Distribution figures below are descriptive of how the Shishin backtest’s 353 gated trades clustered across the five-year window. They are not a projection of any forward trade outcome. Live execution will produce a different distribution for reasons that include slippage, fills, and the normal divergence of sample from model.
The narrative around systematic trading is the ten-bagger. The reality is the boring middle. Most trades are neither triumphant winners nor blow-ups; they settle in a quiet range either side of zero, hold a few days, and close. The curve is built in that range, not in the tails.
The shape of the distribution
Across the full five-year backtest, the four-engine stack took 353 trades. The blended hit rate sits below half, roughly a coin flip. The headline number that gets quoted is the CAGR; the number that actually describes the experience of running the system is the shape of the trade-level distribution underneath that CAGR.
It is not a normal distribution. It is right-skewed, stops cap the loss side into a tight cluster of small, similar-sized losses, while the win side runs out into a long, fat tail. Most trades sit in a narrow central band. The boring middle , roughly the central eighty percent of all trades by outcome, produces modest returns either side of zero. The largest losers and the largest winners live out in the tails, each making up a small fraction of the trade count but absorbing or contributing a disproportionate share of the pnl.
Why this matters more than people think
The intuition most readers bring to a quant strategy is: the system finds rare exceptional setups and sizes into them. The reality is the opposite. The system takes a large number of setups that look ordinary, and the math plays out across the population. Any given trade is more likely to land in the middle than in either tail. The operator’s job is not to predict which trade will be the outlier, that is unknowable in advance.
The operator’s job is to keep taking setups that have positive expected value, sized consistently, exited consistently, regardless of whether each individual one feels exciting. The boring middle is the price of admission. You pay it on every trade. The tails, the few that compound your year, arrive without warning, scattered through the population, and you cannot afford to miss them by becoming selective about which setups you participate in.
What changes in practice
Three operational habits fall out of taking the distribution seriously.
- You stop chasing. If most trades land in the boring middle, the trade that looks “the most obvious” in real time is statistically no more likely to be the outlier than any other. The urge to size up because a setup feels strong is the urge to make a categorical guess about which trade will be in the tail. The data does not support that urge.
- You let small wins and small losses accumulate. Every trade in the boring middle is doing exactly what the system needs it to do. The curve is the integral of all of them, not the sum of the tail outliers minus everything else.
- You design for the floor, not the ceiling. Stops, position size caps, and re-entry rules are tuned so that the worst trades stay small enough to be absorbed by the middle’s drift. If a single loser can swallow a month of boring-middle gains, the tail risk is poorly contained. We size and stop so that any single bad outcome is recoverable from ordinary forward operation.
What the tail outliers actually look like
We will not publish the per-trade pnl distribution row by row, the specific thresholds that separate “middle” from “tail” are part of the engine’s edge, and writing them down hands downstream observers a parameter set they can stress-test against ours. What we will say is that the biggest winners are usually held longer than the average, exit at a moving-average-based trail rather than at a stop, and almost without exception involve the position being topped up during the run as the composite score continued to climb. The biggest losers, conversely, are usually short holds, the stop hit early, the position closed, the system moved on.
That asymmetry, long winners, short losers, is the structural shape of any expectancy-positive trend-following or breakout strategy. It is not unique to Shishin. What is more specific to our work is the discipline of letting the long winners run without intervention while the rest of the population stays in the boring middle, quietly producing the floor.
The institutional point
Quant marketing rewards content about the spectacular trade. The actual work of running a system is keeping the spectacular trade from being the only thing the system is organised around. Most days are unremarkable. Most trades are unremarkable. The discipline is to value that. The operator who learns to find the middle uninteresting is the operator most likely to over-fit the system to the tails, and that is the surest way to make the long-run curve worse, not better.
The boring middle is the work. Everything else is the consequence of staying in it.