WhitmanTrading

What Is an Order Book? How It Works, Read on a Real Bitcoin Book

An order book is the list of every resting buy and sell order in a market, sorted by price. The best bid and best ask are its top two lines, the only part a price chart shows, and the depth beneath them decides what a large order actually pays.

Covered on this page: TradingView.

An order book is the live list of every resting order to buy and sell one market, sorted by price: bids from the highest down, asks from the lowest up. The two best prices make the quote you see on a chart; everything behind them decides what a larger order actually pays, and none of it promises to stay.

How it forms

Two queues, sorted in opposite directions. Every limit order that does not trade at once is stored on one of two sides. Bids, the orders to buy, are kept highest price first. Asks, also called offers, are kept lowest price first. The exchange’s matching engine owns the list; a broker’s screen shows a copy of it.

A trade is the moment the queues meet. An incoming buy priced at or above the lowest ask, or a sell at or below the highest bid, fills against what is resting there and removes it. Whatever is left over joins the queue on its own side. That is why the best bid always sits below the best ask: if the two touched, they would already have traded. The gap between them is the bid-ask spread.

Price first, then time. A better price goes to the front of the whole queue. Among orders at one price, the earliest fills first. On most venues that is the entire priority rule, and it is why a limit order that joins a crowded price late may never be filled.

What the book never holds. Market orders do not rest, so they never appear; they only remove lines. Hidden and midpoint orders, and the unseen part of an iceberg order, are matched but not displayed. Opening and closing auctions set aside the queue and compute a single clearing price for everyone.

Stop orders stay out of this list until they trigger, so traders guess where they cluster from chart levels, the idea behind liquidity pools; on IWM since 2000, equal lows were traded through within 20 sessions in 13 of 23 cases (56.52%).

Level 1, Level 2 and Level 3

Three depths of the same list. Level 1 is the top line only: best bid, best ask, the size at each and the last trade, which is what most charts are built from. Level 2 adds the prices behind the best ones and the total size resting at each, which is what most people mean by the order book. Reading it in practice is a step-by-step job: how to read the order book found the largest BTC/USD order near the price stayed a median 30 seconds on 28 Oct 2026.

Level 3 lists each order on its own, with an ID, so a reader can tell one large order from fifty small ones at the same price. Coinbase Exchange, for one, publishes its BTC-USD Level 3 book free (checked 2 November 2026). The same list can also be drawn as a vertical ladder, the depth of market window futures traders use, or as a cumulative depth chart.

A worked example

One real book, one second. This site saved Kraken’s public BTC/USD book every five seconds on the night of 1-2 November 2026. The snapshot taken at 05:59:59 UTC (just before 10 p.m. on Sunday in Seattle, 1 a.m. Monday in New York) is the example, chosen by its time before anything in it was read.

A ten-row table of Kraken's BTC/USD book at 05:59:59 UTC on 2 November 2026: five thin asks from $78,250.30 to $78,256.60 above five bids from $78,250.20 down to $78,249.00, with 1.582 bitcoin at the best bid.
The five best asks and bids in Kraken's public BTC/USD book at 05:59:59 UTC, 2 Nov 2026, with the running total outward from the best price. Source: Kraken public order book (raw snapshot kraken-book-262).

The top two lines. The best bid was $78,250.20 for 1.582 BTC; the best ask was $78,250.30 for 0.033 BTC. The spread was $0.10, one tick, about 0.0128 basis points. Behind the best bid sat 0.237 BTC at $78,249.70, 0.001 at $78,249.60, 0.192 at $78,249.30 and 0.001 at $78,249.00. Counted outward from the best price, the asks added up to 0.033, 0.034, 0.037, 0.062 and 0.063 BTC, and the bids to 1.582, 1.819, 1.820, 2.012 and 2.013 BTC. A chart at that second would show one price, about $78,250.

