Mechanism: define session volatility and what you measure
Session volatility is a descriptive measure of how much an exchange rate tends to vary within a specific trading session window (for example, from a session open to a session close). It is usually expressed as a price range (high minus low) or as a percentage move over the window.
Key idea: session volatility is about the observed dispersion inside a time box, not about predicting a future move.
To work with the concept, you must decide:
- Session window: the exact start and end timestamps you include.
- Data basis: whether “high” and “low” come from bid/ask, mid-price, or another reference.
- Metric: range in price units, or percent movement.
Worked example (fully numeric) with explicit assumptions
Assume you want to estimate session volatility for a hypothetical market during one session.
Assumptions (state everything you use):
- You define the session window as a single period from 10:00 to 12:00.
- You use a reference price series where each candle provides a high and low for the session.
- You compute volatility using the simplest “range” approach:
- Absolute range = session high − session low.
- Percent range = (absolute range ÷ session low) × 100%.
- You assume no transaction costs are included in the price series (this matters later in limitations).
Example input (from your assumed dataset):
- Session low = 1.1000
- Session high = 1.1120
Calculations:
- Absolute range = 1.1120 − 1.1000 = 0.0120
- Percent range = (0.0120 ÷ 1.1000) × 100% = 1.0909% (about 1.09%)
Interpretation (non-promotional, descriptive):
- In this example session window, the exchange rate moved through a spread of about 0.0120 units, which is about 1.09% of the session low.
- If you repeat this process over multiple similar sessions, you can compare typical ranges (for example, average range across sessions).
How this example “works” in practice
The example “works” because volatility is computed directly from agreed measurements (session high/low) and clearly defined assumptions (window, price reference, and metric). There is no need to claim future direction; you are only describing variability during that window.
You can also refine the metric while staying transparent:
- Use the median of session ranges instead of the mean.
- Compute range from hourly highs/lows inside the session window.
- Compare sessions by grouping them consistently (same session window, same data basis).
Limitations and failure modes (what can break)
At least one material limitation is that session volatility is sensitive to choices and conditions that are outside the concept itself:
-
Window selection error: If the session window is too short, you may capture noise; too long, you may mix different market regimes. The resulting “volatility” may not reflect what you intended to study.
-
Price reference inconsistency: High/low values depend on whether you use mid-price, bid, or ask. Mixing references between periods can change the computed range.
-
Costs and execution: Even if the price series shows large movement, the real tradable outcome can be reduced or altered by bid-ask spread, slippage, and order execution constraints. Our numeric example assumed none of these.
-
Non-stationarity: Historical relationships do not establish future results. A session that is volatile in one period can be quieter later due to changing liquidity, macro events, or risk sentiment.
-
Overgeneralization: Using a single worked example as if it represents an entire market can mislead. Volatility should be summarized across many sessions to be meaningful.
Verification and next question to ask
To independently verify session volatility as a concept, you can replicate the same computation on your own chosen dataset:
- Pick a fixed session window.
- Extract session highs and lows from the same price reference.
- Recompute absolute and percent ranges.
- Compare results across multiple sessions rather than relying on one window.
If you want to go one step further, a useful next question is: How does your computed session range change when you alter the session window and price reference? This directly tests whether your “session volatility” measure is robust or overly dependent on assumptions.