How to Reduce Drawdowns Without Destroying Expectancy
Introduction: The Real Problem With Drawdowns
Every trading system experiences drawdowns.
The problem is not drawdown itself —
the problem is excessive drawdown relative to expectancy.
Many traders attempt to reduce drawdowns by:
Tightening stop-loss levels
Skipping trades
Reducing position size randomly
Over-optimizing entries
The result?
They reduce volatility — but also destroy their statistical edge.
The goal is not to eliminate drawdowns.
The goal is to reduce drawdowns without damaging expectancy.
Step 1: Understand Expectancy Before Modifying Risk
Expectancy formula:
(Win Rate × Average Win) – (Loss Rate × Average Loss)
Before changing anything, ask:
Is the system profitable long-term?
Are drawdowns within historical norms?
Is the drawdown structural or accidental?
If expectancy is positive and robust, you must preserve its core structure.
Drawdown control should enhance stability — not distort edge.
Step 2: Position Sizing Is the Most Efficient Lever
The fastest way to reduce drawdown without affecting expectancy:
Adjust position sizing.
Expectancy per trade remains unchanged.
Risk exposure per trade decreases.
For example:
Risk per trade 2% → Max drawdown 25%
Risk per trade 1% → Max drawdown ~12–15%
The strategy logic remains intact.
In CFD markets, leverage amplifies volatility.
Controlling leverage is often the cleanest solution.
Step 3: Use Volatility-Based Position Sizing
Instead of fixed lot size, use:
ATR-based sizing
Volatility-adjusted exposure
Equity-based scaling
When market volatility expands, reduce size.
When volatility contracts, normalize size.
This stabilizes equity curve variance without altering signal structure.
Step 4: Improve Risk-Reward Structure — Carefully
Many traders tighten stop losses to reduce drawdown.
But tighter stops often:
Lower win rate
Increase whipsaws
Reduce average win size
A better approach:
Keep stop structure logical
Improve reward targeting
Use partial exits
Small structural improvements can smooth equity without eliminating edge.
Step 5: Diversification Reduces System-Level Drawdowns
One of the most powerful methods:
Strategy diversification.
Instead of modifying a single strategy, combine:
Trend following system
Mean reversion system
Different asset classes
Uncorrelated systems reduce portfolio-level drawdown.
This preserves individual expectancy while stabilizing total equity.
Step 6: Avoid Over-Optimization
A common mistake:
Optimizing parameters specifically to reduce historical drawdown.
This creates curve-fitting bias.
Symptoms:
Unrealistically smooth backtests
Poor out-of-sample performance
Strategy collapse in live trading
Drawdown reduction should come from structural risk control — not parameter manipulation.
Step 7: Understand Drawdown Duration
Depth is only half the story.
Duration matters.
A 20% drawdown lasting 3 months is psychologically different from one lasting 18 months.
To reduce duration:
Trade multiple uncorrelated instruments
Avoid capital concentration
Use regime filters
This improves capital efficiency without destroying expectancy.
Step 8: Evaluate Risk-Adjusted Performance
Instead of focusing only on drawdown percentage, evaluate:
Sharpe ratio
Return-to-drawdown ratio
Equity curve smoothness
Monthly variance
Sometimes reducing drawdown slightly while maintaining return significantly improves risk-adjusted returns.
That is the real objective.
What Not to Do
Do not:
Skip trades emotionally
Change rules mid-drawdown
Add discretionary overrides
Reduce size randomly
These actions alter statistical distribution and damage expectancy.
Consistency preserves edge.
The Strategic Perspective
Drawdowns are the cost of accessing positive expectancy.
If you eliminate drawdown completely, you eliminate opportunity.
The objective is controlled drawdown — not zero drawdown.
Professional systematic CFD traders design systems where:
Drawdown is tolerable
Expectancy remains intact
Risk is proportional to capital
This creates sustainable compounding.
Why Tightening Stops Backfires
The instinctive response to a large drawdown is to reduce the loss per trade by moving stops closer. It feels like a direct reduction in risk, and it is the most reliably counterproductive change available.
