Why Baseball Favorites Lose So Often: The Nature of the 162-Game Season
The best baseball teams win about two games in three and the worst win about one in three, so favorites lose constantly. Worked examples show why, how odds convert to probability and what a series adds.
The best team in baseball wins roughly 65 percent of its games over a season. The worst team wins roughly 35 percent. That is the whole range, and it is narrow. When the 2001 Seattle Mariners won 116 games and lost 46, one of the great regular seasons in history, they still lost 46 times. When the 2019 Detroit Tigers went 47-114, one of the worst records of the decade, they still won 47 times. Every day of the season, a team that looks hopeless beats a team that looks unbeatable, and nobody should be surprised.
This article explains why. It uses two complete seasons from the historical record, 2001 and 2019, to show how tightly baseball results cluster, then works through the arithmetic of single games, full seasons and playoff series. It also explains how betting odds convert to probabilities, because the numbers in an odds listing only make sense once you know how rarely even the best team wins.
A responsible-gambling note before we begin. This is an educational article about probability. It makes no picks, offers no tips on any game and promises no profit; betting on baseball carries a real risk of loss, and it is for adults of legal age in places where it is legal. If gambling is causing you or someone close to you a problem, the national helpline in the United States is 1-800-GAMBLER, and the National Council on Problem Gambling has resources.
The range of baseball: two seasons side by side
Look at the standings from any full season and the same shape appears. The top teams cluster around .600 to .650, the middle is near .500 and the bottom clusters around .350 to .400. Here are the five best and worst records from the two seasons in our data, with winning percentages calculated from the standings.
| Season | Best records | Worst records |
|---|---|---|
| 2001 | Seattle 116-46 (.716), Oakland 102-60 (.630), New York Yankees 95-65 (.594), Houston and St. Louis 93-69 (.574) | Pittsburgh 62-100 (.383), Tampa Bay 62-100 (.383), Baltimore 63-98 (.391), Kansas City 65-97 (.401) |
| 2019 | Houston 107-55 (.660), Los Angeles Dodgers 106-56 (.654), New York Yankees 103-59 (.636), Minnesota 101-61 (.623) | Detroit 47-114 (.292), Baltimore 54-108 (.333), Miami 57-105 (.352), Kansas City 59-103 (.364) |
Two things stand out. First, a .716 team (2001 Seattle) is an outlier; apart from that club, the best record in either year was .660 or lower. Second, even the worst teams win about a third of their games. In 2019 the league scored 4.83 runs per team per game, and in 2001 it was 4.78. In a sport where each side scores about five runs and one swing or one error can move the result, the distance between a great team and a terrible one is smaller than in most sports.
The result is a blunt rule of thumb. When a very good team plays an average team on a neutral field, it wins about 60 to 65 percent of the time, which means the average team wins 35 to 40 percent.
Combining two teams: the log5 method
How often should a .650 team beat a .400 team? The standard tool is a formula named log5, which Bill James introduced in the early 1980s. If team A's winning percentage is a and team B's is b, the chance that A beats B in a given game is:
a(1 - b) / [a(1 - b) + b(1 - a)]
Plug in some numbers.
| Team A | Team B | Chance A wins |
|---|---|---|
| .650 | .500 | .650 |
| .650 | .400 | .736 |
| .650 | .350 | .775 |
| .600 | .450 | .647 |
| .550 | .450 | .599 |
| .650 | .550 | .603 |
Use the actual seasons. If the 2019 Astros (.660) played the 2019 Tigers (.292), the formula gives the Astros about an 83 percent chance, so the Tigers win roughly one game in six. That is the extreme case: the best team in the data against the worst. For nearly every other pairing, the favorite's edge is much smaller, in the 55 to 65 percent range.
The formula is a starting point, not a prediction. It ignores who is pitching, the ballpark and the lineup that day. It also uses full-season records as if every team were the same team in every game. Those adjustments move the number a few points in either direction, but they do not change the shape of the answer: the favorite in a baseball game is rarely a heavy favorite.
