Direct answer: what to measure (and what not to assume)
“Volatility in Account Base Currency” is not about predicting future price moves. It is about quantifying how much the account’s value, expressed in your chosen base currency, tends to fluctuate over time.
A practical way to measure it is to compute the variability of the account’s base-currency returns over a defined horizon. You first decide how the account value will be translated into the base currency (valuation rule), then measure how variable the resulting returns are.
Mechanics: define account value and base-currency returns
1) Define the base-currency value series
Let your account base currency be B. At each time t, define an account value in base currency, V_B(t). In a simplified educational model, you can treat V_B(t) as:
- the local mark-to-market value in the traded instrument currency, converted into B using an exchange rate at time t, and
- optionally including cash movements from deposits/withdrawals and trading costs (only if you can define them consistently).
If you want to isolate pure market-driven fluctuation, you typically exclude deposits and withdrawals from the variability calculation, because those are external cash flows rather than market risk.
2) Compute returns at consistent intervals
Pick a sampling interval (for example, every hour, every day, or every week) and compute returns:
- Simple return: r_t = (V_B(t) / V_B(t-1)) − 1
- Log return: r_t = ln(V_B(t) / V_B(t-1))
Log returns are often used because they convert multiplicative changes into additive changes, which can make aggregation over time easier. The key requirement is that you apply the same definition every time.
3) Translate returns into a volatility metric
Once you have a time series of returns, compute a dispersion measure, commonly:
- Standard deviation of returns over the chosen window.
To compare volatility across different horizons, you need a scaling assumption. A common educational approach is volatility scaling by the square root of time, but that assumption can fail when returns are not stable over time.
Evidence or example: two valuation choices that change the number
Scenario: measuring “account value” with spot vs. conversion timing
Assume you have an account valued in base currency B, but your positions are exposed to other currencies. To convert those positions, you need an exchange rate. Two different valuation choices can produce different volatility results:
- Spot rate at the measurement timestamp: convert using the exchange rate observable at each t.
- Conversion tied to execution or funding events: convert using rates related to when positions are opened/closed or when currency conversion happens.
Both are defensible, but they measure different things. The first captures volatility from continuous revaluation. The second can fold in timing effects and transaction-related effects.
Material limitation: costs can look like “volatility”
If you include spreads, commissions, or other costs inside V_B(t), the volatility metric can increase even when market price movement is unchanged, because execution and friction affect account value. Therefore, a useful measurement plan should state whether V_B(t) includes costs and whether you are measuring:
- market-driven variability, or
- total variability including operational/friction effects.
Limitations and risks: assumptions, failure modes, and uncertainty
Limitation 1: sampling frequency and window length
Different sampling intervals can change volatility materially. High-frequency sampling may capture short-lived noise; low-frequency sampling can smooth away real variability.
Limitation 2: non-stationary behavior
Returns often change character over time (for example, during stressed vs. calm market conditions). A single volatility number over a long window can hide this.
Limitation 3: scaling and distribution problems
Standard deviation assumes that summarizing variability with a single dispersion number is meaningful. If the return distribution has heavy tails or skew, the “volatility” number may under-represent rare but large moves.
Limitation 4: historical relationships don’t predict future outcomes
Even if you compute a stable-looking historical volatility, that does not establish how volatile the account base currency will be later.
Failure mode: inconsistent treatment of cash flows
If deposits/withdrawals are accidentally included, the measured variability can reflect cash flow timing rather than base-currency risk.
Verification: how to check your measurement independently
- Document your valuation rule for converting position value into base currency (what exchange rate and what timestamp). 2. Document your return definition (simple vs.