Position sizing: the variable that decides whether you survive.

Two traders run the identical strategy with the identical signals. One risks 1% per trade and compounds steadily. The other risks 10% and is finished inside a normal losing streak. Same edge, same entries, opposite outcomes — the only variable was size.

This is why sizing deserves more attention than entry logic and almost never gets it. Entries are interesting. Sizing is arithmetic, and arithmetic is what actually ends accounts.

Trading a fixed number of shares or lots means your risk changes every time volatility does, without you deciding anything. The same 100 shares is a small risk in a quiet market and an enormous one in a violent week.

Fixed fractional sizing inverts that: you decide the risk, and position size follows from it. Risk per trade stays constant while quantity adapts.

A strategy that wins 50% of the time will, across a few hundred trades, produce a run of eight or nine consecutive losses. Not as a disaster scenario — as ordinary variance. At 40% win rate, streaks of twelve are unremarkable.

Size for that, not for the average case. The question is never "what happens if this trade loses?" It is "what does my account look like after the worst run this strategy will normally produce?"

• >1% risk, 10 straight losses: down about 10%. Recoverable, uncomfortable.
• >2% risk, 10 straight losses: down about 18%. Painful, still survivable.
• >5% risk, 10 straight losses: down about 40%. You now need a 67% gain to break even.
• >10% risk, 10 straight losses: down about 65%. Effectively over.

The Kelly criterion gives the bet size that maximises long-run growth given a known edge. It is mathematically correct and almost nobody should trade it directly.

Two reasons. It assumes you know your edge precisely, and you do not — you have an estimate from a backtest that is probably optimistic. And full Kelly produces drawdowns most people cannot psychologically hold. Overestimate your edge slightly and Kelly tips from optimal to ruinous quickly.

Half Kelly or quarter Kelly is the practical compromise: most of the growth, a fraction of the volatility, and far more tolerance for having overestimated yourself.

A refinement worth the effort: scale by how much the instrument actually moves. Sizing off a volatility measure such as average true range means a quiet stock and a violent one carry comparable risk, rather than the violent one silently dominating your portfolio.

This matters most when trading several instruments at once. Without it, whatever is moving most becomes your largest exposure by accident.

Five positions at 1% each is not 5% of risk if all five are the same trade wearing different tickers. Long four tech stocks and a tech ETF is one position in five pieces, and it will behave like one on the day it goes wrong.

Cap exposure by theme, sector or underlying driver, not only by individual position. The account-level number is the one that matters.

Risk 0.5–1% per trade while you are still learning what a strategy does live. Cap total open risk at something like 5% of the account. Set a daily loss limit that stops trading for the day, and a total drawdown limit that stops it entirely pending review.

None of this is exotic, and that is the point. Sizing does not need to be clever. It needs to be small enough that ordinary bad luck is survivable, and automatic enough that you cannot talk yourself out of it.