How to Read a Crypto Order Book & Spot Whale Activity
This walkthrough covers reading an order book level by level, spotting a genuine wall and judging whether it will hold. In one live snapshot a single bid held $7.7M, 21% of all depth within 2% of mid.

Introduction
The best bid in a live bitcoin order book held $2,299. One cent higher, the best ask held $436,585. Both came from the same snapshot, taken at 18:50:09 UTC on 10 September 2026. That 190-fold gap is the first thing an order book teaches anyone who looks past the top row. Below that row sat a single resting bid worth $7,715,482, more than a fifth of everything the buy side displayed within 2% of the price. This article reads one real snapshot from top to bottom. What the levels show, how to find the walls, what the imbalance between the two sides is worth, and how much of it was still there 30 seconds later. Every figure below was measured rather than quoted.
Key Takeaways
- The best bid held $2,299 against a best ask of $436,585 at prices one cent apart, so the top of the book describes almost nothing about the market beneath it.
- Seven of the first eight bid levels together held less than $2,500, which is queue behaviour rather than a thin market.
- One resting order of 100.0711 BTC, worth $7,715,482, made up 21.2% of all bid depth within 2% of the mid price.
- Four of the five largest bids sat on round hundred-dollar prices, because human and algorithmic order placement anchors on the same figures.
- Re-reading the same book 30 seconds later, 94.4% of 5,000 bid levels were unchanged and 4.7% had gone, in completely calm conditions.
What Does a Crypto Order Book Actually Show?
An order book is a live list of every resting buy and sell offer on a venue, each with a price and a size. That is precisely the information an OHLCV candle compresses into four numbers ↗ and then discards. The book is the layer underneath the chart, and it answers a different class of question.
Rows, sides and sizes
Every row in a book is one price level carrying the total quantity offered there. Bids sit below the current price and asks sit above it, sorted outward from the middle. A single row says two things: at what price someone is willing to transact, and how much they are willing to transact there. Neither of those facts survives into a candle. Exchanges publish this as a snapshot through a public endpoint, usually with a depth limit. The snapshot describes the instant it was taken rather than any period. Reading it is a matter of reading a list, and the difficulty lies entirely in knowing which parts of the list carry information.
What the candle threw away
A candle records where price went and how much traded. The book records where price could go and what waits at each step along the way. Those are complementary rather than competing views. A candle cannot tell a reader whether a rally ran into a large resting offer or ran out of buyers. Both produce the same wick. The book shows the offer sitting there before the move, which is the entire reason traders watch it. It also shows the imbalance between the two sides, the size of individual participants, and how quickly the whole structure changes.
Data current as of September 2026.
The first thing most readers look at is also the least informative part of the list.
Why Is the Top of the Book Almost Always Tiny?
The best bid in a live bitcoin book held $2,299. The best ask held $436,585. That is a ratio of 190 between the two halves of a single quoted spread. It explains why quoting the top level tells a reader almost nothing about the market underneath it.
The best bid was worth $2,299
At 18:50:09 UTC on 10 September 2026, Binance BTC/USDT showed a best bid of $77,110.00 for 0.02982 BTC. The best ask was $77,110.01 for 5.66184 BTC (Binance API, 2026-09-10) . In dollar terms that is $2,299 resting on one side and $436,585 on the other, at prices one cent apart. The first eight bid levels held $2,299, $54, $5, $100, $6, $5, $2,181 and $5. Seven of the first eight bid levels together came to less than the price of a laptop. Anyone reading only the top of that book would conclude the market was almost empty. Anyone reading only the ask side would conclude the opposite.
Why the first level is dust
Tiny top-of-book sizes are a structural feature rather than a sign of a thin market. Price-time priority rewards being first at a price. Participants post small orders to hold a queue position cheaply, then cancel or add as conditions change. Real size sits further back, where it is less likely to be picked off by a fast-moving price. The consequence for a reader is direct. The spread describes the first fraction of a coin, and the sizes at the top describe queue behaviour rather than market capacity. Both are real and neither generalises to the order a trader intends to send.