Walk a 1 BTC market buy up the asks. It takes the 0.033 at $78,250.30, then 0.001 at $78,252.60, 0.003 at $78,254.70, 0.025 at $78,254.80, 0.001 at $78,256.60, 0.192 at $78,259.30 and 0.376 at $78,260.00, which is 0.631 BTC so far, and finishes with 0.369 of the 1.004 resting at $78,260.10. Eight price levels. The bill is $78,259.43 for the one coin, $9.13 more than if all of it had filled at the best ask.

Now sell 1 BTC instead. The best bid alone held 1.582 BTC, so the whole sale fills at $78,250.20 in one line. Same market, same second, same size: the buyer paid $9.13 for crossing a thin side and the seller got the best bid on every coin. That asymmetry is invisible on a price chart and obvious in the book.

Bigger sizes, same snapshot. A 5 BTC buy needed 11 levels and averaged $10.40 above the best ask, $52.00 in all; a 5 BTC sell needed 8 levels and averaged $1.40 below the best bid, $6.98 in all.

The original data

The question this section tests. Traders often read a book with more bids than offers near the price as pressure to go up. The how-to page measured how long big orders last; this one asks whether the lean of the whole book near the price lined up with where the price went next.

What was measured. Kraken’s public BTC/USD book (500 price levels a side) every 5 seconds for 75 minutes, 05:38:09 to 06:53:04 UTC on 2 November 2026: 900 snapshots, none missed. For each one, the bitcoin bid within 0.1% (10 basis points) of the midpoint and the bitcoin offered within the same distance gave the lean: (bid − offered) ÷ (bid + offered), from −1 for all offers to +1 for all bids. The outcome was the change in the midpoint 30 seconds later. Leans from −0.2 to 0.2 count as balanced; −0.5 to −0.2 as offer-heavy; 0.2 to 0.5 as bid-heavy; the one window that leaned past −0.5 is left out of the groups. The rule was written before any outcome was counted. The check repeats it on Coinbase Exchange’s full book, every 10 seconds over the same hour: 450 snapshots.

A falling, quiet hour. Bitcoin’s Kraken midpoint slipped from $78,268.65 to $78,070.65, about 0.25%, and the spread sat at $0.10 in 895 of the 900 snapshots. Near the price the book was deep: a median $6.84 million bid and $7.28 million offered within 0.1%.

On Kraken, the lean lined up with the next 30 seconds. Of 894 usable windows, the price rose in 238, fell in 459 and did not move in 197, so it rose in 34.15% of the windows where it moved. When the book was offer-heavy (79 windows, in 37 separate stretches) it rose in 10 and fell in 62: 13.89% of moves. When balanced (774), 34.12%. When bid-heavy (40 windows, in 22 stretches), it rose in 25 and fell in 4: 86.21%. The correlation between lean and the next 30-second change was 0.33.

Paired bars showing how often bitcoin's price rose in the next 30 seconds, by how the order book leaned: on Kraken 13.89% when offer-heavy, 34.12% when balanced and 86.21% when bid-heavy; on Coinbase 50.65% when offer-heavy and 46.31% when balanced.
Share of 30-second windows in which the BTC/USD midpoint rose, of those in which it moved, by the lean of the book within 0.1% of the price, Kraken (900 snapshots) and Coinbase Exchange (450), 05:37 to 06:53 UTC, 2 Nov 2026. Source: Kraken and Coinbase Exchange public order books (m105-btcusd-book-lean-vs-next-30s-kraken-coinbase-2026-11-02.csv).

It faded fast. Thirty-second windows that start five seconds apart share most of their path, so counted windows overstate the evidence. Using only every sixth snapshot, so no two windows overlap, leaves 149: offer-heavy books then rose in 3 of 16 moves, and the correlation was 0.26, with too few bid-heavy cases (4) to say anything. Over 60 seconds instead of 30, bid-heavy books rose in only 16 of 35 moves.

Coinbase did not confirm it. Same rule, same hour: offer-heavy Coinbase books (80 windows) were followed by a rise in 39 and a fall in 38, 50.65% of moves; balanced ones 46.31%; only 6 windows were bid-heavy. The correlation was 0.08. A pattern that shows on one exchange’s book and not on another’s in the same hour is not something to trade on.

