MFE/MAE for Crypto Perpetuals: Reading the Trade You Actually Made
Most traders judge a trade by one number: did it close green or red. That number is the smallest amount of information any trade produces. Between the entry and the exit there is an entire path the price walked. It went farther in your favor than where you exited. It went farther against you than wherever you set your stop. The shape of that path is where the actionable information lives, and ignoring it is the difference between a trader who improves quarter over quarter and one who runs the same mistakes for years and calls it variance.
Maximum Favorable Excursion (MFE) and Maximum Adverse Excursion (MAE) are the two numbers that capture that path. MFE is the best unrealized PnL the trade ever showed during its life. MAE is the worst. Together they say: this trade had this much edge available, and you carried this much risk to extract it. Whether you took it home or not is a separate question.
This post is about why those two numbers matter more for crypto perpetuals than for almost any other instrument, what to actually compute, and how to read the result without falling into the easy traps. The point is not "you should journal MFE." Most journals already have MFE and most traders ignore it. The point is what you do with it once you see it.
Why this matters: definitions on Bybit perps
The textbook definitions are simple, but the perp context adds detail.
For a long trade opened at price E and closed at price X, walking through every tick (or candle) between open and close:
- MFE = the highest price reached, expressed either as price, as PnL on your size (USDT), as a percent of entry, or as an R-multiple if you have a stop loss recorded.
- MAE = the lowest price reached, in the same four units. By convention MAE is signed negative — it represents adverse movement.
- Capture ratio (or "captured percent") = realized gross PnL divided by MFE PnL. If you took $100 from a trade where MFE was $400, your capture is 25%.
- Pain ratio = the average loss on losing trades divided by the average MAE on those same trades. Tells you how much of the worst-case drawdown actually became a realized loss versus how much you escaped.
For shorts the directions flip but the logic is identical: MFE is the lowest price (best outcome for a short) and MAE is the highest (worst).
A clean ASCII view of one trade's life:
Long entry $100, SL $95, exited $105
$115 ─ * <- MFE (best price reached)
$110 ─ *
$105 ─ X <- exit
$100 ─ E ─ <- entry
$98 ─ *
$95 ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ SL <- stop never hit
$94 ─ * <- MAE (worst price reached)
Time -->
In R-multiples, with risk-per-unit = $5: MFE_R = +3R, MAE_R = -1.2R, realized = +1R. The trade "worked" — it was a +1R winner. But you walked away from $10 of unrealized to take $5. Capture was 33%. Worth noting? Yes, especially if it happens consistently.
A few crypto-perp-specific wrinkles. First, the "high" and "low" of any candle in your trade window are reachable prices, not derived numbers — Bybit perps trade 24/7 with deep books on majors, so wicks are real and stop runs are common. Second, perps usually have a pair of leverage and notional inputs, which means MFE and MAE in PnL terms scale with both size and leverage in the same direction, but capture and pain ratios are dimensionless — they're percentages of what was achievable. That's important: capture and pain are the comparable metrics across different position sizes. Third, the maximum-favorable timestamp matters. A trade where MFE happened in the first ten minutes and the rest was a slow grind down is a very different trade than one where MFE happened just before exit — even if the MFE number is the same.
Real numbers: when MFE/MAE rewrite the trade story
Let's price a few patterns where the MFE/MAE view changes the conclusion versus the closed-PnL view.
Pattern A: the "winner" that wasn't
You went long ETHUSDT at $3,500 with size 5 ETH (notional $17,500). Stop at $3,470 (risk $30/ETH = $150 total). Exited at $3,520 — a +$100 gross winner, capturing roughly +0.7R.
But MFE during the trade reached $3,580 (notional gain $400 = +2.7R) and MAE never went below $3,495 (drawdown $25 = -0.17R).
Capture: $100 / $400 = 25%.
In closed-PnL view this is a +0.7R win and your weekly summary records it as such. In MFE/MAE view this is a +2.7R available trade where you took 0.7R. If 25% capture is your average across winners, you have a major execution leak — your reads are good enough to find +2.7R setups but your exits monetize barely a quarter of that. The intervention is not "more good entries." It's "fix exit framework." Trail wider, scale out later, hold to a daily close. The diagnosis is impossible without the MFE number.
Pattern B: the "loser" that wasn't really
You shorted BTCUSDT at $70,000 with size 0.2 BTC. Stop at $70,700 (risk $700 × 0.2 = $140). Exited at $70,140, a -$28 loss. Looks like a small clean loss.
But MAE during the trade was $70,650 — drawdown $130, almost 1R, very close to your stop. MFE was $69,800 — gain $40, about 0.3R.
The trade barely worked at all in your direction. You almost got stopped, then you fled near the worst point of recovery. Capture: meaningless because gross was negative. Pain ratio on this single trade: $28 / $130 = 21%. Not bad — you escaped most of the drawdown.
But the bigger story is in MFE: this trade was almost never positive. The thesis didn't play out. You had a position that spent most of its life underwater, briefly poked into profit, and you cut. That's not an execution issue — that's a setup quality issue. You ran a thesis that the market did not validate at any point. If 30% of your "small losers" share that signature — never really worked — you have an entry-quality leak, not an exit-quality leak. The diagnosis goes the opposite way of Pattern A.
Pattern C: the leveraged stop run
You went long SOLUSDT at $185 with 10× leverage, size 100 SOL ($18,500 notional). Stop at $182. MAE on the day: $181.80 — twenty cents below your stop. You got liquidated/stopped out for a -$300 loss. MFE before the stop run: $187.50, $250 of unrealized.
