Module 6 · Lesson 15 of 17 · Risk, process, and mindset · About 5 minutes
Reviewing Your Own Trade Statistics: Expectancy, MAE and MFE
On this page
Before you read: Trading Plans, Journaling and Honest Backtesting
What you will learn
- How win rate, average win and average loss combine into expectancy.
- What maximum adverse excursion (MAE) and maximum favorable excursion (MFE) describe.
- Why small samples and selective records make statistics misleading.
How do the basic numbers fit together?
Can a high win rate still lose money?
Expectancy per trade is win rate × average win − loss rate × average loss. In a fictional record of 40 trades, 24 wins averaging $80 and 16 losses averaging $150 give 0.6 × 80 − 0.4 × 150 = 48 − 60 = −$12 per trade. Winning 60% of trades did not make this record positive. These are past, fictional results and say nothing about future ones.
What are MAE and MFE?
How far did each trade go against you before it ended?
Excursions measure the path between actual entry and exit, not what the chart did after exit. Convention used here: MAE/MFE are nonnegative gross price-excursion magnitudes, including entry and exit endpoints; final P&L is net of stated costs. We use the complete invented paths below, with no unlisted between-point extreme. A real sparse bar/quote record might miss excursions, and bid/ask or platform definitions can differ.
For long: MAE=max(entry−within-hold low,0); MFE=max(within-hold high−entry,0). For short: MAE=max(within-hold high−entry,0); MFE=max(entry−within-hold low,0). Multiply points by dollars/point/unit and quantity for dollar magnitudes.
Fictional linear instruments, $10/point/contract, fixed quantity per trade, no scale-ins. All times CST; independent sessions, not actual market data. Gross initial-risk denominator R=|entry−planned stop|×$10×quantity, excluding fees. This chosen R is not an account recommendation or a guaranteed stop fill. Costs are $4 per contract round trip, including all assumed entry/exit charges; no additional slippage in these paths.
| Event | Long session price | Short session price |
|---|---|---|
| Entry, 09:00 / 10:00 CST | 100 | 200 |
| During, 09:01 / 10:01 CST | 99 | 201.5 |
| During, 09:02 / 10:02 CST | 103 | 197 |
| Exit, 09:03 / 10:03 CST | 101 | 201 |
| AFTER exit, 09:04 / 10:04 CST | 106 — exclude | 190 — exclude |
| Metric | Long, 2 contracts; stop 98 | Short, 1 contract; stop 202 |
|---|---|---|
| Initial R dollars | (100−98)×10×2=$40 | (202−200)×10×1=$20 |
| MAE points / dollars / R | 1 / $20 / 0.5R | 1.5 / $15 / 0.75R |
| MFE points / dollars / R | 3 / $60 / 1.5R | 3 / $30 / 1.5R |
| Gross realized P&L | (101−100)×10×2=$20 | (200−201)×10=−$10 |
| Round-trip costs | 2×$4=$8 | 1×$4=$4 |
| Net realized P&L / R | $12 / +0.3R | −$14 / −0.7R |
Long calculation uses within-hold low 99/high 103, not later 106. Short uses high 201.5/low 197, not later 190: a lower short price is favorable, not adverse. The short can have a 1.5R favorable excursion and still finish with a loss. MFE is not profit captured; MAE is not final loss. Gross excursions and net outcomes are different measures, explicitly labeled here. A hypothetical net-liquidation excursion would require a consistent fee/bid-ask mark at every point, not silently reusing the gross column.
This two-trade record only: one $12 win and one $14 loss: total −$2; win rate 1/2=50%; mean net=(12−14)/2=−$1/trade; 0.5×12−0.5×14=−$1. Mean normalized net=(0.3−0.7)/2=−0.2R, using different per-trade R denominators. Do not turn the mean R into dollars using an unspecified common risk. These sample statistics are not forecasts. For risk units see Risk Management Arithmetic: Position Size, Risk-Reward, Drawdown.
Statistics diagram
Read the diagram step by step
- Authored fiction. Education only, not market observations or a forecast.
- Gross MAE versus NET realized outcome
- Exactly TWO supplied observations. x=gross nonnegative MAE dollars; y=net realized P&L dollars. Gross excursions and net outcomes are deliberately different measures.
- Declared measurement convention
- $10/point/contract; costs $4 per contract round trip. Gross initial-risk R excludes fees. MAE/MFE include entry/exit endpoints, not later prices. No unlisted between-point extreme or additional slippage in these invented paths.
- Complete LONG path · CST
- 09:00 entry 100 → 09:01 price 99 → 09:02 price 103 → 09:03 exit 101; quantity 2, stop 98. Gross +$20; costs $8; net +$12 (+0.3R).
- Complete SHORT path · CST
- 10:00 entry 200 → 10:01 price 201.5 → 10:02 price 197 → 10:03 exit 201; quantity 1, stop 202. Gross −$10; costs $4; net −$14 (−0.7R).
