WhitmanTrading

How to Read the Order Book (Level 2): 7 Steps, and How Long Big Orders Last

To read the order book, treat each level as resting limit orders that can be canceled at any moment. It shows what participants say they will do rather than what they have done, which is why large visible orders are the least reliable part of it.

To read the order book, check the spread first, then the size resting near the price, and treat every order on it as an intention that can be canceled at any moment. In 40 minutes of bitcoin order books on 28 Oct 2026, the largest order near the price stayed on screen a median 30 seconds, and 79 of the 80 that vanished went before any trade reached their price.

The order book lists the limit orders currently resting at each price, bids on one side and asks on the other. It is the closest thing to a live picture of supply and demand, and everything on it is provisional until it is filled.

Before you start

An understanding that resting orders are intentions and can be canceled instantly. Nothing on the book is a commitment, and everything else follows from that.

A decision about what you are looking for, because the display updates faster than you can read it. Spread width, size near the best prices, or size at one level: pick one question.

An instrument liquid enough that the book means anything. In something thinly traded the book is a handful of orders and tells you very little.

The steps

1. Read the spread first

The gap between the best bid and the best ask is what a round trip costs immediately. With a bid of $50.00 and an ask of $50.02, buying and selling straight back costs $0.02 a share, or $2 on 100 shares, before commission. The bid-ask spread page covers how it moves.

2. Read the size near the best prices

Size resting a few ticks from the best price is what your order will actually meet. Orders twenty levels out are unlikely to be there by the time price arrives.

3. Note the largest order near the price, then look again a minute later

A single large order is the easiest thing on the screen to notice and the easiest thing in the market to cancel. Watch whether it is still there a minute later before it shapes any decision.

4. Compare the book with the trades

The book says what people intend; the tape says what happened. When the two disagree, the trade record is the one that already occurred.

5. Check your own size against the depth

If your order is a meaningful fraction of what rests at the prices it would consume, your order moves the price itself, and the book is describing a market you are about to change.

6. Keep stops away from visible clusters

Prices where orders visibly pile up are the obvious ones to reach. A stop a little beyond the obvious concentration costs a small distance and takes it out of the crowd.

7. Take structural levels from the chart

The book shows this instant. A level price has respected three times over a month is a different kind of evidence, and it survives the next thirty seconds.

How to tell it worked

The spread was checked before every entry, every time.

Your order was less than 10 percent of the size resting at the prices it would consume.

0 stops sat at a visible cluster of orders.

And any large order you noticed was checked again 60 seconds later before it affected a decision.

The original data

Sample: the public BTC/USD order book on Kraken (its 500 best prices a side) and its full trade record, saved every 15 seconds from 14:57 to 15:37 UTC on Wednesday 28 October 2026, 160 snapshots. Coinbase Exchange’s book (every price level) and its latest trades were saved in the same rounds as the check. Bitcoin is used because both exchanges publish their books free and without an account; a stock’s book on one exchange is a different market with the same mechanics.

The rule. In each snapshot, on each side, the largest single price level within 0.25% of the midpoint between the best bid and best ask was marked. Each marked price was then followed through every later snapshot until the size at that exact price fell below half of what was first seen. If any trade printed at that price or through it in the gap before it went, a trade reached it; if not, the size was canceled or moved before price got there.

Horizontal bars for the largest BTC/USD order within 0.25% of the price on 28 Oct 2026: on Kraken 41.46% were gone by the next look 15 seconds later, 30.49% within a minute, 21.95% within five minutes, 3.66% lasted longer and 2.44% were still showing; on Coinbase 35.37%, 47.62%, 13.61%, 0.00% and 3.40%.
How long the largest order within 0.25% of the price stayed at least half its size, BTC/USD, Kraken and Coinbase, snapshots every 15 seconds, 14:57 to 15:37 UTC, 28 Oct 2026. Source: Kraken and Coinbase Exchange public order books (m101-btcusd-largest-resting-orders-kraken-coinbase-2026-10-28.csv).

