Cumulative Delta and Its Limits

Before you read: Bid/Ask Volume Charts: Reading Volume at the Bid and Ask by Price

What you will learn

• How cumulative delta is built from bar delta and where it resets, including session delta.
• What agreement and delta divergence between price and cumulative delta describe.
• Why cumulative delta depends on data, settings and classification, and what it cannot show.

How is cumulative delta built?

If you kept a running total of each bar’s delta, what would the line show?

CVD adds each selected bar’s buy-category minus sell-category volume to its prior total. With deltas +200, −50, +100 and initial 0, the totals are 200, 150, 250. The interpretation is conditional on classification: reported aggressor data and inferred intrabar categories are not identical measurements. In this lesson, “session delta” means a running total reset at the declared session start; an across-session anchor is a separate recorded choice.

For IDENTICAL increments over the SAME overlap, changing only the initial constant shifts levels, not increments or slope:

Both rise 200, fall 50, rise 100. A value above zero is not the same as rising. If another chart RESETS before observation 3, it shows 100 there rather than 250; joining prior 150 to reset 100 draws a discontinuity, not −50 signed activity during the third bar. Different windows, missing trades, aggregation, classification or reset policies can create other differences; an offset alone cannot reverse the same-interval change.

What do agreement and divergence describe?

Price makes a new high but cumulative delta does not. What changed between the two highs?

Agreement/divergence compares specified observations; it is not an identity of buyers or causes. Suppose, under the SAME running-series anchor and classification, a first fictional price high 105 has CVD 250 and a later higher price high 106 has CVD 200. CVD fell 50 over that defined intervening interval, so its summed signed increments there are −50. This does not separately tell gross buy quantity, gross sell quantity, activity in a differently defined “push,” participant count or cancellations.

Passive liquidity, different activity or classification artifacts are possible hypotheses requiring other evidence, not diagnoses established by the two endpoints. A divergence can persist or be followed by several outcomes; no reversal frequency is measured here. The existing question’s possible explanations are hypotheses to investigate, not observed motives.

Cumulative delta diagram

Illustrative diagram constructed for teaching. Not real market data.

The diagram shows a fictional price path above a cumulative delta panel that starts at zero at the session open. Both lines rise into the first high. On the second, higher price high, cumulative delta makes a lower high, labeled as a divergence, with several possible outcomes listed. All values are made up.

Why do different platforms show different numbers?

If two data feeds classify the same trade differently, can their delta agree?

Delta depends on how each trade is classified as buyer- or seller-initiated. Feeds that deliver every individual trade and feeds that bundle trades together can give different totals. Some platforms classify by the quote at the time of the trade, others by the price change from the last trade. Large orders split into many small fills appear as many prints. These differences add up in a running total, so two cumulative delta lines of the same session are often similar in shape but not identical.

Record feed and calculation mode, not merely the word “delta.” As checked in this course’s bid/ask lesson, TradingView’s default estimate, optional native tick/aggressor mode with fallbacks, and Pine’s documented default footprint calculation must not be generalized into one identical feed. Footprint is a registered trademark of its owner. Actual availability/history/entitlements are not audited here. TradingView’s volume footprint guide (rechecked November 8, 2026) adds that its tick-based mode also applies to its Volume Delta and Cumulative Volume Delta indicators, and that when available ticks do not cover the whole session, session totals and CVD may differ from other sources.

What does cumulative delta leave out?

What part of the order book never shows up in delta?

Delta categories summarize executed quantity, not the unfilled book or canceled interest. Consider a completed fictional bar whose prior reference price is 100: one sell-initiated trade of 5 contracts at 100 is followed by a buy-initiated trade of 2 at 100.25. The last price is higher while delta=2−5=−3 contracts. This stipulated sequence demonstrates coexistence without identifying why liquidity changed.

Cancellations can change available liquidity and coexist with such trades, but they are not counted in delta; CVD alone cannot diagnose pulled offers, absorption, identity, hedging or motives. With inferred categories, even the aggressor interpretation needs qualification. See Bid/Ask Volume Charts: Reading Volume at the Bid and Ask by Price.

What this does not tell you about cumulative delta

• Whether a divergence will be followed by a turn. Not every divergence is followed by a turn, and no published figure says how often one is.
• Who was trading or why. Delta treats every aggressive order the same.
• The resting and canceled orders that shape the book. Delta is a trade-only measure.
• Nothing here suggests buying, selling or holding at any reading.

Method in Practice: divergence means wait, not reverse

Fictional example. John, Harbor Index futures, the prices, times, costs and results are all invented for teaching. This shows how a written plan can be built and tested. It is not a recommendation, and nothing here says how often this idea works.

Try it on a real chart. Test this idea yourself on past sessions with bar replay. Keep your own count, including the days with no trade.

