Spread By Pair: definition and what you measure
“Spread by pair” refers to the difference between the buy (ask) and sell (bid) prices quoted for a specific currency pair at a specific moment or over a defined time window. In practical terms, you assess spread using bid–ask quote data and a clearly stated rule for how you convert that difference into a reportable metric (for example, points, pips, or a percent of price).
Before discussing implications, keep the measurement mechanics separate from what causes the quotes to change.
Data inputs you need (and how to structure them)
To assess spread by pair in a verifiable way, collect these inputs:
- Bid and ask quotes for the pair
- Minimum: bid price and ask price for the same timestamp (or for the same aggregation window).
- Include the quote currency/price format if available (e.g., number of decimals).
- A consistent spread metric definition Common choices include:
- Absolute spread: ask minus bid (in quote-price units).
- Pip/point spread: spread converted using the instrument’s pip/point convention.
- Percentage spread: absolute spread divided by an agreed reference price.
If you use any conversion (pips/points or percentages), state the exact assumptions (pip size, reference price used) because different conventions can change results.
- Quote timing information (timeliness)
- Timestamp of each quote, or the start/end times for any averaging window.
- Time zone and whether quotes are server time or local time.
- The sampling interval (for example, every second, every tick, or aggregated per minute).
- Data provenance (where the quotes come from)
- The source identity: data vendor, exchange/venue feed, broker quote feed, or platform export.
- Any transformation applied by the source (normalization, rounding, bid/ask definitions).
- Quality and completeness markers
- Handling of missing data (gaps, zero/invalid quotes, stale quotes).
- Flags for abnormal ticks (outliers) and rules for excluding them.
- Whether the feed includes spread-relevant details beyond bid and ask.
Evidence or example: a simple, checkable calculation
Assume you have bid and ask quotes for EUR/USD sampled once per minute for one hour. For each minute you can compute:
- Absolute spread = ask − bid.
- If you want pip spread, apply your pip-size assumption consistently.
To assess “by pair,” repeat the same steps for other pairs using their own pip conventions if they differ. Then compare summaries such as the median spread and an interquartile range across the hour.
A key point: this method only describes spread under the collected conditions and measurement rules. It does not imply future behavior, and it does not include any execution or trading-cost components that may exist outside the quoted bid–ask difference.
Limitations and risks (what can go wrong)
Several failure modes can make spread assessments misleading:
- Timestamp mismatch: bid and ask must correspond to the same moment. Using asynchronous data can inflate or deflate the computed spread.
- Inconsistent pip/point conventions: converting to “pips” with the wrong pip size can distort comparisons across pairs.
- Stale or low-quality quotes: feeds that repeat old quotes can make spread look artificially stable.
- Variable market conditions: spreads change with liquidity and volatility; historical averages may not represent current conditions.
- Provider-specific quote behavior: different quote sources may use different definitions or rounding, so results may not be comparable.
Verification and next question
To verify your assessment, independently check that:
- Your bid and ask data contain valid, correctly paired timestamps.
- Your conversion assumptions (pip size, reference price) are explicitly stated and applied uniformly.
- Your summary statistics handle missing values and outliers using a documented rule.
If you want to go further, the next question is often how “spread by pair” relates to total dealing costs, which may include other fees and execution effects beyond the bid–ask difference.