Intuition says a tighter stop means less risk. You lose earlier, you lose less.
We measured it. Intuition is wrong, and not by a little.
The numbers
Same configuration, same period, only the per-basket stop limit changes:
| Limit | Max drawdown |
|---|---|
| No stop | $606 |
| $60 | $224 |
| $20 | $2,331 |
Setting the stop at $60 cut the drawdown by 63 % at no cost in net result. Dropping it to $20 quadrupled it relative to having no stop at all.
Why
The market has a normal fluctuation range — noise that means nothing. If your stop sits below that range, it is not protecting you from adverse moves: it is cutting you in the noise.
And cutting in the noise has a cost that accumulates in a way you cannot see by looking at a single trade:
- You close at a loss a position that was going to recover on its own.
- The logic re-arms and enters again.
- The noise touches you again. You close again.
Each cut is small. The problem is the frequency. A stop inside the noise does not produce one big loss, it produces a constant bleed that adds up to more than the big loss you were trying to avoid.
There is a second effect: because the system goes flat far more often, it enters far more often. You multiply transaction costs and slippage exposure while gaining nothing.
The usable range
We measured the full range. The usable plateau sits between $60 and $100 for this configuration. Inside that band the stop does what it promises. Below it, the sign inverts.
The number itself is not the point — it depends on the instrument, the position size and the timeframe. The point is that a threshold exists, and below it the effect changes sign. A stop is not a knob that runs from "less protection" to "more protection" as you tighten it. It is a U-shaped function.
How to find yours
Do not deduce it: sweep the full range, including values that look absurd at both ends. We tested 20, 40, 60, 80 and 100. Had we only tested 40 and 60, we would have seen a smooth line and concluded that tighter is better up to a point. The disaster sits outside the window a timid sweep explores.
And measure equity drawdown, not balance drawdown. Balance only finds out when you close. Equity drawdown is what margin-calls you out of the market and what makes you shut the system off at three in the morning.
The honest limit of all this
The basket stop was the one change that improved across every regime tested and under both simulation models. It is the most solid piece of the system.
It still does not save a negative expectancy. The same configuration with the stop properly set loses $6,608 over 2024. A good cutout bounds what you lose per episode; it does not turn a losing strategy into a winning one.
From building Cerberus at Tuurt Labs. High-risk instrument: trading leveraged CFDs can cost you your entire capital. This is not investment advice.