A backtest can finish well in profit and still have a narrow foundation. If a handful of trades, a single day, or one friendly market regime supplied most of the result, the total return is describing history accurately but not describing how broadly the edge appeared.
Profit concentration in a prop backtest is the share of historical profit produced by its largest contributors. It matters because prop accounts live through the sequence: an account that needs one exceptional patch of history to look viable has a different risk profile from one that earned across many independent opportunities.
What profit concentration actually measures
Profit concentration measures dependence on the winners that did most of the work. It can be calculated at trade, day, week, instrument, strategy, or regime level, provided the unit is stated before the result is inspected.
The simplest question is useful precisely because it is unglamorous: what share of net profit came from the largest contributor? A second useful question is whether the backtest remains positive when that contributor is removed. Neither test proves the remaining system will work live. Both expose how much the headline total leans on one part of the sample.
| Unit inspected | Question it answers | What it can reveal |
|---|---|---|
| Trade | Did a few exits create most of the result? | Dependence on rare payoff events |
| Day | Did one session carry the period? | Calendar and event clustering |
| Strategy or instrument | Is one sleeve doing all the work? | Hidden portfolio concentration |
| Regime | Did profits arrive only in one market condition? | Regime dependence |
A profitable curve can still be a narrow one
Imagine an illustrative backtest with 100 units of net profit. If its best day made 70 units, the other days made only 30 units combined. The result is still 100 units. But the reader now knows that removing, delaying, or weakening one day changes the story substantially.
That is not an argument for deleting the best day. Real strategies do have unusually good trades. The mistake is treating the accumulated total as if every part of the sample contributed equally.
The same problem appears at a larger scale. A system may look stable over several years but have earned most of its gains during one volatility regime, one directional stretch, or one instrument's unusually favourable conditions. The curve does not label that dependence by itself. A concentration check makes it visible.
This is adjacent to market-regime backtesting, but it asks a different question. Regime analysis separates conditions; concentration asks how much of the result each condition supplied.
Measure the dependency before reading the answer
A useful concentration review is simple enough to reproduce. Start with the same net returns used in the backtest, including costs. Sort the chosen unit from largest profit contributor to smallest, then report the share of total net profit supplied by the largest one and by a small top group.
largest-contributor share = largest net contribution / total net profit
top-group share = sum of the largest chosen contributors / total net profitThere is no universal safe percentage. A threshold chosen after seeing the result is only a new parameter to optimise. The useful discipline is consistency: define the unit, group size, costs, and handling of losses before comparing systems or test windows.
Losses must stay in the calculation. Reporting gross profit concentration while ignoring gross loss can make a strategy with violent offsetting trades look broader than it is. Use net contributions after the same spread, commission, swap, and slippage assumptions applied to the equity curve.
| Check | A transparent report states | A weak report hides |
|---|---|---|
| Unit | Trade, day, strategy, or regime | What was counted |
| Basis | Net result after stated costs | A frictionless profit total |
| Window | The full period and any validation segment | Selective dates |
| Removal test | What changes when a top contributor is excluded | Reliance on the headline curve |
Why the prop-account path makes this relevant
Prop-style constraints care about the route, not just the finish. A large historical winner may have arrived after a losing spell, during a high-exposure period, or in conditions that are not frequent. If it arrives later in a different sequence, an account can face a very different drawdown path before it receives that profit.
That is why concentration belongs next to losing streaks in prop backtesting, portfolio heat, and time-to-target. These checks do not predict a funded outcome. Together, they ask whether the historical route depends on a narrow set of favourable moments.
The exact rules of any prop firm are its own and can change. The durable point is simpler: a total return cannot tell you whether the account had to wait for one exceptional contributor while carrying the risk of everything before it.
The concentration check has limits
Concentration is a diagnostic, not a verdict. A trend-following approach may rationally earn a large share of profit from a few extended moves. Removing those moves and declaring the strategy broken would confuse its design with a flaw.
It can also overstate fragility when related contributions are split arbitrarily. A single macro event can create many trades, while a daily grouping may show only one entry. That is why the chosen unit needs a reason, not merely a convenient ratio.
The stronger reading is comparative. If a strategy is profitable only because of a tiny group of contributors, test why those contributors occurred, whether their execution assumptions are credible, and whether the rest of the sample supports the same thesis. Then inspect untouched data rather than repairing the report with a nicer ratio.
realbacktesting is a trading-software studio for cTrader that publishes reproducible cBot backtests rather than asking readers to take a curve on trust. Its methodology page sets out the execution and validation approach; the funding page explains the account context. Neither makes a historical result a live promise, and neither should.
Frequently asked
Does high profit concentration invalidate a backtest?
No. High concentration does not invalidate the arithmetic or prove that the strategy has no edge. It shows that the reported result depends heavily on a small part of the historical sample and needs a more careful robustness review.
Should the best trade be removed from a backtest?
Removing the best trade is a sensitivity check, not a new official performance result. It helps show whether one observation dominates the total; it does not establish that the trade was impossible or illegitimate.
Is profit concentration the same as a losing streak?
No. A losing streak measures a run of losses. Profit concentration measures how narrowly gains are distributed among winners, days, strategies, or regimes. Both affect how a prop-style account experiences the path.
Can diversification eliminate concentration?
No. Several strategies can still earn most of their profit from the same market condition or correlated event. Diversification needs to be tested in the realised returns, not assumed from the number of strategy names.
The stubborn takeaway: do not ask only how much a backtest made; ask how many parts of the history had to cooperate for it to make it.