The short version
The Kelly criterion estimates how much of a bankroll to put at risk when you believe you have an advantage and know the odds being offered.
The formula only produces a positive result when your estimated win probability is above the odds’ break-even probability. If the edge is uncertain, the output can look more precise than the underlying assumption deserves.
Start with break-even probability
American odds already contain a break-even probability. At +100, the break-even point is 50%. At −110, it is about 52.38%.
Your estimate has to clear that number before Kelly sees a positive edge. A 55% estimate at +100 is 5 percentage points above break-even. A 55% estimate at −110 is only about 2.62 points above it.
The full-Kelly formula
In the formula, b is the net profit on a one-unit win, p is your probability of winning, and q is your probability of losing. The reference is f = (bp − q) ÷ b.
For +100, b is 1.00. With p = 0.55 and q = 0.45, the result is 0.10, or 10% of the bankroll. That is a mathematical output—not a guarantee, a recommendation, or proof that the estimate is correct.
Expected value answers a related question: what average net result do the price and probability imply? Read expected value, explained or use the expected value calculator to explore that calculation.
When Kelly applies
Kelly is designed for situations where you have a repeated wager, a known price, and a credible estimate that your true probability is better than the price implies. FadeIt’s calculator is built around betting prices and binary outcomes.
Ordinary casino games with a house edge are different: the player does not start with a positive edge, so Kelly should not be used to make a normal casino session look mathematically favorable. Use the house edge calculator to understand the game’s long-run cost and the bankroll calculator to plan controlled exposure.
Why people use fractional Kelly
Full Kelly is highly sensitive to the probability estimate. If the estimate is too optimistic, full Kelly can make the error expensive. Fractional Kelly uses a portion of the full result, such as one-half or one-quarter.
A fraction reduces the size of the reference stake and the impact of estimation error. It does not remove variance, create an edge, or make a losing result impossible.
The estimate is the hard part
The calculator can do the formula exactly. It cannot validate your model, sample size, injury information, market assumptions, or whether the line will still be available when you play.
That is why a Kelly result should be treated as a way to audit your assumptions. If you cannot explain where the probability estimate came from, the percentage on the screen should not feel authoritative.
What the calculator can and cannot do
FadeIt’s calculator compares your probability estimate with implied probability and shows full or fractional Kelly as a reference. When your estimate is at or below break-even, it shows no positive Kelly stake.
It does not know your finances, your tolerance for losses, or whether gambling fits your life. Only use money you can afford to lose, and stop if the plan starts turning into pressure to recover losses.