MAE and MFE record the largest adverse and favourable excursions while a trade is open. Reviewing those paths can help you test alternative exit rules, but it does not reveal what a stopped-out position would have done under a different rule. This article explains the measurements, selection biases and how to evaluate a proposed change.

What MAE and MFE are

MAE, Maximum Adverse Excursion, is the greatest adverse deviation while the position is open; MFE, Maximum Favorable Excursion, is the greatest favourable deviation in that interval. This article records MAE as a positive loss magnitude, and MFE as a positive gain magnitude, floored at zero if there was no favourable excursion. Other reports can use signed values, pips, money or initial risk R; state the convention before comparing them.

The measurements add path information to final P&L, but they do not establish optimal stop loss and take profit levels. For a fixed long OTC Forex position, use the actual entry and the relevant liquidation Bid; for a short, use Ask. A one-sided chart, commissions or currency conversion can make a price excursion differ from the greatest net monetary loss.

Mean capture can differ from aggregate capture

Both positions have positive results and positive MFE. Before comparing reports, decide whether each trade has equal weight or is weighted by its MFE.

Scroll sideways to see the full diagram

Mean of trades or share of total MFE?Two fictional fixed-size positions have results 8 and 20, with MFE 10 and 100 in the same money unit. Their individual ratios are 80% and 20%. Equal trade weights give 50%; the ratio of sums gives about 25.45%.TWO WAYS TO SUMMARISE MFEA: result 8 / MFE 10 = 80%B: result 20 / MFE 100 = 20%One length scale for both positions.Mean of ratios: (80% + 20%) / 2 = 50%.Ratio of sums: (8 + 20) / (10 + 100)= 28 / 110 ≈ 25.45%.

Each position has fixed volume; one money unit, no costs or conversions, and MFE only during the position's life. Numbers are fictional, not exit targets or predictions of future performance. TradingView — Favorable excursion confirms excursion is measured while a trade is open; the displayed aggregations are our own arithmetic.

Light bars show MFE; dark bars show achieved results on the same scale.

50% answers the question about the average individual ratio across two trades. 25.45% answers what share of total MFE was realised in aggregate. In the second calculation, larger MFE receives more weight: B contributes 100 of the denominator's 110 units. Do not select the more favourable measure after seeing results. Specify method, unit and treatment of zero MFE before comparing periods. Neither ratio establishes that the maximum was recognisable in real time.

What we measure on a single trade

Consider a hypothetical long EUR/USD entered at 1.0850 with a stop at 1.0820. Its exit target is a planned discretionary rule, not an already active take-profit order: price reaches 1.0905 before the trader later exits at 1.0890 for 40 pips. This permits the 55-pip MFE below. With an active take-profit at 1.0890, the execution sequence and fill would need separate explanation.

  • Lowest Bid 1.0830. MAE is 20 pips in the positive-magnitude convention: two thirds of the initial 30-pip stop distance.
  • Highest Bid 1.0905. MFE is 55 pips; the later realised price gain is 40 pips.
  • Capture ratio. 40 / 55 ≈ 72.73%. This describes the stated path, not an achievable maximum or a universal target.

A single trade can reveal a data inconsistency, but it cannot establish a persistent edge. A fixed threshold of one hundred observations does not resolve dependence, selection, market coverage or how many candidate rules were tried. Record those limits before using the sample to choose exits.

Calibrating the stop loss with MAE

The existing stop limits which paths remain observable as open positions. In a model with a fixed 20-pip stop, continuous quotes and execution without slippage, a winning position cannot remain open through a 28-pip adverse excursion. First reconcile time bounds, units, stop modifications, costs and execution. A winners-only mean does not show the cost of losing trades or what stopped positions would have done under another rule.

  1. Collect a defined sample. Record the strategy, timeframe, trade count and market coverage. A hundred observations can still be selected, dependent or unrepresentative; fewer observations are not automatically meaningless. State uncertainty before using excursions to select new rules.
  2. Inspect winners and losers separately. A winners-only group describes selected outcomes; retain the losing trades when testing what a changed stop would do to the whole strategy.
  3. Inspect the distribution, not just its mean. Winning trades are a selected group, and active stops truncate their observable paths. A 25-pip winners-only mean does not establish how many trades a 20-pip stop would clip under another strategy; reconstruct valid complete paths and test the alternative.
  4. Test a candidate stop. A winners-only mean is affected by selection and by the existing stop, which truncates paths. Compare candidate rules on complete historical paths and separate evaluation data. Recalculate volume for each distance; a mean multiplier alone cannot establish that a wider stop improves expectancy.
  5. Recalculate volume and costs. A wider stop with the same chosen cash-risk budget generally requires less volume under the stated linear model. Round to the instrument’s permitted step and test the resulting complete strategy.

