Size the bet like the math says.

The Kelly formula is exact, for the inputs you give it. Feed it a 58% win probability when the truth is 53% and it will confidently size you into bets you shouldn't make at all. That asymmetry is why every serious bettor runs a fraction: half-Kelly keeps ~75% of the theoretical growth at half the variance, and quarter-Kelly, Whet's default, trades a little more growth for an estimate-error cushion that survives being wrong.

If you don't have a defensible win probability, the stake isn't the problem. Start with the de-vigged market consensus instead of a hunch.

What is the Kelly criterion?

A formula for bet sizing that maximizes long-run bankroll growth: f = (bp − q) ÷ b, where b is the decimal odds minus 1, p is your win probability, and q is 1 − p. It stakes more when your edge is bigger and nothing when you have no edge.

Why ¼-Kelly instead of full Kelly?

Full Kelly is only optimal if your probability estimate is exactly right. Overbetting an overestimated edge is far more damaging than underbetting a real one. Fractional Kelly gives up a little growth for a lot of protection against your own estimate being wrong. Whet sizes every suggestion at ¼-Kelly for exactly that reason.

Where do I get the win probability?

That's the hard part. The formula is only as good as the probability you feed it. De-vigging the market (see the no-vig calculator) is the starting point; Whet's board gives you one for every pick and sizes it with this same math.

The calculator says 'no stake.' Why?

At the odds you entered, the win probability you typed doesn't clear the breakeven point. The bet is -EV, and Kelly's answer to a -EV bet is to not place it. No stake IS the output.