How to Read a Crypto Order Book

What an order book is

An order book is the list of unfilled limit orders resting on a venue for one trading pair, grouped by price. Exchanges expose it directly. Binance's depth endpoint returns a lastUpdateId together with a bids array and an asks array, where each entry is a price and quantity pair given as strings, and requests can ask for anywhere from 1 to 5,000 levels with a default of 100.

That structure tells you what the book is and what it is not.

What it is: a record of orders that have already been placed and have not yet been filled or cancelled, on one venue, for one pair, at one instant.

What it is not: a record of all available liquidity. Every order that will be submitted in the next second is absent, and in liquid markets the flow arriving continuously is far larger than the stock sitting displayed. The book is a photograph of a river.

Bids asks and the spread

The book has two sides and they are asymmetric in what they mean.

• Bids are resting buy orders. The best bid is the highest price someone is currently willing to pay. If you sell immediately, you sell into the bids.
• Asks, also called offers, are resting sell orders. The best ask is the lowest price someone is currently willing to accept. If you buy immediately, you buy from the asks.
• The spread is the best ask minus the best bid. It is the immediate round-trip cost of entering and exiting at market.
• The mid price is the average of the two. It is a convention, not a price you can trade at.

Anyone placing a resting limit order is a maker. Anyone crossing the spread to execute immediately is a taker. This distinction determines your fee tier on most venues, which is covered in maker and taker fees.

Reading a book snapshot

Here is a book for a liquid pair, with one feature that deserves attention.

Read it in three passes.

Pass one: the spread. Best bid is $3,000.00 and best ask is $3,000.50, so the spread is 50 cents on a $3,000 asset, about 0.017%. That is tight, which tells you competitive market making is present.

Pass two: the cumulative totals. The total column is the running sum. To buy everything up to and including $3,002.00 you would need 47 ETH. That is the actual liquidity available within 5 cents of a percent of the mid.

Pass three: the shape. Of the 447 ETH resting up to $3,005.00, the single level at $3,005.00 holds 400 ETH, which is 89% of all displayed ask-side depth in that range. The book is not deep. It is shallow with one large object sitting in it.

That concentration is the whole lesson, and the next section shows why.

What happens when a large order arrives

A market order consumes resting orders from the best price outward until it is filled. Walking it through by hand is the only way to see what the displayed depth is really worth.

Working it through

Take a market buy for 200 ETH against the book above.

Scenario A: the wall is genuine and stays.

The average fill is $3,004.13, about 0.12% above the best ask. The last print is $3,005.00.

Scenario B: the wall is cancelled as the order starts consuming the book. Behind it, the genuine resting orders are 30 ETH at $3,012.00, 40 ETH at $3,020.00, 60 ETH at $3,035.00 and 100 ETH at $3,060.00.

The average fill is $3,023.51, and the last print is $3,060.00.

The difference: $19.38 per ETH, or $3,875 on a 200 ETH order, which is about 0.65% of notional. The order also printed a high 55 dollars above where it would otherwise have stopped.

What this actually tells you

The wall was not liquidity. It was the appearance of liquidity. Nothing about the two scenarios differed except whether a resting order stayed in place, and a resting order can be cancelled at effectively zero cost. The trader who sized their order by looking at the book saw 447 ETH of depth and received the execution of a book with 47.

Concentration is the risk, not depth. A book showing 447 ETH spread evenly across twenty levels behaves completely differently from one showing 447 ETH with 400 at a single price, even though both report the same total. The summary number hides the thing that matters.

Why walls disappear precisely when tested. A large resting order is a standing offer to trade at that price to anyone who wants it. If the market is moving toward it, the person who placed it has the most information-rich moment to reconsider, and cancelling is free. Genuine size therefore rarely sits displayed for long; it is worked in pieces, hidden in iceberg orders, or executed away from the book entirely.

The actionable version. Before sizing any order against displayed depth, apply a haircut test: recompute your expected fill assuming the single largest visible level is not there. If the answer is acceptable, the order is safely sized. If removing one level changes your execution materially, you are relying on a stranger's revocable claim, and you should either reduce size, split the order, or use a limit order and accept the risk of not filling. The order type choices are covered in crypto order types explained, and the resulting execution cost in crypto slippage explained.

Depth charts and what they add

A depth chart plots cumulative quantity against price, producing two curves rising away from the mid: bids to the left, asks to the right.

It adds one thing over the raw book, and it hides another.

What it adds: shape at a glance. A steep curve means quantity accumulates quickly as you move away from the mid, so large orders can be absorbed within a small price range. A flat curve means the opposite. Asymmetry between the two sides is visible immediately.

What it hides: the concentration that matters. A depth chart is cumulative, so a single 400 ETH level appears as one vertical step, and the eye reads the area under the curve rather than the step. On a compressed chart that step can look like gradual accumulation. The raw book shows you that 400 units sit at exactly one price; the depth chart smooths it into a slope.

Use the depth chart to judge overall shape and the raw book to judge whether that shape is built from many orders or from one.

Thin books and thick books

The single most important property of thin books is that their thinness is worst exactly when you most need liquidity. Market makers widen or withdraw when volatility spikes, because the risk of being adversely selected rises. So the depth you measured in calm conditions is not the depth that will be there during the move you want to trade.

How market makers use the book

Market makers quote on both sides simultaneously and aim to earn the spread while remaining roughly flat in inventory. Understanding their constraints explains most of what you see in a book.

