Playoff Odds Explanation

The FanGraphs Playoff Odds are projections of how likely teams are to make the playoffs or win the World Series during the major league season. These odds are an adaptation of the coolstandings.com playoff odds originally created by Greg Agami and Sean Walsh.

To generate the playoff odds, we take the current standings, the remaining schedule, and each team's projected performance. We use those inputs to simulate the remaining season 20,000 times. We aggregate these outcomes to find the probability of winning the division or a Wild Card spot, along with winning the World Series and various playoff rounds. If a team has a 90% chance to make the playoffs it means that 18,000 out of the 20,000 simulated seasons end with the team making the playoffs.

We also report a projected W and L record. This is the average of the total wins in each of the simulated seasons for the team.

A value of 100.0% does not necessarily mean a team has officially clinched a playoff spot. It only means that a team made the playoffs in at least 19,990 of the simulated 20,000 seasons (which rounds up to 100%). Similarly, a value of 0.0% doesn't always mean that a team cannot mathematically make the playoffs. It means that they finished the season in the playoffs in less than 10 of our 20,000 simulations.

Projection Modes

We have four different projection modes. These determine our projections of each team's strength in the simulation. All the other factors are held constant.

FanGraphs Projections Mode

This mode is forward looking and uses the FanGraphs Depth Charts projections for rate statistics (a 50/50 blend of ZiPS and Steamer) and playing time to estimate the neutral-opponent winning percentage of each team -- in other words, how likely a team would be to beat a .500 opponent on a neutral field. These winning percentages are then used to find the odds of each team winning each remaining game in the major league season.

ATC Projections Mode

This mode is forward looking and uses the ATC projections for rate statistics (a combination of multiple projection systems) and our FanGraphs Depth Chart playing time to estimate the neutral-opponent winning percentage of each team -- in other words, how likely a team would be to beat a .500 opponent on a neutral field. These winning percentages are then used to find the odds of each team winning each remaining game in the major league season.

Season to Date Stats Mode

This mode is backward looking and uses current season stats, weighted more heavily towards the most recent games, to calculate the neutral-opponent winning percentage of each team -- in other words, how likely a team would be to beat a .500 opponent on a neutral field. These winning percentages are then used to find the odds of each team winning each remaining game in the major league season.

Coin Flip Mode

This mode is completely neutral and uses a coin flip to determine the winner of each game, giving each matchup 50/50 odds.

Strength of Schedule (SOS)

This is our estimation of the difficulty of each team's remaining schedule. We take the average neutral-opponent winning percentage of each of a team's remaining opponents, weighted by how many times the teams play. This is affected by the projection mode -- if you select FanGraphs Projections mode, strength of schedule will use opponents' forward-looking projections, rather than their performance in the season to date. Note that the winning percentages used for each opponent are not the same as the "ROS W%" column. That column takes the remaining schedule into account, while SOS calculations are performed on neutral-opponent win percentages (which you can find on our projected standings page).

One reason we decided to include SOS is that it can be useful to explain why certain teams are projected to do better in the future, especially in the last few weeks of the season, when remaining schedules can diverge dramatically. Early in the season, teams share so many opponents that SOS divergences are generally small.

We would caution against using this for determining how relatively difficult two division opponents' schedules are. If you are comparing a team projected to do very well (let's say the White Sox) to a team projected to do poorly (like maybe the Tigers), we aren't removing the common games between the two teams, so the Tigers appear to have a tougher schedule because they have to play the White Sox (high opponent winning percentage), while the White Sox in those games get to play the Tigers (low opponent winning percentage).

Viewing Changes in Playoff Odds

Have you ever wondered how much the Rays have increased their playoff odds since Opening Day? You can answer this question by viewing the changes in playoff odds. This finds the difference in our playoff odds between two different dates. There are a few preset date ranges to choose from -- yesterday, one week, the start of the month, and the start of the season. You can also enter your own starting and ending dates to find out, say, how your team's winning streak has vaulted it into contention, or how a star's injury marked the turning point in a playoff chase.

Distribution Mode

Rather than see our point estimates, distribution mode allows you to view the range of outcomes for each team's win total. These win totals average out to the win projection we report in the default version of our odds, but they show how varied our simulated outcomes can be. We also report the 25th and 75th percentiles for each team's win totals. The dispersion of these distributions will steadily fall as the season goes on -- the more games that have been played, the less uncertainty remains in how many games a team will win.

ZiPS Postseason Game-by-Game Odds

During the postseason, Dan Szymborski produces predictions for each series that use ZiPS projections and postseason roster and lineup information to produce game-by-game win probabilities. These are then aggregated to create series win probabilities. They are calculated round by round, rather than for the postseason as a whole.

You can read more about the methodology here. Generally speaking, the methodology is similar to the way we calculate playoff odds, but the rosters are based on projected game rosters rather than an aggregate team strength. The differences between these projections and the overall playoff odds are summarized here: