What is a worked example of performance metrics?
A worked example of performance metrics is a fully transparent, step-by-step scenario where you define the terms, state every assumption, and then compute metrics from the given numbers. The goal is not to predict future results; it is to show how the mechanics work so you can apply the same definitions to your own data and independently verify the arithmetic.
Performance metrics are measurable values that summarize performance using agreed formulas. In a forex context, the “inputs” often include profit or loss, the time span being measured, position sizing, and costs such as spreads or commissions. Because market conditions, execution quality, and provider-specific fee structures vary, the same formula can produce different metric values even when the underlying approach is unchanged.
How a worked example works (mechanics and definitions)
A simple way to structure a worked example is to separate stable mechanics from variable conditions.
Stable mechanics (definition + math):
- Period profit (P): total net profit or loss for a fixed time window.
- Number of trades (N): count of completed trades included in the calculation.
- Average profit per trade (AP): AP = P / N.
- Win rate: fraction of trades with net profit greater than zero.
- Average win / average loss: separate the mean outcomes of profitable and losing trades.
- Profit factor (PF): PF = total gains / total losses, using absolute values for losses.
Variable conditions (what can change reality):
- Costs: spreads, commissions, swap/rollover, and any financing charges.
- Execution: slippage, partial fills, and differences between backtest and live fills.
- Jurisdiction and reporting: how trades are grouped into periods and how deposits/withdrawals are treated.
To keep the worked example verifiable, you must state which costs are included, how trades are counted, and what “net profit” means.
Evidence or example: one fully numeric worked scenario
Assume the following, and use exactly these assumptions for every calculation:
Given
- We measure performance over one fixed period.
- The period includes N = 10 trades.
- Total net profit P = 180 (currency units).
- Costs are already included inside the net profit for each trade (so we do not subtract costs again).
- Trade outcomes by net profit (after costs) are:
- 6 winning trades: +40, +35, +30, +25, +20, +10
- 4 losing trades: -30, -25, -20, -15
Step 1: Win rate
- Winning trades = 6 out of 10.
- Win rate = 6/10 = 0.60 = 60%.
Step 2: Average profit per trade
- AP = P / N = 180 / 10 = 18.
Step 3: Average win and average loss
- Total gains = 40 + 35 + 30 + 25 + 20 + 10 = 160.
- Total losses (absolute value basis) = 30 + 25 + 20 + 15 = 90.
- Average win = 160 / 6 ≈ 26.67.
- Average loss (magnitude) = 90 / 4 = 22.50.
Step 4: Profit factor
- PF = total gains / total losses = 160 / 90 ≈ 1.78.
Step 5: Consistency check
- Net profit should equal total gains minus total losses with signs:
- 160 − 90 = 70, which conflicts with the stated period net profit P = 180.
That mismatch reveals an important worked-example principle: if your numbers do not reconcile, either the assumed trade list, the assumed total net profit, or the inclusion/exclusion of costs is inconsistent. For verification, you must correct the inputs until the totals match.
To repair the scenario while keeping the same trade outcomes, we would replace P with the reconciled net profit:
- Reconciled P = 160 − 90 = 70.
- Then AP = 70/10 = 7 instead of 18.
This shows how a worked example can expose data or assumption errors.
Limitations and risks of performance metrics
Material limitations usually come from how performance metrics are defined and from the data that feeds them.
- Cost and accounting mismatch: If some costs are included in trade net profit but you also subtract costs again (or omit costs that were present in reality), metrics become distorted. 2. Period selection effects: Metrics depend on the chosen time window. Changing the start/end dates can change win rate, PF, and averages. 3. Execution differences: Backtests and live results can diverge because fills, slippage, and timing vary. 4. Incomplete trade inclusion: If you accidentally exclude certain trades (e. g. , those opened but not closed within the period), the calculations no longer represent the intended strategy behavior. 5.