They must be able to cancel. A maker who cannot withdraw a quote when conditions change is guaranteed to be picked off by better-informed flow. Cancellation is not abuse; it is the basic risk control that makes quoting possible at all. This is why the book is inherently revocable and why treating displayed depth as committed is a category error.

They manage inventory, which skews the book. A maker who has accumulated a long position will quote more aggressively on the offer side to reduce it. A visibly lopsided book is frequently an inventory artefact rather than a directional opinion held by anyone.

They hide size. Showing genuine large size invites everyone to trade against it on their terms. Real institutional size is worked through iceberg orders that display a fraction of their quantity, through algorithmic slicing over time, or off-book entirely. This is the deepest reason not to read the visible book as the whole market: the largest participants have the strongest incentive not to appear in it.

They quote across venues. A maker's risk is aggregated across every exchange they quote on, so their behaviour in the book you are watching depends on conditions in books you are not.

Spoofing and layering

The practices have a precise legal definition worth quoting, because the wording identifies exactly what makes them undetectable to a reader of the book.

United States law prohibits, on registered entities, conduct that is, is of the character of, or is commonly known to the trade as spoofing, which the statute defines in the same sentence as bidding or offering with the intent to cancel the bid or offer before execution.

Layering is the related pattern of placing multiple orders at several price levels on one side to create the appearance of pressure, with the same intention of cancelling before execution.

What this means for you as a reader of the book

Intent is the element, and intent is invisible. The statutory definition turns on the intent to cancel before execution. Nothing in the displayed book records intent. A 400 ETH offer placed by a genuine seller and a 400 ETH offer placed to create an impression are byte-for-byte identical in the depth feed. No amount of staring at the book distinguishes them, and no indicator built on book data can either.

Cancellation on its own is not evidence of anything. Legitimate market makers cancel constantly, as described above, and cancellation rates in electronic markets are extremely high for entirely proper reasons. Concluding that a withdrawn order was manipulative because it was withdrawn is not a supportable inference.

Venue coverage is uneven. The statute quoted applies to registered entities under US commodities law. Crypto spot venues sit in a patchwork of regimes depending on jurisdiction and product, so the protections and surveillance you can assume differ by where you trade. That is a question to answer about your venue rather than about the book.

The practical response is defensive, not detective. Do not try to identify manipulation from the book, because you cannot. Instead, assume any single large level may not be there when you arrive, and size accordingly. The haircut test described earlier makes you indifferent to whether a wall was genuine, which is a far better position than being right about it.

What the order book cannot tell you

• It cannot tell you intent. Displayed orders carry no information about whether they will be honoured.
• It cannot show hidden liquidity. Iceberg orders, algorithmic flow and off-book execution are all absent, and they are usually the largest components.
• It cannot show other venues. Depth on other exchanges and in correlated pairs does not appear.
• It cannot predict direction. A heavier bid side is at least as often an inventory artefact as a directional signal.
• It cannot survive its own age. The book changes continuously, so a snapshot is stale on arrival.
• It cannot tell you what depth will be during stress. Every measurement taken in calm conditions is an upper bound.

How to read a book in practice

Conclusion

An order book is a precise object with a narrow meaning: the resting limit orders on one venue, for one pair, at one instant. Everything about how to use it follows from taking that definition seriously. It excludes all flow that has not yet arrived, all hidden and iceberg quantity, all off-book execution, and every other exchange. In liquid markets those omissions are larger than what the book displays.

The most consequential property is that resting orders are revocable at no cost. The worked example makes the size of that gap concrete: a 200 ETH market buy against a book displaying 447 ETH of ask depth fills at an average of $3,004.13 if the 400 ETH wall holds, and at $3,023.51 if it is withdrawn as the order arrives. That is $3,875 of difference on one order, produced entirely by whether a stranger left an order in place. A wall is a claim, not a commitment, and the person who placed it has both the right and the incentive to reconsider at exactly the moment it would be tested.

That is why the practical technique is defensive rather than analytical. You cannot identify spoofing from the book, because the statutory definition turns on the intent to cancel before execution and intent does not appear in a depth feed. A manipulative order and a genuine one are identical bytes. What you can do is make yourself indifferent: recompute your expected execution with the largest visible level removed, and size so that the answer is still acceptable. That single habit converts an unanswerable question about someone else's honesty into an answerable question about your own order size.

Beyond execution, the book is a good liquidity gauge and a poor directional one. Spread width and cumulative depth tell you what trading will cost and whether conditions are deteriorating. Side imbalance mostly tells you about a market maker's inventory. And every depth measurement taken in a calm market is an upper bound on the depth that will exist during the move you actually want to trade.

Frequently asked questions

Sources and further reading

Primary sources:

• United States Code Title 7 Section 6c, prohibition on disruptive practices
• Binance — Spot API order book depth endpoint

Related CoinBeaver articles:

• How to read crypto market signals
• What trading volume tells you
• Crypto slippage explained
• Crypto order types explained
• Maker and taker fees
• How crypto liquidations happen
• DEX and CEX compared

This article is educational and is not financial advice. The ETH/USDT book, the levels behind the wall, and both fill calculations are constructed illustrative figures chosen to demonstrate how a market order consumes depth, not observations of any real market. The spoofing definition quoted applies to registered entities under United States commodities law; the regulatory status of any particular crypto venue differs by jurisdiction and product, and should be checked directly with that venue.