PLAYOFF PUSHHow Playoff Odds Are Calculated: Inside a Monte Carlo Season Simulation
A number like "33% to make the playoffs" sounds precise and a little mysterious. Where does it come from? Not from a formula. It comes from playing the rest of the season out, over and over, with dice. This is how Playoff Push produces the odds on every team page, and how every serious projection system works.
The idea in one paragraph
Take today's standings and the remaining schedule. For every game left, estimate the probability that the home team wins. Then simulate the whole remaining season: for each game, draw a random number and award the win accordingly. When the schedule is exhausted, apply the playoff format and tiebreakers and record who qualified. Do that a few thousand times. The share of simulated seasons in which a team qualifies is its playoff odds.
This technique is called a Monte Carlo simulation, after the casino. It replaces an impossible calculation, the exact probability across trillions of combinations of outcomes, with a large sample of random ones.
Step one: how good is each team?
The simulation needs a strength estimate for every team. Winning percentage is the obvious input, but raw winning percentage overstates how good a hot team is and how bad a cold one is, especially early in a season. A team that is 12-4 is not a .750 team.
Our engine regresses each team's record toward the league average by adding a fixed number of imaginary games at .500. For baseball that is a couple of weeks' worth of games; for football, where the whole season is 17 games, the regression is heavier relative to the schedule; for basketball and hockey it sits in between. The effect is that a team's estimated strength moves toward its record as the season goes on, which is what you want: by September the standings are mostly signal.
Step two: who wins a given game?
Knowing that Team A is a .580 team and Team B is a .470 team is not the same as knowing how often A beats B. For that the standard tool is the log5 formula, popularised by Bill James:
P(A beats B) = (A − A×B) / (A + B − 2×A×B)
where A and B are the two teams' winning percentages. A .580 team beats a .470 team about 61% of the time by this formula, not 58% and not 53%. It correctly gives a .500 team a 50% chance against another .500 team, and it never returns a probability outside 0 to 1.
On top of log5 we add a home edge. In baseball the home team wins about 54% of the time all else equal; in the NFL and NBA the edge is a little larger; in the NHL a little smaller. In soccer we also need draws, so a fixed share of matches are drawn before the winner is decided.
Step three: play it out, thousands of times
Each simulated season runs every remaining game once, updates every team's record, then hands the final standings to the same engine that computes our live standings tables. That engine knows the playoff format: how many division winners, how many Wild Cards, whether there is a Play-In, whether standings are on points. It applies the format and tiebreakers and produces a set of seeds.
We record, for the team you are following:
- whether it made the playoffs at all;
- whether it won its division;
- whether it got in as a Wild Card;
- its seed, and whether that seed carries a first-round bye;
- for the NBA, whether it landed in the Play-In range;
- for soccer, whether it finished in the relegation places.
Then we repeat. Playoff Push runs 1,500 to 2,000 seasons per team page, which is enough for the odds to be stable to about a percentage point. The seed for the random number generator is fixed per day, so the number you see does not wobble every time the page refreshes.
Why the odds move so much late in the year
Early in a season a single game changes little, because thousands of games remain to be drawn and any one result is diluted. Late in a season the opposite is true. When two teams are tied with ten games left, a head-to-head result cuts off a huge fraction of the branches in which the loser catches up. Odds that were 50-50 on Friday can read 68-32 on Saturday. That is not the model being twitchy. It is the season running out of room.
Elimination shows the same effect. A team's odds do not glide gracefully to zero. They hang at a few percent for a week, then a loss combined with a rival win takes the elimination number to zero and the odds snap to nothing.
What the model does not know
A season simulation is deliberately simple, and it is worth being honest about what it leaves out:
- Injuries and roster changes. A team that just lost its best pitcher for the season is treated as the team its record says it is.
- Pitching matchups. The engine does not know who is starting tonight.
- Rest and travel. A team on its fourth city in ten days is not penalised.
- Motivation. Teams that have clinched sometimes rest players; the simulation plays every game to win.
More elaborate systems fold some of this in. The gain is real but smaller than people expect, because most of the signal in a playoff race is simply the standings and the schedule.
How this compares to sportsbook odds
Playoff Push shows two different numbers side by side on a team page: our simulated odds, and the sportsbook's futures price for the championship. They answer different questions. Ours is a probability of qualifying, from the model above. The sportsbook's is a price that embeds a margin and moves with betting volume, and it is for the title, a much harder thing than qualifying.
The two are worth comparing anyway. If our odds have a team at 33% to make the playoffs and the market implies a 1% chance of winning the World Series, nothing is wrong: roughly a third of qualifying teams win a round, and the champion comes from a field of twelve. But if a team's playoff odds are high and its title price is unusually long, or the reverse, that tension is exactly what a bettor is looking for. Our guide to reading futures odds explains the conversion.
Reading the odds card
On every team page the Playoff odds card shows:
- Make playoffs, the headline probability of qualifying by any route.
- Win division or, in the NBA and soccer, the probability of a direct berth versus the Play-In or European places.
- Wild Card, the probability of getting in by the back door.
- First-round bye or top seed, depending on the league's format.
- Expected finish, the average number of wins across all simulated seasons.
The footnote records how many seasons were simulated and the regression and home-edge settings, so nothing is hidden. Pick a team on the home page and watch the number move with the standings.
Frequently asked questions
- How are playoff odds calculated?
- By playing out the remaining schedule thousands of times with a random number generator, using each team's estimated strength to set the win probability of every game, then counting how often each team qualifies.
- Why do playoff odds change so much after one game?
- Late in a season one result eliminates whole branches of possible outcomes. When two contenders are tied with ten games left, a single head-to-head win can move each team's odds by ten points or more.
- Are playoff odds the same as betting odds?
- No. Playoff odds are a model's estimate of the probability of qualifying. Betting odds are prices set by a sportsbook that include its margin and react to where money is being bet.