Everything worth reading sits below that first row.
How Do You Read Depth Beyond the First Level?
A book is read downward as a running total rather than across as a price list. The useful question is never what sits at any one level, but how much size has accumulated by the time a given price is reached. That accumulated quantity is the shape worth reading.
Reading downward, not across
Scanning individual rows produces noise, because sizes at adjacent levels vary wildly and mean little in isolation. Accumulating them produces the shape that matters. Starting at the best bid and adding downward gives the total value available above each price. That is the quantity a market sell order would consume before reaching that level. Most exchange interfaces display this as a cumulative column beside the raw sizes. The depth chart many venues draw is the same running total plotted as a curve. A steep curve means size is concentrated near the price. A shallow one means an order has to travel to find it.
Cumulative size is the real shape
Two books with identical top levels and identical spreads can accumulate at completely different rates. That is why execution cost varies between venues which look the same on a quote. Converting that accumulated shape into the cost of a specific order is a separate exercise. Its arithmetic is covered in the companion article on crypto liquidity ↗. For the purpose of reading a book, the shape is enough. It says whether the market is holding size close in, and at what distance the offers thin out. That is a structural reading rather than a measurement, and it is what the rest of this article builds on.
Within that accumulated shape, single orders occasionally dominate everything around them.
What Is a Wall and How Do You Spot One?
A wall is a single resting order large enough to dominate everything around it. The live snapshot held an unmistakable one: 100.0711 BTC at $77,100.00, worth $7,715,482, sitting one basis point below the mid price. Nothing else on the bid side came close to it.
One order, a fifth of the book
That single order represented 21.2% of all bid depth within 2% of the mid price (Binance API, 2026-09-10) . One participant, at one price, held more than a fifth of everything the buy side displayed across a two-percent band. The scale of it is what makes it a wall rather than a large order. In a book where typical levels hold a few dollars to a few thousand, an order worth $7.7M is three orders of magnitude out of family. It appears on any depth chart as a single flat shelf. Spotting one requires no tooling beyond sorting the levels by value and looking at the top of the list.
Telling a wall from a cluster
A wall is one order at one price. A cluster is many orders spread across a band, which produces a similar total but behaves differently. The distinction matters because a cluster cannot be removed by a single cancellation, while a wall can vanish the moment its owner changes their mind. Public depth data does not always separate the two, since some venues aggregate all size at a price into one row. Where the data is aggregated, a very large level might be one participant or fifty, and a reader cannot tell which. The honest approach is to treat any single oversized level as a wall for risk purposes and to remember that its composition is unknown.
Where those large orders sit turns out to follow a pattern.
Why Do Large Orders Cluster at Round Numbers?
Four of the five largest resting bids in the snapshot sat on round hundred-dollar prices. That is not coincidence. The reason it happens also explains why those levels tend to matter rather more than their raw size alone would justify.
Round numbers collect orders
The five largest bids sat at $77,100.00, $76,500.00, $76,458.00, $76,600.00 and $76,800.00 (Binance API, 2026-09-10) . Only $76,458.00 broke the pattern. On the ask side the largest order sat at exactly $78,000.00, holding $1,051,063, which is the same behaviour mirrored above the price. People choose round numbers when deciding where to buy, and the algorithms built to work human orders inherit the same anchors. Stop levels, take-profit levels and manual limit orders all bunch at the same places, so size accumulates there without anyone coordinating it. The clustering is a property of how decisions get made rather than anything about the asset.
What the clustering is for
Because orders gather at round prices, those prices become harder to move through. The loop is self-reinforcing rather than a mystical property of the number. A reader can use this practically. When scanning a book, round levels are where to look first for accumulated size. A large order at an unremarkable price like $76,458.00 more likely reflects one participant's calculation than a crowd anchoring on a familiar figure. Neither reading is a prediction. Both are useful context for interpreting what the book is showing. A round-number wall says a crowd converged on a familiar level, while an odd-number wall says somebody chose that price deliberately.
Data current as of September 2026.
Counting the two sides against each other gives a different kind of reading.
