The Lab · The math · updated 2026-10-22
How often do favorites win? Four sports, measured
Across 20,360 games, favorites won 58.7% in baseball, 62.2% in hockey, 68.0% in basketball and 67.1% in football — each within a point and a half of what their prices said. Why the sports differ, and why betting the favorite still loses.
Favorites win about 59% of baseball games, 62% of hockey games and about two in three in basketball and football — almost exactly as often as their prices say. Because the price already knows, betting every favorite loses about the fee.
Record figures marked as today's come live from the public record, refreshed as games are graded — as of Oct 30, 2026. Numbers we published on earlier dates stay as written, dated, so you can check them against today's.
How often does the favorite win? It's the first question anyone asks about betting, and the answer depends almost entirely on the sport.
We first answered it in July with a number from our own board — 54.7%, across 311 baseball games — under the title "Why baseball favorites only win 54%." That number was honest and far too small to lean on. This is the measured answer, and further down, what happened to the old one.
Four sports, measured
| Sport | Games | Favorite won | Its price said | Games under 60% |
|---|---|---|---|---|
| Baseball (MLB) | 9,270 | 58.7% | 58.5% | 64% |
| Hockey (NHL) | 5,126 | 62.2% | 60.7% | 53% |
| Basketball (NBA) | 4,887 | 68.0% | 67.5% | 30% |
| Football (NFL) | 1,077 | 67.1% | 65.8% | 34% |
Every game of four seasons, 2021 through 2024–25, priced at least an hour before the start by the consensus of every sportsbook quoting it, with the fee removed. "Its price said" is the average of those fee-free chances. (The most recent finished season is held back for testing, and isn't in these numbers.)
Two things are true in every row. The favorite wins more than half the time — about three games in five in baseball, two in three in basketball and football. And it wins almost exactly as often as its price said: within 1.6 points in all four sports.
Every band lands where it says
Sorting baseball games by how confident the market was:
| Market gave the favorite | Games | Its price said | Favorite won |
|---|---|---|---|
| 50–55% | 3,007 | 52.6% | 54.0% |
| 55–60% | 2,923 | 57.4% | 56.3% |
| 60–65% | 1,998 | 62.3% | 61.4% |
| 65–70% | 975 | 67.2% | 70.8% |
| 70–75% | 336 | 71.8% | 69.6% |
| 75% and up | 31 | 76.2% | 64.5% |
That's what calibration looks like: when the market said 55–60%, the favorite won 56.3%; when it said 60–65%, 61.4%. The widest miss with a real sample is the 65–70% band, 3.6 points over its price — and its neighbors on both sides missed the other way, which is what noise looks like and a real bias usually doesn't. The top band holds 31 games; that few can miss by 15 points on luck alone, so we print it rather than read it.
The table also explains the number. Almost two baseball games in three sit under 60%. The average baseball favorite wins 58.7% because the average baseball "favorite" is a team the market gives a coin flip and a nudge.
Why the sports differ
Same sportsbooks, same sharp bettors, very different rates: 58.7% in baseball, 62.2% in hockey, 67.1% in football, 68.0% in basketball. Baseball isn't priced worse — it's structurally closer to even, for reasons built into the sport:
- The best player is capped. An ace throws about six innings, then hands the game to a bullpen much closer to league average. A dominant pitcher can't own a game the way a dominant basketball player owns forty minutes.
- The talent band is narrow. Excellent baseball teams win around 60% of their games; bad ones around 40%. In basketball that spread runs from 80% down to 20%.
- Scoring is low and lumpy. One bloop single or one error decides a 3–2 game. Fewer scoring events leave less room for the better team to prove it. Hockey shares this — and a hot goaltender can steal a game the way a hot pitcher can.
- Even the at-bat is nearly a coin flip. An elite hitter succeeds around 30% of the time, a poor one around 20%. Ten points of separation, compounding very slowly over nine innings.
The clinching evidence is the schedule itself. Baseball plays 162 games precisely because one game tells you so little. Football settles its standings in 17. If a single baseball game reliably revealed the better team, the season wouldn't need to be ten times longer.
The trap: accurate is not profitable
This is where "favorites win most of the time" gets dangerous, because it's easy to read as "so bet favorites."
Favorites cost more. A favorite's price already demands its win rate — plus the fee. Here's every favorite in those four seasons, bet one unit each at the typical sportsbook's price:
| Sport | Favorites won | The price needed | Betting every favorite |
|---|---|---|---|
| Baseball | 58.7% | 60.6% | −3.2% |
| Hockey | 62.2% | 63.1% | −1.3% |
| Basketball | 68.0% | 70.3% | −3.0% |
| Football | 67.1% | 68.5% | −1.8% |
Favorites won more than half the time in every sport, and lost money in every sport. That isn't a cold streak — it's the arithmetic of the bet, every time you make it. Taking the single best price any book in our data offered shrinks the result to between −1.0% and +0.7%: shopping recovers most of the fee and leaves no edge behind.
Which is why a model can be right and still lose money. Being accurate gets you to the market's number. Profit requires being more accurate than the market, by more than the fee it charges to play. Those are very different achievements, and almost every betting pitch you'll ever read quietly swaps one for the other.
What happened to 54.7%
This page first went up on July 28. It reported our board's record at the time: 311 graded baseball games, favorites winning 54.7% against a claimed 55.9%. We're leaving that here, dated, because it's what we published.
