Take two traders. Give them the exact same signals — same entries, same exits, same win rate. One ends the year up 40%. The other blows up in March. The only difference between them is how much they risked on each trade. That's not a footnote to strategy. That is the strategy. The indicator everyone obsesses over is the costume; position sizing is the body underneath.
Every edge, no matter how real, comes with losing streaks. Flip a coin that lands heads 55% of the time and you'll still hit runs of five or six tails. If your position size is large enough that a normal losing streak takes you below the point of no return, the math of your edge never gets a chance to play out. You don't lose because you were wrong — you lose because you were too big to survive being temporarily wrong.
You can survive a bad strategy with good sizing. You cannot survive good strategy with bad sizing.
The Kelly criterion answers a precise question: what fraction of your capital maximizes long-run growth, given your edge? For a simple win/loss bet it's:
f* = W − (1 − W) / R
where W is win probability and R is your reward-to-risk ratio. A strategy winning 50% of the time at 2:1 reward:risk gives f* = 0.5 − 0.5/2 = 0.25 — Kelly says risk 25% per trade for maximum growth.
But almost nobody should bet full Kelly. Full Kelly is brutally volatile — it assumes your edge estimate is perfectly accurate, and in trading it never is. Overestimate your win rate by a few points and full Kelly turns from optimal to ruinous. The standard practice is fractional Kelly: a half or quarter of the number. It sacrifices a little growth for a massive reduction in drawdown and in the damage done by a wrong edge estimate.
Most disciplined traders skip raw Kelly and use a simpler, robust rule: risk a fixed small percentage of capital per trade — typically 0.5–2%. The mechanics:
This quietly does what Kelly does — it scales exposure to conviction and volatility — without requiring you to trust a fragile win-rate estimate. And because your risk per trade is fixed and small, no single trade and no normal streak can end you.
"Adaptive" sizing that ramps up after a winning streak feels smart and is dangerous. It maximizes the damage of the loss that always eventually comes — you're largest right before you're wrong. If you size adaptively, cap it hard and floor it, and test it honestly against the regime where it gets punished, not just the bull run where it looks brilliant.
The reason to automate all of this: a bot computes the right size every time, dispassionately, and never "feels lucky." It encodes the survival math once, in the open, where you can read and backtest it.
👉 nexus-bot.pro — build & own a transparent bot where sizing is code you can read, not a gut feeling.
📊 Our live, auditable results: nexus-bot.pro/proof/rvv/