By William Harris · Last reviewed
Trusting Your EA After Drawdown — The Hardest Part of Automated Trading
Why drawdowns feel different than they look
A 20% drawdown reviewed on a 5-year backtest chart appears as a small dip among many. The same 20% drawdown experienced over 6 weeks of declining equity feels existentially threatening. The asymmetry is real: humans evaluate ongoing loss with much greater emotional weight than the same loss reviewed in hindsight.
Three factors amplify the emotional intensity: (1) Real money. The drawdown is your actual capital, not a backtest abstraction. (2) Powerlessness. The EA is making decisions you can't override without breaking discipline. (3) Uncertainty. You don't know if the drawdown will deepen or recover. Each factor makes the experience meaningfully harder than reviewing similar drawdowns in past data.
The implication: drawdowns are when emotional decision-making most threatens strategy adherence. The traders who succeed have pre-committed to specific actions at specific drawdown levels — the commitments are made when calm and applied when distressed. Without pre-commitment, the in-drawdown trader makes emotional decisions that almost always damage long-term outcomes.
Distinguishing normal variance from genuine degradation
Normal variance: the EA's current drawdown is within or close to its backtest maximum drawdown. The individual losing trades fit the strategy's expected pattern (system trades hitting normal stops, not unusual outcomes). The drawdown has been building over days to weeks rather than appearing suddenly. Other EAs running similar strategies are also in drawdown (regime-level explanation).
Genuine degradation: the current drawdown exceeds 1.5× backtest maximum drawdown. Losing trades show patterns not present in backtest (e.g. EA placing trades outside expected session times, sizing wrong, missing scheduled news filter activations). The drawdown appeared suddenly without proportional market move. Similar EAs are not in equivalent drawdown.
Practical procedure during any drawdown: (1) compare current drawdown depth to backtest max DD, (2) audit the most recent 10-20 losing trades for pattern consistency, (3) check broker condition (spread profile, execution quality), (4) compare to other EAs or community reports of similar strategies. If 3+ of 4 checks indicate normal variance, the drawdown is within expected range. If 3+ of 4 indicate degradation, investigate and consider intervention.
Most drawdowns fall into the 'normal variance' bucket. Most abandonment decisions made during drawdowns were premature — the EA would have recovered if held. This pattern is so consistent across retail EA operations that it's the strongest predictor of long-term success: traders who hold through drawdowns succeed at higher rates than traders who exit during them, controlled for EA quality.
Pre-commitment framework
Before deploying any EA, document specific drawdown thresholds and the specific responses at each:
Threshold 1 — Within backtest max DD (e.g. up to 18% for an EA with 18% backtest max DD). Response: continue running, no changes. Drawdowns of this magnitude were observed in backtest and the EA recovered from them. The current drawdown is within proven variance.
Threshold 2 — 1.0-1.3× backtest max DD (e.g. 18-23% for the same EA). Response: continue running but increase monitoring frequency. Daily review becomes twice-daily. Document the drawdown progression. No intervention yet but heightened attention.
Threshold 3 — 1.3-1.5× backtest max DD (e.g. 23-27%). Response: investigate. Audit recent trades for pattern divergence. Check broker conditions. Reach out to vendor or community for diagnosis. Continue running while investigating; don't override yet.
Threshold 4 — Above 1.5× backtest max DD (e.g. above 27%). Response: pause EA. Close all positions. Conduct full investigation before resuming. Most legitimate EAs don't reach this threshold; if yours does, something has likely changed (regime, broker, EA itself) that requires resolution before continuing.
Threshold 5 — Above 2× backtest max DD (e.g. above 36%). Response: replace EA. The strategy's edge has either degraded materially or never existed as backtested. Don't resume the same EA; transition to a different strategy.
The conversation with yourself during drawdown
When drawdown is happening, the mind generates strong narratives for action: 'This time is different', 'The EA is clearly broken', 'I just need to override this one trade'. These narratives feel like reasoning but are usually rationalizations of emotional urgency.
Counter-narrative practice: when you notice the urge to override or abandon, ask yourself: 'What does my pre-commitment document say to do at this drawdown level?'. The pre-commitment is the calm-state decision; honor it during the distressed state.
If the pre-commitment doesn't fit the current situation (e.g. drawdown hit faster than expected), pause for 24 hours before any action. The 24-hour cool-down separates the urge from the action. Most urges don't survive the cool-down.
After the drawdown — whether it recovers or doesn't — journal the experience. What did you feel? What were you tempted to do? Did the pre-commitment hold up? The journal builds self-knowledge across multiple drawdown cycles; over time, your responses become more aligned with the strategy's success rather than against it.
Frequently asked questions
How long do typical EA drawdowns last?
Drawdown duration is often more psychologically difficult than depth. A 15% drawdown that recovers in 6 weeks is bearable; the same 15% drawdown that takes 18 months to recover is torture. Look at the EA's Maximum Drawdown Duration in the backtest report — if your current drawdown is taking longer than the backtest's worst duration, that's a meaningful signal. Most healthy EAs have 12-26 week worst drawdown durations; durations beyond 40 weeks suggest the strategy may not recover.
What if I genuinely think 'this time is different'?
The diagnostic for 'this time is genuinely different' vs 'I feel like this time is different': run the 4-check audit (drawdown depth vs backtest, trade pattern audit, broker condition, peer EA comparison). If 3+ checks indicate degradation, your intuition has data support and intervention may be warranted. If 3+ checks indicate normal variance, your intuition is being fooled by emotional intensity. Trust the data-based audit, not the feeling.
The vendor says my drawdown is normal — should I trust that?
Vendors have incentive to reassure customers during drawdowns — abandoned EAs hurt the vendor's reputation and create refund requests. Some vendors are honest about strategy degradation; some default to 'normal variance' regardless of what's actually happening. Verify by cross-checking the vendor's own live track. If the vendor's Myfxbook shows similar drawdown to yours, the variance is genuinely strategy-level and likely recoverable. If the vendor's account is profitable while customers are in drawdown, the difference is your-specific-setup level and needs investigation independent of vendor's general reassurance.
I've blown up an EA before. How do I trust the next one?
Post-blow-up trust rebuilding is structured time. Most former blow-up traders need 6-12 months of consistent operation on a new EA before the emotional skepticism subsides. The accelerator: very small initial capital where individual losses don't trigger blow-up trauma responses. As the account grows through successful months, capital additions can match the demonstrated stability. Don't try to short-circuit by depositing large capital based on intellectual confidence; the emotional pattern needs accumulated evidence to update.
When should I replace an EA vs just rest it?
Resting an EA provides operational distance to investigate without further losses, lets you observe whether market regime evolves favorably, and tests whether the underlying strategy edge is regime-specific or persistent. Replacement should be reserved for EAs that fail to recover after multiple rest cycles — typically 2 rest periods of 4-8 weeks each with similar drawdown patterns each time. EAs that fail this test have likely lost their edge and replacement is warranted. EAs that recover after rest have demonstrated persistent edge; continue running them.
Calculate the recovery you're facing
Our drawdown calculator shows the recovery percentage required from any current drawdown. Use it to set realistic recovery expectations.
Open Drawdown Calculator →