FX drift is an exposure, not an edge
Net R was defined as "after all costs, converted to account currency", which conflates two different things. A GBP account holding a USD stock for ten days incurs an FX cost — the spread on converting in and out — and takes an FX position: it is long USD for the duration, and nobody asked for that.
Including drift in measured edge tests a joint bet, the strategy plus an unhedged currency position. That currency position has no declared mechanism, no falsifier and no registered hypothesis — it is an undeclared information axis taken with real money, inside a system built to refuse exactly that.
Net R is measured in the instrument's currency, with FX conversion spread included as a cost. FX drift is excluded from measured edge and reported as a separate declared exposure with its own limit in configuration.
Considered options
- Net R in account currency, drift included. One number, no reconciliation, and it measures what is actually banked. Rejected: the same Strategy would receive a different verdict depending on the account's currency, and the currency component would never have been registered.
- Hedge the exposure. Cleanest conceptually — account-currency and instrument-currency net R become equal. Adds hedging cost to every trade, requires venue support for the hedge instrument, and is likely unavailable on a retail equity account.
Consequences
Three quantities stay distinct that were previously one: what the Strategy earned, what the currency contributed, and what was banked. Reported P&L will therefore differ from measured edge, and that difference must be explained wherever both appear rather than quietly reconciled.
Verdicts remain comparable across a universe spanning several currencies, which a GBP-denominated net R would not be.