The finding: in this sample, the biggest order near the price behaved like a quote that moved, not a wall that waited. The rule marked 82 largest orders on Kraken, with a median size of $2.00 million of bitcoin. 34 (41.46%) were below half their size by the next look, 15 seconds later; 25 (30.49%) went within a minute; 18 (21.95%) within five minutes; 3 (3.66%) lasted longer; 2 (2.44%) were still showing at the last snapshot. The median order stayed about 30 seconds. Of the 80 that went, 79 went with no trade at their price, while price was still a median 20 basis points away, about $160 at a bitcoin price near $80,500. The one exception was an ask of 17.16 BTC at $80,593.90 that stayed 16.5 minutes until trades reached it. Of the 82, 73 were within 1% of exactly $2.0 million, which looks like one automated quote sized in dollars and re-priced as the market moved.

The check. Coinbase’s book, saved in the same rounds, shows the same pattern with smaller orders: 147 largest orders with a median size of $224,448, 52 (35.37%) gone by the next look, 70 (47.62%) within a minute, 20 (13.61%) within five minutes, 0 (0.00%) longer and 5 (3.40%) still showing at the end. None of the 142 that went was reached by a trade first, and the median order again stayed about 30 seconds. The check covers a second exchange’s book in the same 40 minutes, not a second day.

What it cannot show. A snapshot every 15 seconds misses anything shorter, so the times are upper bounds. An order that vanished may have been moved to a new price rather than withdrawn: in a market that keeps moving, the people quoting large size re-price all the time. That is the point for a reader: the size was a quote, not a promise to stay at that price. Forty minutes on one weekday morning in New York is one slice of one market. And 73 of Kraken’s 82 orders looked like the same dollar-sized quote, so the count may describe one participant’s quoting more than 82 separate traders. The figures describe this sample, not how long large orders last in general.

Across the 24,971 unique videos in the finance search study this site keeps, 3 mention the order book in the title, at a median of 2,367 views across 3 channels. The mechanism every order passes through has almost no instructional coverage.

Why the visible book is incomplete

Some size is hidden. Iceberg orders show a small portion and refill, and some order types do not display at all, so what you see is a subset.

And for stocks it is one venue’s book. US share trading is spread across many venues, so the depth on one screen is a slice of the total. The depth of market view aggregates what a feed can see.

What it is useful for

Deciding how to execute. Whether to use a limit or a market order, and whether your size will move the price, are questions the book answers well.

Measuring the real cost of a trade. The spread and the size near it tell you what entering and leaving will cost.

Not forecasting. Market makers quote both sides to earn the spread and manage inventory, so much of the size near the best prices reflects a business being run, not a view on direction. The sample above shows how quickly that size moved in one 40-minute window.

When it fails

The failure is treating a large resting order as support. A big bid appears a few ticks below, which reads as a floor. A position goes on against it with a stop just underneath. As price approaches, the order is gone, and the level that justified the trade disappears in the moment it was supposed to matter. In the Kraken sample 79 of the 80 largest orders that disappeared went before any trade reached their price.

The second failure is reading size far from the best prices. It is unlikely to be there when price arrives.

A third is an order too big for the book. You become the price movement.

A fourth is stops at visible clusters. They are the obvious prices to reach.

A fifth is treating one venue’s book as the market. Trading is spread across venues.

And a sixth is using the book to predict direction. It is a snapshot of intentions.

Questions people ask

What is Level 2?

The order book display: every bid and ask price with the size resting at each. Level 2 covers the name and what different feeds show.

Can large orders on the order book be fake?

They can be canceled at no cost, which is not the same as fake. On Kraken’s BTC/USD book on 28 Oct 2026, 80 of the 82 largest orders near the price disappeared within the 40 minutes, and 79 of them went before any trade reached their price.

What does a big bid wall mean?

That someone is showing size at that price right now. It says nothing about whether the size will still be there when price arrives; the largest order near the price lasted a median 30 seconds in the sample.

How do I read the order book for crypto?

The same way as for stocks: spread first, then size near the best prices, then check the trades. Crypto exchanges usually publish the full book free, which is why the data on this page uses one.

Order book covers the structure in detail, including what a large market order costs as it walks the book. Level 2 is the common name for the display. And depth of market is the aggregated view of the same data.

The practical check

The lesson was that a large order sitting on the book is not support. It is a message, and it can be canceled the instant price gets close. What actually happened is in the trade record, not in the list of things people say they might do.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.