Trading performance should be measured as a combination of return, losses, drawdown, risk-adjusted results and execution quality—not by profit alone. A strategy can make money and still expose the account to unacceptable drawdowns. It can also look impressive because a handful of unusual trades carried most of the result.
The purpose of performance analysis is to turn a trade history into evidence. Instead of asking only, “Did the account make money?”, ask how the result was produced, what risk was required, how consistent the process was and which assumptions would have to remain true for the result to be useful.
Start with a record you can trust
Every performance metric inherits the quality of the underlying data. A useful trading journal records the market, setup, entry, exit, position size, costs, initial risk, result and any process notes in a consistent way. Deposits, withdrawals and unrelated account changes should be separated from strategy profit and loss where possible.
Just as important, the record should preserve what was known before the trade. If the original thesis is rewritten after the result, the journal stops being evidence and becomes a story shaped by hindsight.
Win rate tells you frequency, not quality
Win rate is the proportion of completed trades classified as winners under a stated convention. It answers one useful question—how often the strategy wins—but it says nothing about how large the wins and losses are.
A lower win rate can coexist with positive performance when winners are sufficiently larger than losers. A higher win rate can still lose money when occasional losses are much larger. There is no universal “good” win rate without the payoff distribution, costs and strategy context beside it.
Average win and average loss show the size of the payoff
Average win and average loss describe the typical size of positive and negative outcomes in the sample. Read them together with win rate. A strategy that wins often but gives back several normal wins in one loss behaves very differently from a strategy that loses often but occasionally captures a much larger move.
Averages can hide skew and outliers. If two or three trades dominate the sample, inspect the full distribution instead of allowing the average to create a false sense of stability.
Expectancy combines frequency and payoff
Expectancy estimates the average outcome per trade from the observed win probability and the average size of wins and losses. With losses expressed as a positive magnitude, a simple two-outcome form is:
E = pw × AvgWin − pl × AvgLoss
The number only makes sense when the unit is clear. Expectancy can be expressed in currency, percentage points or units of initial risk such as R. Costs should be treated consistently. Most importantly, historical expectancy is an estimate from a sample; it is not a promise about the next trade.
Profit factor compares gross gains with gross losses
Profit factor is gross profit divided by the magnitude of gross loss. It gives a compact view of how much gross profit the sample generated for each unit of gross loss, but it should not be treated as a complete score for a strategy.
A similar value can come from very different trade distributions and drawdown paths. If the sample contains no losing trades, the denominator is zero, so the ratio should be reported as undefined or handled explicitly rather than advertised as an infinite edge.
Drawdown shows what happened to capital along the way
Two strategies can finish with the same return while exposing capital to very different journeys. Drawdown measures the decline from a previous equity peak to a later low. Maximum drawdown is the deepest such decline observed over the chosen sample and measurement frequency.
That word “observed” matters. A historical maximum is not a guarantee that the strategy cannot experience a deeper decline in the future. Drawdown should therefore be read alongside the broader Risk Management framework, position sizing and the amount of loss the capital base can realistically absorb.
Sharpe and Sortino ratios answer narrower questions
The Sharpe ratio compares differential return with the standard deviation of that differential return. It compresses return and variability into one measure, which can be useful for comparison when the return convention and period are consistent. It does not describe drawdown, tail risk or portfolio concentration by itself.
The Sortino ratio uses downside deviation relative to a stated target or minimum acceptable return instead of total variability. Different implementations can use different targets, periods and downside-deviation conventions, so the formula and inputs should be disclosed before two values are treated as directly comparable.
MAE and MFE show what happened inside each trade
Maximum adverse excursion (MAE) records the largest unrealized movement against a trade while it was open. Maximum favorable excursion (MFE) records the largest unrealized movement in its favour.
These measures can help evaluate stop placement, exit behaviour and trade management because they describe the path before the final result. They are not the same as account-level maximum drawdown. Data resolution also matters: coarse bars can hide the sequence of intrabar extremes.
The equity curve connects individual outcomes through time
An equity curve shows the cumulative path of strategy or account results. Look beyond whether the line slopes upward. Review drawdown depth, recovery time, flat periods, changes in variability and whether performance is concentrated in a particular market, setup or regime.
A smooth historical curve is still a description of the observed sample. It does not prove that the same path will continue.
Separate mistakes from normal losing outcomes
A losing trade is not automatically a mistake, and a profitable rule-break is not automatically a good decision. Performance review becomes much more useful when process deviations are classified separately from normal strategy variance.
For example, record whether the setup was valid, whether size followed the risk rule, whether execution matched the plan and whether a discretionary change had a defined reason. This connects performance measurement back to Trading Systems: the system defines what should happen, while analytics shows what actually happened across a sample.
Review performance in a sensible order
- Verify the data. Make sure trades, costs and account adjustments are recorded consistently.
- Define the unit and period. Decide whether results are being measured in currency, percentage, R or another stated unit.
- Read win rate with payoff size. Neither number is meaningful enough on its own.
- Calculate expectancy and profit factor. Then inspect whether a few unusual trades dominate them.
- Measure drawdown. Look at both depth and the path of recovery.
- Use risk-adjusted metrics carefully. State the return, target and volatility conventions behind them.
- Inspect the trade path. MAE, MFE and the equity curve can reveal information that the final P&L hides.
- Separate process errors from strategy variance. Do not reward a rule-break simply because it made money.
- Segment only when the sample supports it. Market, setup and regime comparisons can help, but tiny groups invite false conclusions.
No single metric proves a trading edge
A metric is useful when it answers a defined question. Win rate describes frequency. Payoff size describes magnitude. Expectancy combines the two. Drawdown describes capital damage along the path. Sharpe and Sortino summarize particular forms of risk adjustment. MAE and MFE describe what happened inside trades. The equity curve shows how results accumulated through time.
None of them, alone, proves robustness or future profitability. A professional review uses several measures because profitability, risk, path, execution and repeatability are different questions. The final judgment should also consider whether the rules were followed and whether the sample is large and varied enough to support the conclusion being drawn.
The practical aim of trading analytics is not to find the most flattering number. It is to build an honest picture of how a process earns, loses, behaves under pressure and uses capital so the next decision can be based on evidence rather than memory.