And it was tiny next to the cost. Even on Kraken, the average 30-second move after an offer-heavy book was 0.85 basis points, less than $7 a bitcoin. Kraken Pro’s taker fee on its lowest tier is 0.80%, or 80 basis points, per side (fee schedule read 2 November 2026): about 95 times that move. Every snapshot from both exchanges, with its lean and what followed, is in the book lean file.

What this cannot say. One hour on one Sunday night, one market, two venues. It does not test stocks, busier hours, other distances from the price or other horizons beyond the three tried, and nothing here explains why Kraken and Coinbase differed.

Earlier measurements, still on file

What size costs. On a quiet Saturday evening, 24 October 2026, this site saved Coinbase’s and Kraken’s full BTC/USD books 30 times, a minute apart, and pushed imaginary market buys up each one. A $10,000 buy filled within a hair of the best ask on both. A $1 million buy needed a median 65 price levels on Coinbase and paid 2.39 basis points over the best ask, about $238.67; Kraken’s book took it in 24 levels for 0.88 basis points, about $87.97. The snapshots are in the Coinbase book file and the Kraken book file.

How thin the top line can be. In listed options every strike and expiry has its own book. In Cboe’s SPY quotes saved after the 25 September 2026 close, 52.23% of the 12,400 two-sided series showed fewer than 50 contracts at the best ask, and META’s median was 8 contracts. The ten-underlying file has the rest.

Order book vs trade book

The order book is what is waiting; the trade book is what happened. A trade book (also called the tape, or time and sales) records each fill: price, size and time. It never shows a canceled order. The order book shows every order still waiting and never says which will fill. Reading the tape next to the book is how traders tell size that is being traded from size that is only on display.

When it fails

Reading size as intent. A big resting order is a conditional offer, not a forecast. It can belong to a market maker hedging something else, an algorithm that will cancel within seconds, or someone who placed it to be seen.

Spoofing is the illegal version. Placing orders with no intent to trade, to push other people around, is banned in US markets. On 29 September 2020 the CFTC ordered JPMorgan to pay $920.2 million over spoofing in precious metals and Treasury futures. A book read for intent can show intent manufactured to be read.

Seeing one venue and calling it the market. A US stock trades on many exchanges at once, and much of it off exchange: 50.63% of US shares traded in September 2026 were reported off exchange, as the stock exchange page measures from Cboe’s venue figures. One exchange’s Level 2 is one slice.

Assuming the depth will still be there. Displayed size moves while you look, and it thins out under stress, which is when liquidity risk bites.

The job it does honestly is a capacity check. Before sending size, the book answers how much can trade near the current price and how far the price must travel to fill the rest. That question has a factual answer.

Questions people ask

What is an order book in trading?

The exchange’s list of every resting buy and sell order for one market, sorted by price, with the quantity at each price.

How do you read an order book?

Start with the spread, then the size at the best price on the side you would trade against, then how far you would have to go to fill your whole order. The how-to guide goes through it step by step.

Is Level 2 the same as the order book?

Level 2 is the order book shown by price level, with the total size at each. Level 1 is only its top line; Level 3 lists each order separately.

Does the order book show hidden orders or stop orders?

No. Hidden orders, the unseen part of iceberg orders and untriggered stop orders do not appear.

Can the order book predict price?

In this test, not in a way anyone could use. Kraken’s lean lined up with the next 30 seconds, Coinbase’s did not in the same hour, and the moves were a small fraction of a basis point to about one, against fees of 80 basis points.

The bid-ask spread is the book’s top two lines and the cost of crossing them. Order types are the different ways of joining or removing from this queue, and market makers are the participants who keep resting orders on both sides of it as a business. When those resting orders pull back all at once under stress, the danger has its own name: liquidity risk.

What I actually do

Do not read Level 2 for direction. Keep it open for the question it answers well: can this size get out, at something like this price, right now. That is a liquidity question, and the size shown at the best price is often smaller than the order you are thinking of sending.

— 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.