Risk, expectancy and drawdown: look beyond win rate

What is trading expectancy?

For a simple model, multiply the win rate by the average winner. Then subtract the loss rate times the average loser and the average trading cost. The result is the estimated average profit or loss per trade.

If outcomes include breakeven trades, use their actual probabilities and include their costs. More generally, the average of all net trade outcomes is the sample net expectancy. Be consistent: do not subtract costs twice when your average wins and losses already include them.

Suppose an illustrative strategy wins 40% of the time, with an average gross winner of $300, average gross loser of $150 and costs of $10 per trade. Its model expectancy is 0.40 × 300 − 0.60 × 150 − 10 = $20 per trade. These are hypothetical inputs, not a measured trading result or a forecast.

Net expectancy = p × W − (1 − p) × L − C
Break-even win rate = (L + C) / (W + L)

Why can a higher win rate be worse?

Consider a second invented model: 70% winners averaging $100, 30% losers averaging $300 and $10 in costs per trade. Net expectancy is $70 − $90 − $10 = −$30. Frequent wins can feel reassuring while occasional larger losses dominate the economics.

Payoff ratio is average winner divided by average loser. It is different from a planned target-to-stop ratio. A nominal 2:1 target does not guarantee a realized 2:1 payoff because trades may exit early, slip, gap or be managed differently. Use actual realized outcomes when evaluating an established sample.

Exit changes affect both frequency and magnitude. Taking profits earlier may increase win rate while reducing average winner. Widening stops may avoid some small losses while increasing the size of losses that remain. Assess the combined distribution instead of declaring either change an improvement from win rate alone.

How certain is an estimated expectancy?

An average from past trades is only an estimate. It depends on the trades, market conditions and costs in that sample. A few large winners can make the average look much better.

Trade outcomes may cluster because market regimes persist. Treating every trade as an independent draw can understate uncertainty. Resampling trades or blocks of trades can illustrate sensitivity to the observed sample, but it cannot invent unobserved crises or prove that the next period resembles the past.

Ask whether the estimated advantage is large enough to survive plausible errors in spread, slippage and average loss. If adding a modest cost allowance erases it, that fragility belongs in the main conclusion. Reporting a precise dollar average should not imply equally precise knowledge.

What does drawdown add to the picture?

Drawdown is the drop from an account high. Falling from $10,000 to $8,000 is a 20% drawdown. Getting back to $10,000 then requires a 25% gain because the account is starting from a smaller number.

Maximum drawdown is the largest observed peak-to-trough decline in the selected series. It is backward-looking and sample-dependent. It is not an upper bound on future losses. State whether equity includes open positions, whether it is sampled intraday or only at trade close, and how deposits and withdrawals are treated.

Two samples with the same trade outcomes in different orders can have different drawdowns under fixed sizing. Under proportional sizing, the order can also affect interim exposure and practical constraints. Average expectancy alone does not describe whether a trader can fund or tolerate the path.

Drawdown % = (prior peak − current equity) / prior peak × 100
Recovery required after loss d = d / (1 − d)

How do fixed and trailing loss limits differ?

A fixed loss floor stays at a specified level. A trailing floor rises according to a defined reference, such as a new equity high. The details matter: a floor based on intraday unrealized peaks can tighten during a trade before any profit is realized.

For illustration, begin at $10,000 with a $1,000 trailing allowance. If the reference peak rises to $10,600, a simple trailing floor becomes $9,600. That differs from a fixed $9,000 floor. Real account programs may cap the trail, use end-of-day values or apply other rules. Verify current terms rather than transferring this simplified example to a particular provider.

The practical lesson is to model the actual constraint as part of the strategy’s path. A rule that looks acceptable on a final equity chart may violate a loss floor along the way. This discussion does not endorse funded-account programs or any particular account structure.

How does position size connect to planned risk?

For stocks and other simple instruments, first find the dollars at risk per unit from the entry, stop and costs. Divide the total risk amount by that number and round down.

For example, a $100 risk budget, a $0.50 entry-to-stop distance and $0.02 estimated cost per share permit 192 whole shares under this simplified model: floor(100 / 0.52). Planned risk is $99.84. This does not establish that the trade is suitable, affordable or within margin limits.

A stop is an instruction, not insurance at a guaranteed price. Gaps, halts and poor liquidity can produce a larger loss. Account for correlated positions and aggregate exposure; splitting one market bet across several symbols does not necessarily diversify it. The position-size tool uses transparent arithmetic and does not model options, nonlinear contracts or portfolio margin.

What should go on a useful risk review?

Keep the units consistent and show assumptions next to the outputs. Distinguish dollars from R-multiples, gross from net and planned risk from realized loss. A concise review should reveal both the average trade and the worst parts of the path.

Before treating an estimate as actionable, evaluate later data, stress costs and study clustered losses. Set limits in light of your actual resources and constraints rather than copying a universal risk percentage. No formula can make that personal decision.

• Net expectancy, trade count and sensitivity to large outliers.
• Average winner, average loser and realized payoff ratio.
• Maximum drawdown, recovery time and open-equity exposure.
• Worst losses, gaps and cost assumptions.
• Aggregate position risk and a documented review process.