What Does Order Book Imbalance Actually Tell a Trader?
The snapshot showed $36,373,788 of bids against $21,212,002 of asks within 2% of mid, an imbalance of 1.71 to one. That describes displayed intent at one instant, and it predicts direction far more weakly than most readings of it assume.
Counting both sides
Imbalance is computed by summing depth on each side within a stated band and dividing. The band matters as much as the result. Widening it pulls in orders parked far from the action, and narrowing it reduces the calculation to queue noise. Within 2% of mid, this book leaned toward buyers by 1.71 to one (Binance API, 2026-09-10) . Read plainly, more capital was displayed wanting to buy near the price than wanting to sell near it. That is a fact about the snapshot and a reasonable thing to note.
What imbalance does not predict
The temptation is to read 1.71 as bullish, and that step is where most order book analysis goes wrong. Displayed orders are the ones participants chose to show, so a heavy bid side can equally reflect sellers hiding their intent and buyers advertising theirs. Imbalance also says nothing about the aggressive flow that moves price, which arrives as market orders and is invisible in a resting-order snapshot. Order-flow measures built on executed trades address that question and are a separate subject with their own literature. Treating a static imbalance as a directional signal attributes predictive power to a number that only describes what was on display.
Any reading of the book, imbalance included, comes with an expiry.
How Stable Is the Order Book You Are Looking At?
Taking the same book twice, 30 seconds apart, answers the question directly rather than by assertion. Of 5,000 bid levels within 2% of mid, 94.4% were unchanged, 0.9% had changed size and 4.7% had disappeared entirely. The result is more reassuring than expected.
Thirty seconds of the same book
Between 18:50:09 and 18:50:40 UTC, 4,720 of 5,000 bid levels within 2% of mid were still present at exactly the same size. Another 46 had changed size and 234 had gone (Binance API, 2026-09-10) . Total bid depth within that band moved from $36,373,788 to $35,954,196, a change of 1.2%. The $7,715,482 wall was still sitting at $77,100.00, unchanged. A calm bitcoin book is therefore stable enough that a reading taken half a minute ago is broadly still true. That is more persistence than the usual warnings about book fragility suggest.
Reading a snapshot as a snapshot
Stability in calm conditions says nothing about stability under stress, and the two differ enormously. The measured 4.7% attrition happened with no news and no volatility event, purely as market makers repositioned around a quiet price. It is the baseline rate of change for a calm book rather than a sign of anything unusual. During a fast move that fraction rises sharply, because the same participants cancel and requote rather than absorb a price gap at stale levels. The practical rule is to treat any book reading as describing the last few seconds and to refresh before acting on it. Distrust any conclusion that depends on the book still looking this way in a minute.
Persistence over half a minute is not the same as a commitment.
Can You Trust a Wall to Hold Its Price?
A resting order carries no obligation to remain in the book. The $7.7M wall in the snapshot survived 30 seconds untouched. That proves it existed at two moments, and proves nothing at all about whether it will still be there when price arrives.
A resting order is not a promise
Limit orders can be cancelled at any moment before they fill, at no cost, and modern systems do it in milliseconds. That makes every wall conditional. A large bid can be genuine accumulation ↗ by a buyer who wants the coin. It can equally be an execution algorithm parking size, or an order placed with the expectation of pulling it before price gets close. All three look identical in a depth snapshot, because the snapshot records price and quantity and nothing about intent. The measured 4.7% level attrition over 30 seconds is the same mechanism operating at small scale, on orders too small to notice individually.
Spoofing and its honest limits
Placing orders with no intention of filling them is spoofing, and it is prohibited on regulated venues and prosecuted in traditional markets. It clearly occurs in crypto, and the honest position is that a public depth feed cannot identify it. Distinguishing a spoof from a cancelled genuine order requires knowing intent, which no snapshot contains. What a reader can do is watch behaviour: a wall that retreats as price approaches, repeatedly, is behaving differently from one that absorbs trades. That is an observation about a pattern over time, not a verdict on any single order. Stating it as more than that is not supportable from the data.
Data current as of September 2026.
