How to Read a Win Probability

A number like “65%” looks precise, but it’s easy to misread. Here’s what a win probability actually tells you — and what it doesn’t.

It’s a frequency, not a promise

A win probability is an estimate of how often a result would happen if the same matchup were played many times. If we list a team at 65%, the honest reading is: “in a long run of games like this one, we’d expect this side to win about 65 out of 100.” That also means the other team wins about 35 out of 100 — which is a lot. A 65% favorite losing isn’t the model being “wrong”; it’s the 35% showing up, exactly as expected.

Favorites lose constantly — and that’s normal

Upsets aren’t flukes; they’re built into the percentages. A team favored at 70% will still lose roughly three times out of ten. Over a full season of slates, you should expect a steady stream of “wrong” favorites — if every favorite won, the probabilities would be broken, not accurate. The right way to judge any probability source is over a large sample, not on a single game.

What the confidence level means

On MatchupLens, each pick carries a High / Medium / Low confidence tag based on how far the favorite sits from a coin flip:

Confidence describes how lopsided the inputs are — nothow accurate the call will turn out to be. A “High” game can still go the other way.

Why we show “market agreement”

When a betting line is available, we note whether our pick agrees with the market. Agreement is a sanity check — two independent reads landing on the same side. A disagreement isn’t a signal to bet against the market; it usually just means the inputs are close, and it’s a cue to dig into the matchup yourself.

Three common mistakes

Win probabilities on MatchupLens are editorial analysis for informational and entertainment purposes only — not betting advice, and not a guarantee of any outcome. If you choose to wager, do so responsibly and only where legal. 21+. Gambling problem? Call 1-800-GAMBLER. See the full methodology →