Trading Statistics Explained: Win Rate, Expectancy, Profit Factor and Drawdown

Win rate alone says almost nothing about a trading method. Expectancy, profit factor and drawdown combine how often you win, how much you win and how badly it can go, and each of them has blind spots.

Published 11 October 2026 · Updated 3 November 2026 · 6 min read

Key takeaways

Win rate only counts how often trades win; without the average size of wins and losses it says almost nothing about results.
Expectancy combines the two: win rate times average win minus loss rate times average loss, the average result per trade, always to be measured net of costs.
Profit factor is gross profit divided by gross loss; above 1, the winning trades made more than the losing trades lost over the sample.
Maximum drawdown is the largest fall of the equity curve from a previous peak, the figure that decides whether a trader or an account can keep going.
Every statistic is an estimate from a sample: few trades, a single market regime or missing costs make all of them unreliable.

Trading statistics: the short answer

Trading statistics summarise a list of trades, real or simulated, into a few numbers that describe how a method behaves: how often it wins (win rate), how much it wins and loses on average, what it earns per trade on average (expectancy), how gains compare with losses overall (profit factor), how deep its worst losing period was (maximum drawdown) and how steady its returns were relative to their variability (Sharpe-type ratios).

None of these numbers is useful alone, and none of them predicts the future: they describe a sample. Learning them is mostly about not being impressed by the wrong figure, whether it is a high win rate, a large total profit or one exceptional month. The definitions below follow those published by NinjaTrader for the performance reports of NinjaTrader 8.

Win rate, average win and average loss

Win rate, which NinjaTrader calls Percent profitable, is the number of winning trades divided by the total number of trades. It is the most quoted statistic and the least informative on its own. A method that wins most of the time can still lose money if its few losses are large; a method that wins rarely can make money if its wins are much larger than its losses.

The missing information is size. Average win is gross profit divided by the number of winning trades; average loss is gross loss divided by the number of losing trades. Their ratio, average win divided by average loss, is often called the payoff ratio, and NinjaTrader reports it as Ratio Avg Win / Avg Loss. Win rate and payoff ratio pull against each other: close targets tend to raise the win rate and lower the payoff, distant targets do the reverse. Neither number is good or bad in isolation.

Before costs, the break-even win rate for a given payoff ratio is 1 divided by (1 + payoff ratio). With wins and losses of equal size, a method must win half its trades just to break even; with wins twice the size of losses, a third is enough. This is arithmetic, not a target, and real costs push every break-even point higher.

Expectancy: the average result per trade

Expectancy is the average amount a method makes or loses per trade over the sample. It can be written as win rate × average win − loss rate × average loss, where the loss rate is the share of losing trades; the result equals total net profit divided by the number of trades, which NinjaTrader reports as Average Trade. It can be expressed in money, in ticks or in multiples of the initial risk, often written R.

Expressing expectancy in ticks or in R makes it comparable across contracts and position sizes, and it makes the cost question concrete. If the average trade is worth about as much as a tick or two of costs and slippage, a small change in execution can turn it negative, a point developed in what scalping futures really involves. Expectancy should always be measured net of commissions, fees and slippage.

Expectancy is an average, and averages hide sequences. Two methods with the same expectancy can have very different losing streaks and drawdowns, which is why it is always read together with the figures that follow.

Hypothetical example — Illustrative figures, not market data. Two invented methods: a high win rate does not by itself produce a positive expectancy.
	Method A	Method B
Win rate	70%	35%
Average win	2 ticks	9 ticks
Average loss	5 ticks	3 ticks
Payoff ratio	0.4	3
Expectancy per trade	−0.1 tick	+1.2 ticks
Profit factor	0.93	1.62

Profit factor

Profit factor is gross profit divided by gross loss, the definition NinjaTrader uses. A value above 1 means the winning trades made more money than the losing trades lost over the sample; below 1, the opposite. It packs win rate and payoff ratio into a single figure, and it is tied to expectancy: a profit factor above 1 always goes with a positive average trade.