Why baseball is so noisy
Several features of the game keep the better team from winning more often.
- The best hitters fail most of the time. Ichiro Suzuki won the 2001 American League batting title at .350. He still failed to get a hit in about 65 of every 100 at-bats. If the league's best hitter fails two times in three, no lineup can guarantee anything on a given night.
- Low scoring. With about 4.8 runs per team per game, a single inning or a single home run decides many games. In higher-scoring sports, results are driven by many more scoring events, so the better team's advantage has more chances to show.
- The pitching matchup changes everything. A team's season record averages its best and worst starting pitchers. A weak team with its best starter on the mound is a different opponent from the same team with its worst, and bettors and oddsmakers know this.
- Luck on balls in play. A hard-hit ball can go right at a fielder. A weak grounder can find a hole. Over 162 games these things average out; over one game they decide the result.
- No clock and no way to run out the lead. A team ahead cannot kill time, and a team behind always gets its last three outs. The structure favors close games.
None of this says baseball is random. It says the skill difference between teams is small compared with the amount of luck in any one game. That is why you often see a heavy favorite, a team with the better pitcher, the better lineup and the home park, lose to a team with a worse record.
From one game to a full season
The luck that dominates a single game mostly cancels out across 162. But it does not disappear. Here is how much it matters.
If a team's true chance of winning any game is 50 percent, the number of wins it records over 162 games has a standard deviation of about 6.4 wins (the square root of 162 times 0.25). Roughly two-thirds of the time, such a team will finish between 75 and 87 wins; roughly 95 percent of the time, between 68 and 94. A team of exactly average ability will win 95 or more games about once in 60 seasons, which sounds rare but across 30 teams over many years it does happen.
The same arithmetic works for the other end. A team with a true winning chance of 55 percent, a good team, has about a 20 percent chance of winning 95 games and about a 9 percent chance of winning 80 or fewer. A true .600 team has less than a 1 percent chance of finishing at 80 wins or below.
What this means in practice is that a team's record is an estimate of its quality, and a noisy one. A team that starts 12-8 is not demonstrably better than average: with a 20-game sample, the standard deviation of a winning percentage is about .112, so a 12-8 record (.600) is within one standard deviation of a .500 team. For this reason the best predictors of future results are not the standings but a combination of run scoring, run prevention and the roster.
Run differential: the better measure of a team
Because records are noisy, analysts look at runs scored and allowed. The simplest version is the Pythagorean expectation, which estimates winning percentage as:
runs scored squared / (runs scored squared + runs allowed squared)
A team that scores 800 runs and allows 700 has an expected winning percentage of 640,000 divided by 1,130,000, about .566, or roughly 92 wins in a 162-game season. If the same team wins 97 games, it is likely to have had some good luck in close games. If it wins 87, it probably had bad luck. The formula is a rule of thumb, not a law, but it has a long record of predicting next-season performance better than the win-loss record does.
Our guide to magic numbers shows how wins and losses turn into the clinching arithmetic at the end of a season, which is a different question from how good a team really is.
From a game to a series
The postseason does not fix this; it makes the effect larger. A short series adds a few more games but not enough to separate teams that are close. Here is the chance that a team with a given single-game probability wins a series against an opponent, assuming that chance is the same in every game.
| Single-game chance | One game | Best of 3 | Best of 5 | Best of 7 |
|---|---|---|---|---|
| 55% | 55.0% | 57.5% | 59.3% | 60.8% |
| 60% | 60.0% | 64.8% | 68.3% | 71.0% |
| 65% | 65.0% | 71.8% | 76.5% | 80.0% |
| 70% | 70.0% | 78.4% | 83.7% | 87.4% |
A team that is a 65 percent favorite in every game, which is an enormous edge against any playoff opponent, still loses a best-of-seven series one time in five. In practice, playoff opponents are good teams, and the single-game edge is closer to 55 percent, which gives the favorite about 61 percent in a best-of-seven. That is better than a coin flip but not by much.
