Outcome is a poor judge of process
A month can be profitable for bad reasons and negative for good ones. Judging a rules-based system purely on its monthly figure trains you to reward luck and punish discipline, which over a year reliably produces worse decisions than having no review process at all.
The better question is whether the system behaved inside its own specification. Did it take the trades it was designed to take, at the size it was configured to take them, with the losses it was expected to accept? If yes, a red month is information about markets. If no, a green month is information about a bug.
The monthly review checklist
- —Was the maximum equity drawdown inside the configured tolerance?
- —Was the number of trades within the normal range for the conditions?
- —Did any single trade contribute more than 25% of the month's result, up or down?
- —Was average slippage stable compared with the previous month?
- —Did anything manual happen — an override, a pause, a resize — and was it recorded?
Record your interventions. The intervention log explains more of your results over a year than the strategy does.
Expected distribution of months
Any honest system produces a distribution, not a streak. Over a twelve-month window, expect the majority of months positive, a meaningful minority negative, and at least one flat month where conditions simply did not present. A vendor showing twelve green months in a row is either extremely fortunate or not showing you everything.
Judging the process when the outcome disagrees
A month can be profitable and badly executed, or negative and executed perfectly. Because outcomes over thirty days are dominated by variance, the only stable thing to evaluate is whether the system did what it was designed to do: respect its exposure limits, place stops where the rules require, refuse trades in prohibited conditions, and recover cleanly from any interruption.
Separate the two verdicts explicitly
- —Process verdict: did every trade comply with the rules as written?
- —Outcome verdict: where did the month land inside the expected distribution?
- —A good process with a poor outcome requires no action at all.
- —A poor process with a good outcome requires immediate action.
The metrics worth tracking month to month
Rather than tracking the profit figure, track the inputs that produce it. Average realised spread, slippage against expectation, number of rule-blocked trades, uptime of the hosting environment and time under water form a dashboard that changes slowly and tells you when something structural has shifted — usually well before the equity curve does.
A five-line monthly log
Everything discussed here is applied on a public, third-party verified account — updated continuously, losing weeks included.