Its weaknesses are those of any ratio built on a sample. With few trades, one large winner can inflate it. It ignores the order of trades, so it says nothing about drawdown. It depends on costs: a profit factor computed before commissions and slippage overstates the result, especially for short-term methods with many small trades. And it rises easily with optimisation, which is why an in-sample profit factor is a weak guide, as explained in how to backtest a futures strategy.

Maximum drawdown and recovery

Drawdown is the fall of the equity curve from a previous peak. Maximum drawdown is the largest such fall over the sample, measured in money or as a percentage of the peak. NinjaTrader computes it on realized profit, from a local maximum to the following local minimum. Related figures include the longest time taken to recover to a previous peak (Max. Time to Recover) and the longest run of losing trades (Max. Consec. Losers).

Drawdown is the statistic that decides survival. Two methods with the same expectancy and the same profit factor can produce very different maximum drawdowns, depending on how their losses cluster. Recovery is also asymmetric: after losing a fifth of an account, the remainder must grow by a quarter to get back to the peak; after losing half, it must double.

A historical maximum drawdown is not a ceiling for the future. It is the worst that happened in one sample, and a longer or different sample usually contains worse. Many traders size positions so that a drawdown well beyond the historical one would remain survivable, a principle covered in futures trading risk management. Monte Carlo simulation, available in NinjaTrader’s Strategy Analyzer, randomises trade results across many simulated runs to show a range of possible outcomes rather than a single path.

Sharpe and Sortino: return relative to variability

The Sharpe ratio divides an excess return by the standard deviation of returns: how much a method earned per unit of variability. NinjaTrader computes it from monthly profits, with the benchmark rate set to zero, and sets it to 1 when there is one month of history or less. The Sortino ratio uses only downside variability in the denominator, so upside swings are not penalised.

These ratios help compare methods measured on the same basis, but their values depend on the period, on the frequency of the returns used and on how they are annualised, so figures from different platforms are rarely comparable. They also assume that variability captures risk. That fails for methods with rare, large losses: a strategy that collects small, frequent gains while carrying the risk of an occasional large loss can show a smooth curve and a flattering ratio until that loss arrives.

Reading trading statistics together: a checklist

A short routine works for any report, whether it comes from a backtest or from a journal of live trades.

Number of trades first: a few dozen trades are rarely enough to separate skill from luck.
Expectancy net of all costs, in ticks or in R, compared with your typical slippage.
Profit factor with and without the largest winning trade.
Maximum drawdown and longest losing streak, and whether you could realistically trade through both.
Results by period or by market condition: does one month or one regime carry the whole result?
Simulated or live: statistics from a backtest are hypothetical and usually better than live ones.

Put it into practice

Frequently asked questions

What is a good profit factor?

There is no universal threshold. A profit factor only means something together with the number of trades, the costs included and the conditions tested. High values from an optimised backtest often shrink out of sample, so compare it with an honest out-of-sample or live figure rather than with a rule of thumb.

Is a high win rate better than a high payoff ratio?

Neither is better in itself. What matters is the combination, measured as expectancy net of costs, and how the method behaves in drawdown. Methods with a high win rate often carry occasional large losses; methods with a low win rate often go through long losing streaks.

How many trades are needed for reliable statistics?

There is no fixed number, but more is better, and the trades should come from varied conditions. The more the results of individual trades vary, the more trades are needed before the average means anything. A few dozen trades rarely settle the question.

What is the difference between the Sharpe and Sortino ratios?

Both divide an excess return by a measure of variability. Sharpe uses the standard deviation of all returns; Sortino uses only downside variability, so large gains do not lower it. In NinjaTrader both are computed from monthly figures, with the benchmark rate set to zero.

Should trading statistics include commissions and slippage?

Yes. Statistics before costs describe price movements, not trading results. For short-term methods with many trades, the gap between gross and net figures can decide whether a method makes or loses money.

Sources

Statistics Definitions (NinjaTrader 8) — NinjaTrader Help Guide
Performance Displays (NinjaTrader 8) — NinjaTrader Help Guide
Monte Carlo Simulation (NinjaTrader 8) — NinjaTrader Help Guide