The idea. The lesson says a divergence describes two readings, not a cause, and no frequency is measured. John uses it only as a reason to wait. His breakout plan needs cumulative delta to agree with price. If it doesn't, he doesn't trade the other way. He simply doesn't act yet.

John's plan

Setup

the morning high (08:30–09:30). Session cumulative delta, reset at 08:30, same feed mode every day

Enter if

a 5-minute bar closes above the morning high, and cumulative delta at that close is above its reading at the morning high. Buy.

Wait if

delta is lower (a divergence). Act only on a later close that passes both tests.

Wrong if

a 5-minute bar closes below 101.25

Manage

take half at 101.95, the rest at 102.35 or by 14:30

Risk

$150 maximum, costs included

Three outcomes

• ✅ Worked: price and delta agreed, and the breakout carried → +$106

• ❌ Failed: they agreed, and price still fell back → −$82, inside the limit

• ⏸ No trade: divergence, and no later close passed both tests → $0

Show the full replays

Item	Plan
Context	Morning high 101.40 (08:30–09:30 CT). Session cumulative delta reset at 08:30, feed's reported aggressor side
Entry condition	A 5-minute bar closes above 101.40 with cumulative delta above its reading at the morning high. Buy 2 at market after that close
Wait condition	Close above 101.40 with lower cumulative delta: no entry on that bar
Wrong if	A 5-minute bar closes below 101.25. Exit at market after that close
Management	Take profit on half at 101.95 (limit) and move the exit for the rest to the entry price (stop). Rest at 102.35 (limit). Time limit 14:30 CT
Risk cap and size	$150. Exiting just under 101.25 is planned to cost about $45 per contract, so 2 contracts
Not trading if	Scheduled major release within the hour, or no setup by 12:00 CT

Replay A: worked

Time (CT)	What happened	John
09:20	Morning high 101.40, delta +1,800	Marks both
10:10–10:15	Closes 101.52, delta +2,300	Both tests pass: buys 2 at market, filled 101.57
10:15–11:30	Rises, trades 101.95 at 11:30	Sells 1 at 101.95 (limit). Exit for the other moves to 101.57
11:30–13:45	Never back to 101.57, trades 102.35 at 13:45	Sells the last 1 at 102.35 (limit)

Result after costs: +0.38 point = +$38, +0.78 point = +$78, commissions −$10. Net +$106. Journal: "Price and delta both made new highs. Both exits were limits."

Replay B: failed

Time (CT)	What happened	John
09:25	Morning high 101.40, delta +1,600	Marks both
10:40–10:45	Closes 101.48, delta +2,050	Both pass: buys 2 at market, filled 101.53
10:45–10:50	Drops, closes 101.22	Wrong-if met: sells 2 at market, filled 101.17
10:50–15:00	Rotates 100.90–101.40	Done for the idea

Result after costs: 2 × −0.36 point = −$72, commissions −$10. Net −$82. Journal: "Delta agreed, and price fell back anyway. Agreement was a condition, not a forecast."

Replay C: no trade

Time (CT)	What happened	John
09:15	Morning high 101.40, delta +1,900	Marks both
10:20–10:25	Closes 101.50, delta +1,500	Divergence: waits
10:25–12:00	Rises to 101.62, falls to 101.10, ends near 101.40. No close passes both tests	No trade. 12:00 cutoff

Result: $0. Journal: "Divergence, so I waited, and nothing qualified. One day can't tell me whether waiting helps. The lesson says no frequency is measured, and my review counts every wait."

Test it yourself

• Fix the reset time, feed and calculation mode, and write them down (this lesson's table shows a reset changes the line).
• Replay every close above the morning high. Split the results by "agreed" and "diverged".
• Check what happened after the waits too. Waiting has a cost when price rises anyway.

Where it breaks

• Agreement isn't a promise (the failed replay).
• Two platforms can disagree on delta for the same bars.
• Delta leaves out the order book and canceled orders completely.

Calculators: position size · risk/reward · expectancy

Reflection question: Does your plan say what a warning sign makes you do: wait, reduce, or do the opposite?

Check yourself

1. Bar deltas are +300, −100, −250 and +50. What does the cumulative line read after each bar? Hint: keep a running total from zero.
2. Why can two platforms show different cumulative delta for the same session? Hint: reset point and trade classification.
3. Price makes a higher high on a lower cumulative delta high. Name two explanations other than “sellers are about to win”. Hint: limit-order buying, or a quieter market.
4. Can a 1000 offset alone make series A fall while B rises on observation 3? What can a reset change?
5. Does a higher last price with −3 delta prove offers were canceled?

Last reviewed November 9, 2026. Spotted a mistake? Report an error.