Risk Management Rules Every Algorithmic Trader Should Follow

An algorithm removes emotional execution errors, but it introduces a new failure mode: it will execute a bad idea perfectly, at machine speed, until the account is empty. Risk management for automated trading is therefore not a feature — it is the architecture. These are the five rules our own systems are built around.

Rule 1 — Risk a fixed fraction, never a fixed lot

Position size must be computed from current equity and stop distance, so that every trade risks the same percentage — typically 0.5–1%. Fixed-lot traders unknowingly increase their risk as the account shrinks, accelerating drawdowns exactly when they can least afford them.

Rule 2 — Install a kill switch

Every bot needs an equity stop: a hard daily or weekly drawdown level at which the system closes positions and disables itself until a human reviews it. Markets occasionally go non-stationary — flash crashes, surprise interventions, broken liquidity. The kill switch is the difference between a bad week and a funeral.

Rule 3 — Cap correlated exposure

Three bots trading EUR/USD, GBP/USD and gold are often one bet on the dollar wearing three costumes. Correlation-capped portfolios limit total exposure per theme, so a single macro shock cannot hit every strategy simultaneously.

Rule 4 — Trust live data over backtests

Backtests assume perfect fills. Live markets slip, spread and gap. Track the delta between backtested and realised performance; when slippage consistently exceeds assumptions, the strategy's true expectancy may already be negative.

Rule 5 — Retire strategies on evidence, not on hope

Define in advance what failure looks like — for example, a drawdown exceeding 1.5 times the historical maximum, or a rolling win rate outside the tested confidence band. When the evidence says the edge is gone, the correct trade is the off switch.