Profit Concentration — P&L
Share of gross profit produced by the top 5% of trades by net P&L. High values mean the result rides on a few outsized winners rather than a broad edge.
- Computed from
- Trades list
- Scope
- Single report
- Range
- 0 – 100%
- Direction
- Lower is better
Profit Concentration measures how much of your winning profit came from just your few best trades. It adds up your biggest winners and divides by the total profit from all your winning trades — so a high number means the whole result rests on a handful of trades. Pull those few out and a concentrated strategy goes flat; a well-spread one barely notices.
How it's calculated
Profit Concentration = (profit from the top 5% of winning trades) / (total profit from all winning trades) × 100
- top 5% of winning trades
- the best ceil(N × 5%) trades — at least one, so on a small record it is literally the single best trade
- total winning profit
- the sum of every winning trade; losing trades are not counted on either side
k = ceil(N × 0.05), so it scales with the record (200 trades → 10; 2,000 → 100). Both numerator and denominator use gross winning profit only (winners summed; losers excluded). Computed per axis: account currency (P&L) by default, plus a sizing-agnostic Return (%) and pips% version. The Return axis shares profit in log space so the figure stays consistent with how Total Gain compounds. A companion top-1% figure shows the same shape at a tighter cutoff.On the P&L axis this is the dollar profit of your best 5% of trades over your total dollar profit. It's affected by position sizing — bet bigger on a few trades and they'll dominate the share.
What it tells you
Check the sample size first. "Top 5%" is only a percentile if you have enough trades to fill it: 20 trades → top 5% is 1 trade, 200 → 10, 2,000 → 100. Below ~100 trades the metric is barely distinguishable from "what did your single biggest winner do," and a low number on a tiny record is uninformative, not reassuring. Then read it with the strategy style — the same number means different things for different systems.
| Value | Reading | Notes |
|---|---|---|
| < 30% | Well spread — robust | Profit comes from across the book, not a few trades. Repeatable. (Trend-following may sit higher by design — read with style.) |
| 30 – 50% | Moderate | Some lean on the best trades, still broadly supported. Normal for many systems — but trend-following is supposed to be concentrated, so read with style. |
| 50 – 75% | Concentrated — fragile | The best handful carry most of the result — check the trade count before trusting it. For trend-following this is expected; for a high-win-rate system it is a warning. |
| > 75% | Extreme — a few trades ARE the strategy | Remove the top slice and little is left. Treat as a few jackpots, not an edge — unless it is trend-following, where a few huge winners carrying it is the design. |
Worked example
A 200-trade record's winning trades produced +$10,000 of gross profit (that's the denominator — winners only, before losses are netted out). But the best 5% — its ten strongest trades — made +$8,000 of those winnings. Concentration is 8,000 / 10,000 = 80%. Strip those ten trades and all the other winners combined made just +$2,000 — so almost the entire winning side came from ten trades. The honest question: is this a 200-trade edge, or ten jackpots wrapped in 190 trades that went nowhere? Now picture the same +$10,000 of winnings with the top 5% contributing only +$2,500 (25%): the other winners still made +$7,500 — spread broadly, that is a repeatable system.
Better than the residual number is the residual curve: re-plot the equity curve with the top 5% of trades deleted. A real edge still slopes up; a jackpot-dependent one goes flat. The number tells you how much rests on a few trades; the curve tells you whether anything's left.
Pitfalls
- High isn't universally bad — the headline. Trend-following and breakout systems are concentrated by design: a few huge winners are supposed to carry a sea of small losers. High concentration there is the strategy working, not failing. Always read it with the style.
- Depends on the trade count. The "top 5%" of 20 trades is one trade; of 2,000 trades it's a hundred. A high share from a tiny record says far less than the same share from a long one — check N first.
- The curve-fit tell. Concentration is also how you smell an over-optimized backtest. When a parameter sweep "finds" a strategy, it often just found a few trades that happened to win — strip them and the edge evaporates. High concentration in-sample, especially with a thin out-of-sample, is a near-certain non-repeater. The real test is whether concentration stays low in the unseen period.
- Silent on why the big trades won. It measures that a few trades dominate, not whether they came from a repeatable setup or pure luck. A few big winners from one news event or one symbol is far more fragile than the same profit scattered across time and instruments. Two records at 70% can be opposites — pair it with the worst-case and the win rate.
- One outlier can dominate. On a small record a single freak trade swings the whole figure. Glance at the top-1% companion: if it's already near the top-5%, the result rides on one trade.
- It's a tail metric — read it with its siblings. Concentration is the right-tail cousin of skewness and tail ratio; together they describe how lopsided the result is.
Concentration and style
The number only means "fragile" relative to what the style should produce — and the same number gives the opposite verdict by style. For mean-reversion (betting price snaps back, many small consistent wins) or scalping (lots of tiny quick trades) — both built on many small wins — high concentration is a red flag: the steady edge those systems claim has quietly died, and a few outliers are masking it. For trend-following (riding a few big moves) or breakout (jumping in when price breaks its range) — both built on rare large winners paying for many small losses — high concentration is expected and fine; low concentration there would actually be the surprise. Seventy percent is the system working as designed on a trend-follower, and a warning on a mean-reverter. So before reading a high number as a weakness, ask what the strategy is supposed to look like.
Related
Tail Ratio and Skewness measure the same right-tail lopsidedness from the equity curve, Expectancy is the per-trade edge concentration can flatter, and Win Rate the frequency a few jackpots can disguise.