Most traders ask the right question ("is this signal an edge?") but skip the next one: how much do I bet?
That's where the Kelly Criterion comes in. It's the math for sizing positions when you have a known edge.
The formula:
f* = (bp - q) / b
Where:
โข f* = fraction of bankroll to bet
โข b = net odds received on the bet (decimal payout - 1)
โข p = probability of winning
โข q = probability of losing (1 - p)
Applied to a NO signal:
If YES is priced at 35ยข, NO costs 65ยข and pays $1 if it resolves NO.
โข Net odds b = (1 / 0.65) - 1 = 0.538
โข If our model says true NO probability is 75%: p = 0.75, q = 0.25
โข f* = (0.538 ร 0.75 - 0.25) / 0.538 = 0.535 โ 53.5% of bankroll
That's full Kelly. It's mathematically optimal for long-run growth โ but it's volatile. A single loss takes 50%+ of your bankroll.
Most serious traders use FRACTIONAL Kelly โ typically half-Kelly (26.7% in the example above) or quarter-Kelly (13.4%). You give up some growth in exchange for much smaller drawdowns.
Why this matters for EdgeFinder signals:
A 66.5% win rate is real edge. But edge without proper sizing is how good traders blow up. Kelly tells you the ceiling. Fractional Kelly keeps you in the game long enough to compound.
Rule of thumb: Never bet more than half-Kelly on any single signal. If your bankroll can't survive a 5-loss streak, you're sized too big.
Learn more in The EdgeFinder Foundation course โ Classroom tab above.