Capture: zero (closed at a loss). MAE-to-realized ratio: 1.0 (you ate the full drawdown).
This is the classic "stop ran two ticks past my level then ripped." If you see this signature on multiple trades — MAE just barely past your stop, then mean-reversion right after — you might have a stop-placement problem. Either you're parking your stop at obvious levels everyone else is also using, or you're sizing too aggressively for the volatility regime and getting stopped on noise. Looking at MAE alone, sorted by "distance past my stop," would surface this pattern in two minutes of journal review.
What to track
The four metrics worth instrumenting on every closed trade:
| Metric | What it answers | Watch when | |---|---|---| | MFE in R-multiples | How big was the available trade? | Average MFE_R per winner — your top-line edge | | MAE in R-multiples | How much risk did you actually carry? | Average MAE_R below -1R = stops getting run | | Capture % on winners | How much of available edge did you take? | Below 50% sustained = exit framework issue | | Pain % on losers | How much of worst-case drawdown became realized loss? | Above 80% = no defense, you eat full MAE |
The frequency to review them is monthly, not per-trade. A single-trade MFE comparison is variance. A 30-trade rolling average tells you whether your execution shape is improving. Looking weekly works if you're high-frequency, monthly if you're swing.
A few comments on the metrics, because the obvious mistake is to over-interpret them:
Capture is bounded by your exit method. A pure breakout-pullback strategy with a fixed 2R target will never have capture above 100% on any single trade and will average somewhere around 60-70% on winners, because by definition you're cutting at a fixed level while MFE keeps walking. That's not a leak — that's the strategy. The capture leak is when you have a "let it run" intent and your average capture is 30%.
MAE_R below -1R is not always a problem. Bybit perps trade across funding cutoffs, news prints, and weekend illiquidity. Wicks past your stop happen. If 5% of your trades show MAE_R between -1.0 and -1.2 and the rest cluster nicely below -1R, that's just market microstructure. If 30% of your trades show MAE_R worse than -1.5R, your stops are getting run systematically — different problem.
Capture and pain interact. If your capture is 80% (great) and your pain is 90% (eating most losers in full), the math says your wins must be much bigger than your losses to be profitable. That is a high-conviction position-trader signature. Whereas a scalper might run 40% capture and 30% pain — they cut both directions fast, take small wins, take small losses. Both can be profitable. The leak is when capture is low AND pain is high — you give up most of your wins and eat most of your losses. That's a "fix the entire framework" signal, not a tweak.
PerpLog computes all four of these numbers per trade, and the dashboard surfaces the rolling capture and pain ratios as Execution Efficiency. Per-trade MFE/MAE can be expensive to compute (it requires backfilling kline candles per trade from Bybit), so the journal does it on demand for closed trades — there's a "compute MFE/MAE" batch action that walks your closed trades and fills the columns. Once filled, sorting your loser list by MAE-distance-past-stop or your winner list by capture-percent surfaces the patterns above in seconds.
Common mistakes
1. Computing MFE on the wrong window. Some trade journals will let MFE "drift" past the actual exit time — e.g., a trade closed at noon but the journal still considers prices through 4 PM. That's not MFE, that's hindsight bias. The MFE window is exactly entry-to-exit, no extra runway. Anything else makes you look worse than you actually executed and dunks on you for a price you couldn't have taken.
2. Mixing position-weighted and full-size MFE. If you scale into a position — say, half-size at $100 and the other half at $102 — your MFE PnL needs to use the actual qty held at each candle, not the final qty applied to the whole window. The naive calculation treats the early candles as if you already had the full position on. That overstates available P&L. PerpLog's MFE engine uses position-weighting whenever scaling executions are present; if your spreadsheet doesn't, your numbers are systematically inflated on scaled trades.
3. Treating capture below 100% as failure. It isn't. Capture above ~80% on every trade means you are exiting at the absolute high tick — which means you are clairvoyant or you are misreading the data. Realistic capture rates for swing traders are in the 40–70% range on winners. The number to optimize is your median, not your max. A median capture that drifts up over six months is improvement.
4. Ignoring time-of-MFE. MFE that arrives in the first 5 minutes of an 8-hour trade is fundamentally different from MFE that arrives 30 seconds before exit. The first is "the market gave me an immediate pop and I didn't take it." The second is "I held through chop and barely made it out." Same MFE number, different lessons. Most journals don't capture time-of-MFE; if yours does, look at the distribution. If too many of your MFE-points cluster at trade open, you have a "take quick wins" leak more than an "exit too early" leak.
5. Using MAE to second-guess every stop. Survivorship bias. The trades you remember as "I should have held — it came back" are the trades that came back. The dozens that hit your stop and kept going are forgotten. MAE is a population-level signal: across many trades, where do my stops cluster relative to MAE? If 90% of my stops are within 0.1R of MAE, that's "stops at obvious levels, getting run, getting reversed." If MAE is consistently far past where I exited the trade, that's "I'm cutting fine but my stops would have caught me anyway." Different actions. Don't tell single-trade stories with population data.
Takeaway
MFE/MAE turn a one-number trade into a four-number trade, and that's the resolution at which leaks become visible. The closed-PnL view tells you whether you made or lost money. The MFE/MAE view tells you whether the trade idea was good, whether your entry was on the right side, whether your exit captured the move, and whether your stop cost you the trade or saved you from a worse one. Tomorrow morning, take your last fifty closed trades and compute capture-on-winners and pain-on-losers. If capture is below 50%, your exits are leaking edge. If pain is above 70%, your defense is leaking too. Pick the larger of the two and spend a month fixing it.