Published 2026-10-14 · Updated 2026-10-25
Risk Management Is Sizing, Not Stop-Losses
Most "risk management" advice is about exits. The bigger lever is how much you put in each position — because a strategy survives or dies on position size long before it depends on the exit rule.
On this page
The common mistake
Ask ten people about risk management and most will describe stop-losses: where to exit, how far, trailing or fixed.
That is one lever. It is not the biggest one.
The lever that dominates
Two strategies with identical entries and identical exits can end very differently — because of how much each put in.
| Sizing | Outcome |
|---|---|
| Fixed fraction | Bounded loss per trade; survivable by construction |
| All-in | One bad trade ends the account |
| Doubling after losses | A classic path to ruin — the losses compound faster than the wins |
| Volatility-scaled | Sizes down when things are choppy |
A stop-loss limits the loss on a trade. Position sizing limits the loss on the account. Those are different problems, and the second one is what ends strategies.
Why exits get all the attention
Stop-losses are visible and intuitive: sell when it drops 5%. Sizing is a quieter decision — and unlike a stop, it is often not stated at all.
In a backtest, sizing is usually implicit. The number of units is chosen somewhere and never reported. Which means:
| If sizing is unstated | Then |
|---|---|
| The result is not reproducible | Someone else would size differently |
| The drawdown is not interpretable | A 40% drawdown at one size is 90% at another |
| Risk is not measured | It is only shown at one arbitrary point |
A backtest that reports a return without its sizing has not reported a strategy.
Sizing is also a check on the result
Once sizing is explicit, it becomes a test the result has to pass:
| Ask | Why it matters |
|---|---|
| What is the rule? | Fixed fraction, volatility-scaled, something else |
| What is the worst drawdown under it? | Not the return — the drawdown you would have had to sit through |
| Does it survive the bad case? | Sizing is about not being wiped out, not about being optimal |
The purpose of sizing is not to maximise the return. It is to make sure there is still an account to trade tomorrow.
Sizing interacts with everything else
Position size is not independent of the rest of the system:
- With turnover — bigger positions cost more to trade
- With liquidity — a size that is fine in calm markets may not fill in stress
- With correlation — ten correlated positions sized independently are one big position
- With capacity — the strategy’s own size becomes the market impact
This is why sizing cannot be chosen after the fact. It changes what the strategy is.
The practical version
State the rule, state the worst case, and check that the result survives it.
# Every input to the sizing decision must be reproducible too.
curl "https://api.ashareapi.com/v1/kline?code=sh600519&period=day&count=250"Then answer, in writing: what is the maximum loss per trade, and what is the maximum drawdown if the rule is followed exactly?
If you cannot state those two numbers, the strategy does not have a risk limit. It has a hope.
Next
- Backtest a portfolio, not just a stock → how sizing meets correlation
- Position sizing → the glossary entry