Calibrating the take profit with MFE

Maximum favourable excursion is known retrospectively. A low capture ratio does not prove that the live rule could identify the peak. In a second hypothetical model, an entry at 1.1000 reaches Bid 1.0980 and 1.1050 while open, then exits at 1.1040: MAE 20 pips, MFE 50, realised price gain 40 and capture 80%. A later 1.1070 is outside the trade. An active take-profit executed on first reaching 1.1040 would instead end the measurement there.

Volume changes require a separate monetary path. In a hypothetical linear contract, buy ten units at 100, sell nine at 110 and the last at 120, without costs or conversion. Before the partial exit, unrealised profit is 100; afterwards, realised 90 plus unrealised 10 is still 100. At 120 it is 90 + 20 = 110. Price MFE of 20 multiplied by the original ten units would give 200, overstating this path’s maximum combined result by 90. Preserve the execution sequence and remaining volume rather than applying a late price peak to units already sold.

  1. Describe the measurement. State price versus money, signed values versus magnitudes, costs and the lifetime boundary. MFE equal to zero makes a capture quotient undefined; with a losing result and positive MFE, the quotient can be negative.
  2. Compute the capture ratio consistently. State whether you use the ratio of total realised gains to total MFE or the mean of individual capture ratios; they need not agree. There is no verified universal retail 50–60 percent or professional 70–80 percent benchmark. Compare alternatives on matched data rather than declaring a fixed target optimal.
  3. Write one testable exit hypothesis. A fixed target, an ATR-based trailing stop or partial exits changes the strategy differently. A 70% ratio or a multiple of mean MFE does not establish which one works best.
  4. Evaluate the complete revised rule. Separate development and evaluation data, record all tried variants and include net outcomes, drawdowns and exposure. Fifty trades is not a proof of persistence or causality.

A hypothetical data-consistency check

The hypothetical table below demonstrates a data-consistency check. Excursions must be measured only while the position is open, using the price side relevant to execution. Before testing a wider stop or another exit, reconcile any recorded excursion that appears incompatible with an active protective order.

Illustrative stop and excursion checks
Original stop loss20 pips in the stated hypothetical baseline; verify whether the stop was active and how it was executed.
MAE consistency checkA 28-pip adverse excursion cannot coexist with an active 20-pip stop and ordinary continuous execution before a winning exit; reconcile the data first.
Candidate stopDefine and test an alternative; 1.1 times a winners-only mean does not establish an optimal distance.
MFE consistency checkA 52-pip pre-exit excursion with an active 30-pip take-profit also requires an execution or measurement explanation.
Capture ratio30/52 = 57.6923 percent only if the realised gain and pre-exit MFE are valid and measured consistently.
Candidate exit logicTest closing half at a defined level and trailing the remainder against the unchanged baseline.
Evaluation resultMeasure on separate evaluation data; no universal professional capture-ratio benchmark is assumed.
Evaluation limitsNo improvement in win rate, reward-to-risk or annual profit follows from the incomplete illustrative data alone.

Changing stops or exits changes the trading strategy even if entry signals remain the same. Evaluate the revised rules on separate data and compare risk-normalised net outcomes. The table above is a consistency exercise, not evidence of an observed improvement.

MAE describes the recorded adverse path while a trade was open. Use it as data for testing candidate rules, with care about truncation by the original stop and selection of winners.

Tools — how to measure MAE and MFE

  • Edgewonk: the current product uses an annual subscription. Check the official pricing page, supported import formats and required price data before assuming that imported statements alone provide accurate excursion measurements.
  • MetaTrader 5 reports: official trading and tester documentation describes MAE/MFE during position lifetimes. Reconcile a few individual records before assuming a broker statement or an imported history contains sufficient price data.
  • TradingView Strategy Tester: built into the platform — under "List of trades" it shows adverse and favourable excursion for every position the script opens.
  • A spreadsheet: retain explicit entry/exit timestamps, price-side observations, volume changes, costs and calculation conventions. No fixed setup time is established, and missing quote history cannot be recovered by adding columns.

What to do tomorrow — a plan for the first week

  1. Add documented MAE and MFE fields. Use the professional trader journal template or your own sheet, specifying units, signs and the measured interval. Mark missing inputs before drawing conclusions from a completed row.
  2. Measure excursions from valid price data. Use a source with sufficiently detailed bid/ask history over each position’s actual lifetime. Trade statements may not contain those extremes. State any approximation and do not guess missing values.
  3. Summarise a defined sample. Record its period, count and coverage, then compare mean, median and distribution for winners and losers. This describes the observed strategy; it does not by itself diagnose an optimal stop or target.
  4. Evaluate the revised rule separately. Freeze the candidate exit and compare it with the baseline over a defined forward or untouched historical period. Include costs and drawdowns and acknowledge uncertainty: fifty to one hundred trades do not prove the observed difference is causal or persistent.
  5. Schedule a review and preserve versions. A quarterly cadence is one possible procedure, not a guarantee that drift will be caught before a loss. Keep the old rules and data for comparison. For context see keeping trading statistics and the Traders’ Workshop in our library.