median bar range.
A 2.01-bar average direction run sets the shelf life of a divergence. A disagreement between price and delta observed over a few bars is being read in a series that changes direction roughly that often.
And the cost of acting is unchanged by the sophistication of the tool. At 2% of a median bar per round trip, a method producing several divergences a session pays that repeatedly, which is the practical limit on how finely it is worth reading the tape.
How the classification goes wrong
Trades inside the spread are ambiguous. A print between the bid and the ask has no clear aggressor, and each rule handles it differently.
Fast markets outrun the quote. When prices move quickly, the quote recorded alongside a trade may be stale, and the classification is then made against a price that no longer existed.
Different platforms use different rules. Two delta charts of the same instrument over the same period can disagree, because the classification method is a choice rather than a standard.
So the number is a model output. It is produced by an algorithm applied to raw prints, which is worth remembering when a divergence is being treated as a fact about who was in the market.
When it fails
The characteristic failure is reading absorption as weakness. Price rises while delta does not, and the conclusion drawn is that the move lacks support.
The same picture is produced by a large patient seller. Someone resting size in the book absorbs every aggressive buyer, so delta stalls while price grinds higher - and that is a market with significant demand meeting significant supply, not a market running out of buyers.
A second failure is using it without tick data, where the indicator is estimating from bar data and producing a shape rather than a measurement.
A third is comparing delta across platforms, which use different classification rules.
A fourth is applying it to fragmented equity markets with the confidence it earns on a single futures book.
And a fifth is finding divergences after the fact. On any delta chart, past disagreements that preceded a turn are easy to point to, and the ones that preceded nothing are not marked.
Related
Order flow covers the wider family of trade-by-trade tools. Volume covers the total this splits into two halves. And VWAP covers the other measure built from price and volume together.
The thing nobody says out loud about delta is that every trade has a buyer and a seller in equal size. What delta measures is which side was impatient - who crossed the spread. That is a real and interesting distinction, and it is inferred from where the trade printed relative to the quote rather than reported by anybody.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.