What Data Is Needed to Assess Liquidity and Spreads?

Direct answer

To assess liquidity and spreads in forex, collect data that explains (1) how easy it is to trade at requested prices (liquidity) and (2) the cost of immediate execution at those prices (spreads). Because both are variable, you also need provenance (where the numbers come from), timeliness (when they were measured), and quality checks (whether the data was computed and recorded consistently).

Mechanism and definition

Liquidity describes how much tradable interest exists near current prices and how quickly prices respond to trades. Common data inputs include:

Order-book depth / standing liquidity: quantities available at or near the top of book (or within a chosen price distance). If a full order book is not available, depth proxies may be used, but they must be clearly defined.
Market impact measures: how prices move when trade size increases. Even simple measures like price change versus trade size are useful, but they rely on consistent assumptions.
Trading activity and turnover: how often and how much the market trades in relevant sessions.

Spreads represent the difference between the best available buy and sell quotes. The key data inputs are:

Bid and ask quotes (or mid-price plus spread), with timestamped observations.
Spread distribution, not only averages: minimum, median, and tail behavior matters during stressed periods.
Execution cost mapping: how quoted spreads relate to actual transaction prices, since real execution can differ from displayed quotes.

Evidence or example (with explicit assumptions)

Assume you want a self-contained snapshot for a given currency pair over a specific session window (for example, one hour). A practical checklist is:

Pick a definition for “liquidity” you will test. For example, you may define it as “depth within a fixed distance from mid-price.” This requires a chosen distance (in pips or basis points) and an assumption about what “near” means.
Collect spread observations with timestamps and compute a spread distribution. If you only use one recorded quote per day, you may miss intraday widening; if you use every available tick, you must ensure the feed is comparable across time.
Validate comparability. Data from different vendors, venues, or estimation methods can produce different spread numbers. Provenance matters: the same market can show different spreads depending on whether data reflects indicative quotes, executable quotes, or aggregated liquidity.
Check cost consistency. Treat spreads as one component of execution cost. Other costs (for example, commissions, financing, or slippage from delayed execution) must be accounted for separately, otherwise “spread-only” conclusions can be misleading.

Limitations and risks (material failure modes)

At least one major limitation is that liquidity and spreads can change rapidly, often due to time-of-day effects, volatility, and order flow. If your data is not timestamped or is aggregated over long intervals, you can mask short-lived but meaningful widening.

Other common failure modes:

Mixing definitions: one source may measure “quoted” spreads while another measures “effective” execution spreads.
Non-comparable samples: different time zones, trading sessions, or data filters can make two datasets look inconsistent even when the underlying market is similar.
Quality problems: missing quotes, stale timestamps, or outlier handling can distort spread distributions and depth proxies.
Historical relationship risk: even if liquidity and spreads correlate historically, that does not guarantee future behavior, especially under regime changes.

Verification or next question

To independently verify liquidity and spread assessments, you should be able to answer these questions from the dataset documentation:

Provenance: What system generated the quotes or depth data (vendor feed, venue, or model), and does it describe “indicative” versus “executable” pricing?
Timeliness: What is the timestamp resolution, and does the dataset correct for delays or stale observations?
Method: How are spreads computed (bid/ask directly versus proxy), and how is liquidity measured (order-book depth versus proxy variables)?
Quality checks: What are the missing-data rules, outlier rules, and coverage across the chosen sessions?

If you tell me what data you already have (for example, only mid-price, or bid/ask quotes, or trade-level records), I can help you map that data to the specific inputs needed to assess liquidity and spreads without turning assumptions into unsupported conclusions.