The limits of the book go further than intent.
What Does the Order Book Never Show You?
Displayed depth omits hidden size, says nothing about who placed anything or why, and covers only a single venue. The visible book is a partial view of the market that presents itself as a complete one, which is the most consequential thing to know about it.
Hidden size and iceberg orders
Most venues offer order types that display only a fraction of their true size. An iceberg order shows a small visible portion and replenishes it from a hidden reserve as it fills. A level that appears to hold $50,000 can absorb many times that before it clears. Fully hidden orders go further and display nothing at all. The consequence is that displayed depth is a lower bound on real depth rather than a measurement of it. A book that looks thin can absorb far more than it advertises. A reader who sizes an order purely on visible levels is working from a systematically incomplete picture.
Intent, identity and elsewhere
The book carries no identity. A $7.7M bid could be one desk, a pooled algorithm or an exchange's own market-making operation, and nothing in the feed distinguishes them. It also carries no reason, so the same order is consistent with conviction, hedging or an obligation being met. Finally it covers one venue while the same asset trades on many. A wall on one exchange sits inside a wider market the snapshot does not see. Every one of these limits is structural and none is a data-quality failure. Better data would not remove them, because the information was never in the feed to begin with.
How Do You Read an Order Book in Practice?
Five passes over one snapshot extract almost everything a book honestly offers. Read the spread, find where size begins, mark the largest single orders, compare the imbalance between sides, then refresh to see what survived. The whole routine takes about a minute.
Five passes over one book
Start with the spread and top level, which price the smallest possible trade and nothing else. Scan downward next to find where meaningful size begins, since the first rows are usually queue noise. Sort the levels by value to surface the largest single orders and note how far each sits from mid. Compare the two sides within a stated band for imbalance. Then refresh and repeat the same five steps, because the difference between two readings is often more informative than either reading alone.
When the book is the wrong tool
The book answers questions about resting intent at this instant on this venue. It is the wrong tool for questions about what has already traded, which candles and trade tapes record. It is also wrong for questions about aggressive flow, which resting orders cannot show. It is also the wrong tool for anything longer than the next few minutes, since the measured attrition rate makes older readings unreliable. Nothing in a depth snapshot supports a view held for days. Knowing when to close the book matters as much as knowing how to read it. Most misuse comes from stretching a snapshot past what it can support.
Data current as of September 2026.
Summary
An order book lists every resting bid and offer with its price and size, which is exactly the information a candle compresses away. Reading it usefully means ignoring the first rows. In the measured snapshot the best bid was worth $2,299 and the best ask $436,585, at prices one cent apart. Seven of the first eight bid levels came to less than $2,500 combined. Small top-of-book sizes reflect price-time priority and queue positioning, not market capacity. Real size accumulates further back, and it is read as a running total downward rather than level by level.
Within that accumulated shape, single orders sometimes dominate. One bid of 100.0711 BTC at $77,100.00, worth $7,715,482, held 21.2% of bid depth within 2% of mid. Four of the five largest bids sat on round hundred-dollar prices. Summing both sides gave $36,373,788 of bids against $21,212,002 of asks, an imbalance of 1.71 to one, which describes displayed intent and predicts direction weakly. Taken again 30 seconds later, 94.4% of 5,000 bid levels were unchanged, 4.7% had vanished and total depth had moved 1.2%. A calm book is more persistent than its reputation, and none of that persistence is a promise.
Conclusion
Most order book analysis fails at the first row, by treating a $2,299 best bid as though it described a $77,000 market. Reading downward instead, sorting levels by value and comparing the two sides within a stated band extracts nearly everything a snapshot honestly holds. It takes about a minute. What it cannot hold is intent, identity, hidden size or any other venue. A wall is a fact about right now, cancellable in milliseconds and worth exactly one refresh. Read the book for structure, refresh before acting, and stop asking it questions about tomorrow.
Why You Might Be Interested?
If you size trades from the top of the book, you are reading a number worth $2,299 and assuming it describes millions. If you treat a large wall as support, note that the one measured here could be cancelled at no cost in milliseconds. If you read imbalance as direction, the snapshot shows what participants chose to display, not what they intend to do.
