How to read order flow: a six-question checklist

Six questions, in this order

1. Where is price? Near a recent high or low, at a round number, or in the middle of a range? The same signal means more at an edge than in the middle.
2. Who is aggressive? Delta and the tape: are takers mostly buying or mostly selling?
3. Do price and CVD agree? If they move together, aggressive traders are carrying the move. If not, look closer.
4. Is something absorbing? Heavy flow that does not move price means waiting orders are taking it — on which side?
5. What does the book show near the price? Where is size waiting, and does it stay when price comes?
6. What would change the reading? Decide in advance what you would need to see to know you are wrong.

One real moment, question by question

The checklist, on a real moment — BTC/USDT on 27 October 2026 at 15:33 UTC, on 1-minute candles. Price was at 80,380, 0.40% below the 80,705 high of 15:24. In those 9 minutes takers bought 32.3 BTC more than they sold, yet price fell. In the recorded order book, the asks within $20 of the price grew from 18.1 BTC at 15:25 to 31.8 BTC at 15:33 while the bids shrank from 29.1 to 12.8 BTC. From 15:34 the book turned — at 15:37, 29.1 BTC of bids against 13.0 BTC of asks — and price climbed to 80,694 by 15:50, then slid to 80,452 by 16:12.

What the example shows

The answers did not forecast anything. They said where the pressure was, what was holding price back, and what to watch for: the asks being used up, or the bids thinning further. After the book turned, price rose — for a while.

It also shows why CVD alone is not enough. Over the whole session takers bought 488 BTC more than they sold — delta was positive in 65 of the 80 minutes — and price ended up only 0.22% higher. On a spot market, takers can be net buyers for hours while waiting sell orders supply all they want. CVD tells you who is in a hurry; the book and the price tell you whether it is working.

Make it a habit

• Ask the same questions in the same order every time, so a strong signal does not make you skip a step.
• Write down the answer to question 6 before you act. A wrong reading caught early is a small loss; one explained away is a large one.
• Then practise on real history: ten real moments in the drill.

Check yourself · 1 of 2Over nine minutes price fell 0.4% while takers bought 32 BTC more than they sold. Which question helps most next?

AWho is aggressive?BIs something absorbing — and on which side?CWhere is price?

Show the answer

You already know who is aggressive (buyers) and that CVD and price disagree. The question is what is taking the buying: on 27 October, the asks near the price had grown from 18.1 to 31.8 BTC while the bids thinned.

Check yourself · 2 of 2Why decide what would change your reading before you act?

AIt makes the reading more accurateBSo a wrong reading is recognised early instead of explained awayCExchanges require it

Show the answer

Readings are often wrong — the divergence counts in lesson 4 showed how often. Knowing in advance what would prove you wrong turns a surprise into a plan.

Takeaways

1. Ask six questions in order: where price is, who is aggressive, whether CVD agrees, what is absorbing, what the book shows, what would change your mind.
2. Disagreement between price and flow is information: find the side that is absorbing.
3. CVD shows who is in a hurry, not whether it is working — the book and price tell you that.
4. Decide what would prove you wrong before you act.

Open BTC in the terminal and run the checklist live

Further reading

• Diff. depth stream — how the live order book was recorded — Binance WebSocket docs ↗
• Kline data with taker buy volume — Binance Spot API docs ↗