The formula
If f* comes out zero or negative, the inputs show no edge and the "right" bet is nothing at all.
A worked example
Your journal shows a 45% win rate, winners averaging 2R and losers 1R, so b = 2.
Even a quarter Kelly is several times the 0.5% to 1% most traders risk. Why the gap? Kelly assumes your 45% and 2R are exact. If the true win rate is 40%, full Kelly drops to 10%, and betting 17.5% on a 10% edge produces brutal drawdowns. Kelly is best read as a ceiling and as a measure of how strong the edge is, not as a size to trade.
Common mistakes
- Using full Kelly. Full Kelly routinely produces drawdowns of 50% or more along the way, which ends any prop account.
- Feeding it a small sample. Twenty trades cannot pin down a win rate closely enough to size from.
- Ignoring the drawdown rules you trade under. A prop firm's daily and max loss limits cap your real size far below Kelly.
How the RB journal tracks it
Pro Metrics shows the Kelly Criterion from your own trades using Win% − (Loss% ÷ (Avg Win ÷ Avg Loss)). It appears from 10 closed trades, is marked "low sample" until 20 or more, and the journal shows the ½-Kelly figure alongside, with a note that full Kelly is far too aggressive to use as is.
See your own Kelly Criterion from real trades
Log trades by hand, import a file, or live-sync MT4, MT5 and cTrader through the RBSync EA. The numbers on this page then come from your own history instead of examples.
Start free, no card