Direct answer
Spread by Session helps explain that spreads can vary by trading time (often called sessions), but it has practical limitations. It is less useful when spreads shift for reasons that are not tied only to time—such as changing liquidity, volatility, or the way a provider calculates costs. It also cannot guarantee future results: a historical or general time pattern may not repeat.
Mechanism and definition
“Spread by Session” typically means grouping trading hours into windows (sessions) and observing how the bid–ask spread behaves within each window. The spread is the difference between the buy and sell price; it acts as a transaction cost because you effectively pay half the spread when entering and the other half when exiting.
To use this idea, you need at least these inputs:
- A clearly defined session schedule (what time range counts as each session).
- A source of spread observations (for example, historical quotes).
- A consistent measurement method (for example, whether you look at average spread, median spread, or extremes).
Evidence or example (assumptions included)
Assume you split the day into two windows and you calculate an “average spread” for each. Even if you find that Session A historically has a lower average spread than Session B, that comparison is still conditional. The relationship depends on the particular days observed, the volatility regime, and the availability of quotes.
Common ways the pattern can mislead:
- Regime changes: If volatility spikes during the “usually calm” session, spreads can widen.
- Liquidity shifts: If fewer participants are active, spreads can widen even in the same time window.
- Different measurements: A provider might show a different spread metric than what you previously observed (for example, average vs. peak during fast moves).
In other words, the “by session” idea describes a past or typical relationship between time and spread, not a stable law.
Limitations and risks
1) Time-of-day is not the only driver
Spreads react to market conditions in real time. Even with the same session label, conditions can differ. Therefore, any session-based expectation can become inaccurate.
2) Execution and total cost may differ from spread alone
Spread is only one component of trading costs. The actual cost you experience can be affected by execution quality, how quickly prices update, and other charges. A low “typical spread” session does not ensure low all-in costs.
3) Historical relationships do not establish future results
Averages and patterns computed from past data do not guarantee future behavior. Changes in liquidity, volatility, or pricing practices can break the historical pattern.
4) Definition and data choices change the conclusion
If sessions are defined differently (time ranges) or if you use different spread definitions (average, median, or worst-case), you can get different “session” outcomes. Two people can both be “measuring spread by session” and still end up with incompatible results.
Verification or next question
To verify whether “Spread By Session” is useful in a specific case, you can independently check whether the session labels and spread measurements are consistent with your data source and whether the pattern holds across multiple time periods. A helpful next question is: Which costs are included in your comparison—spread only, or all-in transaction costs including other fees—and how sensitive are the results to the session definition?