Quick Stats
- $2,299 vs $436,585 — best bid against best ask at prices one cent apart, a ratio of 190
- $7,715,482 — the largest single resting bid, 100.0711 BTC at $77,100.00
- 21.2% — share of all bid depth within 2% of mid held by that one order
- 4 of 5 — largest resting bids sitting on round hundred-dollar prices
- 1.71 to 1 — bid depth of $36,373,788 against ask depth of $21,212,002 within 2% of mid
- 94.4% — of 5,000 bid levels unchanged after 30 seconds, with 4.7% withdrawn
Data current as of September 2026.
FAQ
?What is a crypto order book, in plain terms?
It is a live list of every resting buy and sell order on one exchange, each showing a price and a quantity. Bids sit below the current price, asks above it. The list is the layer underneath the chart: a candle records where price went, while the book records where price could go and how much waits at each step. Exchanges publish it free through a public depth endpoint.
?Why is the top of the order book so small?
Because being first at a price is worth holding cheaply. Price-time priority rewards whoever queued earliest at a level. Participants post minimal size to claim that position, then adjust as conditions change. In the measured snapshot the best bid held $2,299 and seven of the first eight bid levels came to under $2,500 combined, in a market where a single order further back held $7.7M.
?What is a buy wall or sell wall?
A wall is a single resting order large enough to dominate the levels around it. The clearest one in this snapshot was 100.0711 BTC at $77,100.00, worth $7,715,482, which was 21.2% of all bid depth within 2% of mid. Finding walls needs no special tooling: sort the levels by value and read the top of the list.
?Can you spot whale activity from the order book?
You can spot unusually large resting orders, which is a narrower claim. The book shows price and quantity and nothing else. It cannot say whether a $7.7M bid is one desk, a pooled algorithm or an exchange's own market making. It also cannot show hidden or iceberg size, and it covers one venue out of many. Large displayed orders are visible; whales as such are an inference.
?Do buy walls mean the price will go up?
Not reliably. A wall is displayed intent that can be cancelled at any moment at no cost. The same is true of book imbalance: the snapshot leaned 1.71 to one toward bids, which describes what was on show rather than what will happen. Aggressive flow that actually moves price arrives as market orders and never appears in a resting-order snapshot at all.
?How fast does an order book change?
Faster than it looks in calm markets, and far faster in stressed ones. Across 30 seconds with no news, 94.4% of 5,000 bid levels were unchanged, 0.9% had changed size and 4.7% had gone. Total depth moved 1.2%. That is the baseline rate for a quiet book. During a fast move the attrition rate rises sharply as market makers cancel and requote.
?What is spoofing and can I detect it?
Spoofing means placing orders with no intention of filling them, to influence other participants. It is prohibited on regulated venues and prosecuted in traditional markets. A public depth feed cannot identify it: separating a spoof from a genuine cancelled order requires knowing intent, and no snapshot contains that. Repeated withdrawal as price approaches is a behavioural pattern worth noting, not a verdict.
?Why do large orders sit at round numbers?
Because people choose round numbers when deciding where to buy or sell, and the algorithms working human orders inherit the same anchors. Stops, take-profits and manual limits bunch at the same figures without anyone coordinating them. Four of the five largest bids sat on round hundred-dollar prices here, and the largest ask sat at exactly $78,000.00.
References / Sources
Platform & Company Data
Live order book depth pulled directly from an exchange public endpoint as a timestamped snapshot, then re-pulled 30 seconds later for the churn comparison.
- Binance: Spot API Order Book Depth Endpoint (binance.com, Sep 2026)
- Binance: 24hr Ticker Statistics Endpoint (binance.com, Sep 2026)
- CoinPaprika: Coin Markets Endpoint (coinpaprika.com, Sep 2026)
Regulatory & Legal
- U.S. Commodity Futures Trading Commission: Prohibition on spoofing, Commodity Exchange Act §4c(a)(5)(C) (cftc.gov)
Původní průvodce CoinPaprika.