Risk Management for Automated Trading: Daily Loss Limits and Guards

Automation removes the human circuit breaker

A discretionary trader having a nightmare session eventually stops - tilted, exhausted, or just horrified. An automated strategy does not tire and does not doubt. If market conditions turn hostile to its logic, it will take signal after losing signal with perfect discipline until something external stops it. That external something is the guard layer, and running automation without one is the single most common way accounts die.

The design principle: guards live on the terminal, independent of any strategy. Whatever the signals say - your strategy, a leader you copy, a misconfigured alert firing in a loop - the guards have the final word.

The core guards

• •Daily loss limit (% or $): the master switch - when the day loss hits the line, trading halts until tomorrow. Set it to survive the worst plausible day, not the average one
• •Max open trades: caps simultaneous exposure and neutralizes runaway alert loops
• •Spread filter: skips execution when the spread exceeds your pip limit - rollover and news spikes are where good entries get terrible fills
• •Trading hours window: your strategy tested on London liquidity should not trade the Asian drift by accident
• •News filter: pauses around high-impact releases, configurable minutes before and after

Guards versus sizing: both, not either

Risk-based position sizing controls how much any single trade can lose; guards control how much a day or a malfunction can lose. They fail differently: sizing assumes the stop gets respected (news gaps disagree), guards assume the day boundary matters (a slow bleed across a week slips past a daily limit). Together they cover each other - 1% risk per trade under a 4% daily limit means even a flash-crash session with slippage stays survivable.

Every skipped signal appears in the feed with its reason - the audit trail that turns a mysterious quiet day into a visible guard doing its job.

A sane starting configuration

For a new automated account: 1% risk-based sizing, 4% daily loss limit, five max open trades, 3-pip spread filter, news filter on with 15 minutes each side of high-impact events, and trading hours matched to whatever sessions the strategy was tested on. Loosen individual guards deliberately as live data earns it - never all at once, and never during a winning streak, which is precisely when overconfidence writes checks the next drawdown cashes.