- AFTER EXIT · EXCLUDE these extremes
- Not within the actual holding interval; do not use 106/190 to enlarge MAE or MFE.
- All plotted observations are supplied in the two-path table. Gross excursions and net outcomes are deliberately different columns; later prices are excluded. Two examples do not validate exits or a predictive edge.
Illustrative diagram constructed for teaching. Not real market data.
All plotted observations are supplied in the two-path table. Gross excursions and net outcomes are deliberately different columns; later prices are excluded. Two examples do not validate exits or a predictive edge.
Why are samples and records a problem?
Are 15 trades enough to judge anything?
Keep missing records, mixed versions, costs and selection visible. Neither these two complete paths nor the separate 40-trade aggregate establishes a repeatable edge. Retuning an exit to the best historical MFE uses information unavailable when the position was managed. Full-bar highs/lows can also include time before an entry or after an exit inside that bar; do not label those as exact within-position excursions without sufficient timing data.
A run statistic needs an event definition. Under a separately stipulated independent 50% loss model, five specified consecutive future trades all losing has probability (0.5)^5=3.125%. That is not the probability of at least one five-loss run anywhere in a longer record; record length and overlap matter. This example does not estimate a future loss probability from the two trades. See Overfitting and Repainting: When Indicator Results Look Too Good and Risk of Ruin: How Losing Streaks and Bet Size Interact.
What this does not tell you about statistics
- Whether any record will continue.
- Which rules or exits to use.
- How many trades are enough for a particular purpose.
- Nothing here suggests any trade or method.
Method in Practice: John's month, and what it doesn't show
Fictional example. John, Harbor Index futures, the prices, times, costs and results are all invented for teaching. This shows how a written plan can be built and tested. It is not a recommendation, and nothing here says how often this idea works.
Try it on a real chart. Test this idea yourself on past sessions with bar replay. Keep your own count, including the days with no trade.
The idea. Win rate, average win, average loss and expectancy fit together (this lesson). Here they are for one month of John's plan from Volume Profile Explained: POC, Value Area, HVN, LVN: 20 sessions, 9 trades. The point is what these numbers can't tell him yet.
Three outcomes (invented numbers: this shows how to count, not how the method performs)
-
Wins: 4, averaging +$90
-
Losses: 5, averaging −$80
-
No trade: 11 of 20 sessions
Show the month's stats
| Measure | Value | Working |
|---|---|---|
| Sessions | 20 | |
| Trades | 9 | 11 sessions had no setup |
| Wins | 4: +$75, +$130, +$55, +$100 | Total +$360 |
| Losses | 5: −$104, −$76, −$70, −$52, −$98 | Total −$400 |
| Net | −$40 | |
| Win rate | 44% | 4 ÷ 9 |
| Average win / average loss | +$90 / −$80 | |
| Expectancy per trade | −$4.44 | 0.444 × $90 − 0.556 × $80 = −$40 ÷ 9 |
| MFE note | 3 winners came within 0.10 of the second reference | Tempting to lower the target. Not on 9 trades |
| MAE note | 2 of 5 losses filled past the planned exit | Both inside the $150 cap |
Journal: "Slightly negative on nine trades. That's not a verdict either way. What I can act on: process. Zero rule breaks this month."
Test it yourself
- Work out expectancy by hand: win rate × average win − loss rate × average loss.
- Check MFE: how far did winners go past your exits? Don't change targets on fewer than 30 trades.
- Write down how many trades you'd need before trusting a number. Then wait for them.
Where it breaks
- Nine trades can't separate skill from luck.
- Averages hide the order of results. A losing streak feels different from the same losses spread out.
- Hindsight makes MFE look like money left on the table.
Calculators: expectancy
Reflection question: How many trades are in the statistics you trust most, and is that enough?
Check yourself
- Why is the short’s MAE 1.5 points, and why does the later 190 price not count?
- Why is long net +0.3R while MFE is 1.5R?
- Recompute mean net across the two supplied paths. Separately, what is the expectancy of a fictional record in which half the trades win, with a $100 average win and an $80 average loss? Hint: 0.5 × 100 − 0.5 × 80.
- Does 3.125% describe any five-loss run somewhere in a long history?
- The supplied long path eventually earns $12 net but first visits 99 after a 100 entry. What does its MAE reveal, and does the eventual win erase it? Hint: how far they went against you first.
- Why are 15 trades a weak basis for a conclusion about a population win rate or expectancy? Hint: chance.
Reflect on your own process
- Which statistics does your journal compute for every trade, and does it record costs separately from gross results?
- How many recorded trades does your written review process require before you change a rule?
- Does your journal capture the full path between entry and exit, or only the result, and how would you know?
- How many trades are in the statistics you trust most, and is that enough?