How Much of an Order Book Can Your Data Actually See?

A market can show an extremely narrow bid–ask spread while providing little visible quantity at the best price. The spread answers how far apart the nearest buyer and seller stand. It does not answer how much can trade there.

Order-book depth makes that second question measurable. It also forces a useful discipline: distinguish the book an exchange maintains from the portion a particular data response reveals.

Start with the right side of the book

Bids represent displayed buying interest; asks represent displayed selling interest. A hypothetical immediate purchase consumes asks, beginning with the lowest price. A hypothetical immediate sale consumes bids, beginning with the highest.

The best ask is therefore relevant to the first portion of a purchase. It need not describe the average price of the entire purchase. Coinbase’s order documentation explains why executions can differ from the most recent trade and why orders can fill in parts.

Last-trade price, midpoint and executable average are three different measurements. Treating them as interchangeable can make a chart look accurate while making an execution estimate wrong.

The snapshot we actually received

CryptX requested 25 levels per side for Bybit BTCUSDT spot. The returned response contained 10 bids and 10 asks. Its server timestamp was 22 September 2026, 19:31:05.753 UTC; retrieval completed at 19:31:06.047 UTC.

This historical snapshot is used for an educational calculation. It is not a quote available on this article’s publication date.

The midpoint was 86,463.05 USDT. Dividing the spread by that midpoint and multiplying by 10,000 gives about 0.0116 basis points. One basis point equals 0.01%.

That small spread is a property of the nearest two prices in this observation. The roughly 0.488 BTC on the returned ask side is a property of the limited data slice. Neither number establishes the exchange’s complete available liquidity.

Walk through quantity, then calculate the average

A mechanical book calculation takes as much quantity as needed from the lowest ask, then moves to the next ask if necessary. It multiplies each consumed quantity by its price, adds the costs and divides by total quantity.

For a hypothetical 0.1 BTC purchase, the first ask contains enough quantity. The indicated cost is 8,646.31 USDT before fees, corresponding to 86,463.10 USDT per BTC.

A 0.4 BTC purchase would extend into the sixth returned ask level. The calculated total becomes 34,585.5149407 USDT, or approximately 86,463.79 USDT per BTC. That average differs from both the first ask and the highest price reached.

These are original CryptX calculations using the saved rows. They assume the displayed quantities remain available, ignore competing orders and apply no execution fees. They describe a static mathematical exercise, not a fill prediction.

The most useful result can be “insufficient depth”

Now consider a hypothetical 0.5 BTC purchase. All returned asks together cover only 0.487517 BTC. The uncovered amount is 0.012483 BTC.

It would be incorrect to fill that remainder at the last visible ask, to assume it costs zero, or to extend the curve with invented levels. We do not have the data for a complete average.

The valid result is narrower: the returned portion costs 42,153.1073497 USDT, and a complete 0.5 BTC estimate is unavailable from this response.

This is an analytical limitation, not proof that the exchange could not execute the order. More levels may exist outside our extract. A request for 25 levels is also not evidence that 25 arrived; the actual response controls the calculation.

Visible depth is conditional

Displayed orders can be modified, cancelled or matched before another participant reaches them. Bybit’s order-book stream documentation describes snapshots and subsequent changes. A screenshot freezes that changing process at one moment.

Coverage matters too. Bybit’s REST documentation states that retail price-improvement orders are excluded from that response. A public extract is therefore a defined view of liquidity, even when it contains many levels.

A depth chart should state its venue, market type, timestamp and price or level range. “Depth within 1% of the midpoint” and “the first 10 levels” are different filters. Our limited response does not justify claiming a complete 1% measurement.

What makes a comparison useful

To compare venues, hold the hypothetical order size, side and quote currency constant. Align observation times as closely as possible. Include the applicable fee treatment and explain missing coverage.

A headline liquidity score can provide context, but it reflects a methodology rather than a universal promise. CoinMarketCap’s liquidity metric evaluates market conditions through its own approach. A specific order still requires a specific calculation.

Fees can also exceed a very narrow displayed spread. Coinbase’s fee guide distinguishes liquidity-taking and liquidity-providing executions; the applicable rate belongs alongside the estimate.

The strongest depth analysis makes its boundary visible. It shows the quantities received, explains the arithmetic and stops when those quantities run out. That restraint is what turns an attractive chart into useful evidence.

Disclaimer: For informational and research purposes only. This is not financial advice. Market data may be delayed or inaccurate; verify material information with primary sources.