The reason is that stop distance and win rate are linked. Every system has a distribution of how far price travels against a position before it moves in your favour. Place the stop inside that distribution and you convert trades that would have won into trades that lose. Reduce the stop far enough and you can drive the win rate down faster than you reduce the average loss — the expectancy falls while the trade count rises and the costs accumulate.
There is a diagnostic for this: plot expectancy against stop distance across a range of ATR multiples. Most robust systems show a plateau — a broad region where expectancy is roughly flat — with a cliff on the tight side. The correct stop sits on the plateau, not at the edge of the cliff. If your system is only profitable at one specific stop value, the stop is not protecting an edge; it is fitting noise.
The Correct Hierarchy of Drawdown Controls
Changes that reduce drawdown are not equally safe. Ordered from least damaging to most damaging to expectancy:
Reduce position size. Scales both losses and gains proportionally. Expectancy in R terms is completely unchanged; only the monetary amplitude changes. This is always the first lever and almost never the wrong one.
Add uncorrelated systems or markets. Reduces portfolio-level drawdown while each component keeps its own expectancy, provided the components are genuinely uncorrelated rather than nominally different.
Add a regime filter. Sits out conditions where the system historically performs poorly. Lowers drawdown and often improves expectancy, at the cost of fewer trades and a new parameter to validate.
Adjust the exit to take partial profits. Modifies the return distribution rather than the edge. Usually reduces both drawdown and total return.
Tighten stops or add entry filters. Directly alters the trade population. This is where expectancy is most often destroyed, and it should be the last resort, never the first.
The pattern: controls that act on size and combination preserve the edge; controls that act on selection change it. Reach for the first group by default.
Measuring Whether a Change Actually Helped
After any modification, most traders compare equity curves and stop there. That comparison is dominated by noise. Use these three checks instead, each of which is more informative than the headline result.
Expectancy in R, before and after. If it drops by more than about 15%, you have traded edge for smoothness. Sometimes that is the right trade — but make it deliberately, with the number in front of you.
Drawdown reduction per unit of expectancy lost. If halving drawdown costs you 40% of expectancy, you would usually have been better off simply halving position size, which halves drawdown at zero expectancy cost.
Parameter sensitivity. Re-run with the new parameter perturbed by ±20%. If performance collapses, you have not improved the system — you have found a sharper peak on a noisy surface.
That last check is the one that catches most "improvements". A genuinely better system is flat across a range of nearby parameter values; an overfitted one is a spike.
Recovery Math: Why Depth Matters More Than It Looks
The asymmetry between losses and the gains required to recover them is the core reason drawdown control is not optional.
10% loss requires 11% gain to recover.
25% loss requires 33%.
50% loss requires 100%.
80% loss requires 400%.
The relationship is not linear, and it accelerates sharply beyond about 30%. A system that periodically draws down 50% must generate four times as much return in the following period just to return to break-even — while running at a smaller capital base, with less size available for the recovery.
This is also the argument against maximum position sizing in the pursuit of compounding speed. The mathematical optimum for growth is almost always far above the level a real trader can sustain psychologically, and the gap between those two numbers is where most accounts are lost.
VaR Calculator
Calculate Value at Risk to understand potential portfolio drawdowns in extreme scenarios.
Drawdowns and the Decision to Stop Trading a System
Not every drawdown should be endured. Distinguishing normal variance from genuine decay requires a pre-committed rule, written while the account is at a high.
A reasonable framework:
Within historical maximum depth: no action. Continue at planned size.
Between 100% and 150% of historical maximum: halve size. Do not stop — reduced size preserves optionality while you gather information.
Beyond 150% of historical maximum: stop the system and review. At this point the backtest no longer describes what you are trading.
Any drawdown accompanied by rule deviations: stop immediately. The system is not being tested; something else is happening.
Writing this down in advance is what makes it executable. Deciding whether a drawdown is fatal while you are inside it, with real money moving, is not a decision most traders make well.
Final Thoughts: Stability Over Perfection
Reducing drawdowns without destroying expectancy requires discipline.
The hierarchy of solutions:
Adjust position sizing
Apply volatility control
Diversify systems
Avoid overfitting
Drawdown control is a risk engineering problem — not an indicator problem.
If you approach it scientifically, you can improve stability while preserving long-term edge.
That is the foundation of professional systematic trading.
