How to measure liquidity: bid-ask spread and order-book depth

Mid price and bid-ask spread

The mid price is halfway between the best bid and the best ask: (best bid + best ask) ÷ 2. The bid-ask spread is the gap between them, expressed as a share of the mid price:

spread = (best ask − best bid) ÷ mid price

Best bid 0.0995 USDT, best ask 0.1005 USDT. The mid price is 0.1000, and the spread is 0.0010 ÷ 0.1000 = 1.0%, or 100 basis points.

The spread is roughly what a small buyer loses by buying and immediately selling. It says nothing about size: a 0.5% spread with a few dollars at the best prices can still be an expensive market for a real order.

Depth within ±1% and ±2%

Depth within ±x% is the total value of resting orders priced within x% of the mid price, in the quote currency (USDT in the examples here). Count each side separately:

• Bid depth: the sum of price × quantity for bids priced at or above mid × (1 − x%).
• Ask depth: the sum of price × quantity for asks priced at or below mid × (1 + x%).

Mid price 0.1000. The ±1% band runs from 0.0990 to 0.1010, and the ±2% band from 0.0980 to 0.1020.

Side	Price	Quantity	Value (USDT)	Within ±1%	Within ±2%
Bid	0.0995	20,000	1,990	Yes	Yes
Bid	0.0985	30,000	2,955	No	Yes
Bid	0.0970	50,000	4,850	No	No
Ask	0.1005	18,000	1,809	Yes	Yes
Ask	0.1015	25,000	2,537.5	No	Yes
Ask	0.1040	40,000	4,160	No	No

Bid depth: 1,990 USDT within ±1% and 4,945 USDT within ±2%. Ask depth: 1,809 USDT within ±1% and 4,346.5 USDT within ±2%. The orders at 0.0970 and 0.1040 add 9,010 USDT to the book but nothing to either band.

Depth tells you roughly how large a market order can be before it moves the price by about x%, assuming no new orders arrive while it executes. That is why it is a better guide than the spread to what a real buyer or seller will pay.

Common mistakes

• Counting orders far from the price. A book with many orders 10% away looks busy but is thin where trading happens. Placing a large number of orders does not by itself create useful liquidity.
• Adding the two sides together. A book can be deep on bids and nearly empty on asks. Report each side.
• Measuring in tokens. Token quantities are not comparable across prices or projects. Use the quote currency.
• Using a stale mid. Recompute the bands from the mid price at each sample.
• Trusting a truncated snapshot. Many exchange APIs return only a limited number of price levels. Check that each snapshot actually reaches the edge of your widest band.

Two-sided presence

A market can have a good average spread and still spend part of the day with one side empty. Two-sided presence measures that: the share of samples in which both sides have orders within an agreed band, and the spread is within an agreed maximum. Write the band and the maximum into the definition. Otherwise the number cannot be compared from one report to the next.

Sampling and summarising

These metrics change from second to second, so any report is a summary of samples. The summary is only meaningful if the method is stated:

• Interval. How often the book is sampled, for example once a minute.
• Aggregation. A time-weighted average hides the worst moments. Add the median and a high percentile of the spread, and the lowest depth, to show them.
• Coverage. Say how many samples were expected and how many were usable. List missing data as missing. Never fill it in or count it as zero.
• Exclusions. Exchange maintenance and API outages should be listed as excluded periods, with their times.
• Source. Use the same data source and method for any before-and-after comparison.

There is no single report template that fits every market: exchanges expose data in different formats, and some projects also need on-chain data. At Lyqui, each order book we operate has a real-time Telegram report, and deeper analysis is delivered as a custom report. See how our reporting works.

Why trading volume alone says little

Volume counts trades that already happened. It does not tell you whether the next buyer can trade near the current price. A market with high volume and a 5% spread is still expensive to trade in. Volume is also the easiest number to fake: in October 2024 the U.S. SEC charged three self-described market makers, alleging they generated artificial trading volume for their clients' tokens. Read spread and depth first, and treat volume as context.

What about liquidity scores?

Data aggregators and exchanges publish liquidity and market-quality scores. These often draw on inputs like the ones above, weighted in ways each publisher defines and can change. A score is outside the control of any project or provider, so do not treat it as a target. Work on the spread, depth and presence behind it, and measure those directly.

A checklist to agree before measuring

1. How the mid price is calculated.
2. The spread formula, and whether it is reported in percent or basis points.
3. The depth bands (for example ±1% and ±2%), reported per side, in the quote currency.
4. The two-sided presence rule: which band, and what maximum spread.
5. The sampling interval, the data source and the time zone.
6. How results are summarised over the period.
7. What is excluded, and how missing data is shown.

With these agreed, two people looking at the same market will get the same numbers. That is the starting point for any useful conversation about liquidity.