The two seasons in the data show the pattern. In 2001 the 116-win Mariners lost the American League Championship Series to the New York Yankees, four games to one, and the Yankees (95-65) then lost the World Series to Arizona (92-70), four games to three. In 2019 the Dodgers (106-56) lost the division series to the Washington Nationals, three games to two, and the Nationals, who had gone 93-69 and reached the postseason as a wild card, won the World Series by beating Houston (107-55), four games to three. The Yankees (103-59) lost to the Astros in the American League Championship Series, four games to two. In both years the best regular-season record did not win the title. For how those series are priced, see betting on the postseason: series prices and game prices.
What the odds are telling you
Betting odds are a way of stating these same probabilities with a price attached. American odds come in two forms. A negative number shows how much you must risk to win 100 dollars. A positive number shows how much you win on a 100-dollar stake. To convert to the implied probability:
- For a negative price, divide the number (without the minus sign) by that number plus 100. A price of -150 implies 150 / 250 = 60 percent. A price of -200 implies 66.7 percent. A price of -250 implies 71.4 percent.
- For a positive price, divide 100 by the price plus 100. A price of +150 implies 100 / 250 = 40 percent. A price of +200 implies 33.3 percent.
A listing of -150 on one team and +130 on the other shows why the numbers do not add to 100. The favorite's 60 percent plus the underdog's 43.5 percent comes to 103.5 percent. The extra 3.5 points are the sportsbook's margin, often called the vig or the juice. You cannot be paid at fair odds on both sides; the margin is how the book makes money. Our fuller guide, converting baseball odds to implied probability, works through more examples, and how to compare odds and shop for lines explains why the same game can be priced differently at different books.
Compare the implied numbers to the baseball facts above. A price of -200 implies about 67 percent, roughly what the very best teams of a season achieve against average opponents. Prices of -300 or longer, which imply 75 percent or more, can appear in baseball, for example when a dominant starting pitcher faces a weak team, but the sport's noise makes them hard to justify. Even a heavy favorite loses about one game in four.
What not to conclude
It is easy to draw bad lessons from this.
- "A favorite that loses was a bad bet." Not necessarily. A 65 percent favorite loses 35 times in 100. One loss tells you almost nothing.
- "An underdog that wins was a good bet." The same logic. A bet is judged by whether the price matched the true chance, not by one result. A single outcome cannot settle that, and nothing here is a recommendation to bet on anything.
- "A hot team will keep winning." Short streaks are within the range of normal luck. A team that wins eight in a row is not necessarily better than it was.
- "The long season means the standings tell the truth." The standings tell you roughly who is good. A 162-game record still carries about 6 wins of luck either way.
Betting, on baseball or anything else, is entertainment at best and a financial risk always. Never wager money you cannot afford to lose, and if you feel you cannot stop, call 1-800-GAMBLER. For the history of the sport's relationship with betting, see the history of baseball and gambling, and for how win-total markets are built on the arithmetic above, see season win totals.
Common questions
How often does the best team win the World Series? Less often than most fans expect. In the two seasons in our data, the team with the best regular-season record won the title in neither year, and the 2019 champion was a wild-card team with the third-best record in its league.
Is a 100-win team a lock for the playoffs? In the two seasons in our data, all six teams that won 100 or more games (two in 2001 and four in 2019) reached the postseason. A 100-win team finishes at .617 or better, which is a large cushion.
Why do oddsmakers price baseball games so close? Because the true edge is small. The best teams are 60 to 65 percent favorites against average opponents, so the large majority of game prices are modest.
What is the single most useful number to remember? About 65 percent. That is roughly the best a very good team does over a season, and it is why a single game is never close to certain.
Does home field help? Yes, a little. The home team wins somewhat more than half of its games in most seasons, usually reported as a few points above 50 percent, and it is built into the odds. Check a current source for the exact figure.
Where to go next
If you want to see current prices and standings, the odds page and the standings page put today's numbers next to each other. For the pitching side of the matchup, our guide to breaking balls explains what starters throw, and the baseball betting topic page gathers more explainers.