It was honest, and it was small. 311 games pin a win rate down to about ±5.6 points (two standard errors), so 54.7% was always consistent with the 58.7% that four full seasons show. The board was also claiming less for its favorites then — 55.9% on average, against 58.5% across those four seasons — from a few weeks of one season, with prices built differently in those early weeks. We can't fully separate those causes, and we won't pretend to.
What we can do is show the board as it has grown. Every figure we've published about it, dated, beside today's:
| Published | Games | Favorite won | Market said | Gap (pts) |
|---|---|---|---|---|
| Jul 27, 2026 | 266 | 54.9% | 55.8% | −0.9 |
| Jul 28, 2026 | 311 | 54.7% | 55.9% | −1.2 |
| Aug 1, 2026 | 332 | 54.5% | 55.9% | −1.4 |
| Aug 15, 2026 | 519 | 55.1% | 56.6% | −1.5 |
| Oct 21, 2026 | 1,439 | 58.9% | 58.2% | +0.7 |
| Today live | 1,595 | 59.7% | 58.6% | +1.2 |
The board today, sport by sport — every game it has graded in public, five sports now:
| Sport | Games | Record | Favorite won | Market said | Gap (pts) |
|---|---|---|---|---|---|
| MLB | 1,149 | 662–487 | 57.6% | 57.1% | +0.5 |
| NHL | 213 | 136–77 | 63.8% | 58.2% | +5.7 |
| NBA | 71 | 49–22 | 69.0% | 65.1% | +3.9 |
| NFL | 107 | 67–40 | 62.6% | 64.7% | −2.0 |
| NCAAF | 55 | 39–16 | 70.9% | 71.3% | −0.4 |
And by how confident the market was, the same bands as the baseball table above:
| Market gave | Games | Price said | Favorite won |
|---|---|---|---|
| 50–55% | 595 | 52.3% | 52.4% |
| 55–60% | 467 | 57.3% | 59.5% |
| 60–65% | 274 | 62.2% | 63.5% |
| 65–70% | 130 | 66.8% | 67.7% |
| 70–75% | 73 | 72.0% | 78.1% |
| 75–80% | 27 | 77.2% | 77.8% |
| 80–85% | 12 | 82.8% | 83.3% |
| 85–90% | 14 | 87.1% | 71.4% |
| 90–95% | 3 | 90.9% | 100.0% |
For the numbers people: the bands in the other three sports
Same method: four seasons, consensus fee-free price at least an hour before the start, bands by the favorite's chance.
Hockey
| Market gave the favorite | Games | Its price said | Favorite won |
|---|---|---|---|
| 50–55% | 1,260 | 52.6% | 54.4% |
| 55–60% | 1,464 | 57.4% | 58.9% |
| 60–65% | 1,050 | 62.4% | 63.5% |
| 65–70% | 728 | 67.3% | 67.7% |
| 70–75% | 422 | 72.1% | 75.1% |
| 75% and up | 202 | 77.6% | 81.7% |
Basketball
| Market gave the favorite | Games | Its price said | Favorite won |
|---|---|---|---|
| 50–55% | 674 | 52.7% | 56.4% |
| 55–60% | 815 | 57.5% | 57.8% |
| 60–65% | 782 | 62.5% | 61.6% |
| 65–70% | 672 | 67.3% | 67.7% |
| 70–75% | 669 | 72.4% | 73.4% |
| 75% and up | 1,275 | 82.3% | 82.0% |
Football
| Market gave the favorite | Games | Its price said | Favorite won |
|---|---|---|---|
| 50–55% | 135 | 52.9% | 53.3% |
| 55–60% | 229 | 57.5% | 59.4% |
| 60–65% | 219 | 62.5% | 67.1% |
| 65–70% | 134 | 67.6% | 68.7% |
| 70–75% | 149 | 72.4% | 71.1% |
| 75% and up | 211 | 80.9% | 80.6% |
The standard error of a band's win rate is √(p(1−p)/n): about 1.4 points for a 1,200-game band at 55%, about 4 points for a 135-game one. Overall, the gaps between "won" and "said" are +0.15 points in baseball (z +0.30), +1.56 in hockey (z +2.31), +0.54 in basketball (z +0.81) and +1.31 in football (z +0.91).
"The price needed" in the trap table is the average break-even of the median book's price on the favorite; "betting every favorite" is the flat one-unit return at that price. Its own standard error runs from 0.85 points (baseball) to 2.2 (football).
What this doesn't tell you
Hockey's gap is the one worth watching. NHL favorites won 62.2% against a price of 60.7% — about 2.3 standard errors, the only sport past two. Favorites priced a shade short is a pattern with a name, the favorite–longshot bias. But notice it still didn't pay: betting every NHL favorite at a typical book lost 1.3%. A lean in the price has to beat the fee before it's worth anything. Our two-week Frankenstein test asks the direct version — every board probability pushed 10% further from a coin flip — with its forecast written down before the first game.
The four seasons are history. They describe how the market behaved, not a promise about tonight. That's why the live tables above exist: the board grades every game in public, and you can watch whether the pattern holds.
None of this says favorites are bad bets or underdogs are good ones. It says the price already knows. Favorite or underdog is the wrong question; the only one worth asking about a bet is whether its price is wrong, and the answer is usually no.
Model output is informational and entertainment content, not betting or financial advice. If you bet, bet what you can afford to lose.
Log what you actually bet — units and dollars — and see it graded beside a model that publishes every call it makes, wins and losses alike. Free, no card.
Start a free tracker