The Mets went into Saturday night's game against the first-place Nationals with a 43 percent chance of making the playoffs, according to Baseball Prospectus.
Or did they go into the game with a 24.8 percent chance of making the playoffs? That's according to FanGraphs.
Two of the leading baseball analytics websites have consistently disagreed lately about how likely it is that the Mets -- who prior to last night's game were two games out in the NL East and 31/2 games back in the wild-card hunt -- will reach the postseason.
The sites agree, by the way, that the Yankees have a great chance of making it. BP put the odds at 92.1 percent; FanGraphs had it at 93.4 going into Saturday.
The Mets just finished one of the nuttiest weeks in their history. It may interest their fans to know that all the drama was not in vain.
According to BP, the Mets' chances of making the playoffs increased 6.2 percent in the previous week and increased a whopping 12.1 percent from Friday to Saturday.
What happened Friday? The Mets acquired Yoenis Cespedes from Detroit and beat the Nationals in extra innings.
Checking the daily playoff odds updates on either (or both) site is one of the fun new ways fans can put a number to their team's chances. But what to do when the numbers aren't crunched the same way? To whom do you turn?
We asked Baseball Prospectus editor-in-chief Sam Miller and FanGraphs managing editor Dave Cameron three questions about their sites' systems. The first was how the playoff odds are calculated:
MILLER:"We establish what we believe to be the true talent level of each team, based on our projections of each player going forward. Then, with an established expected winning percentage of each team, we run simulations of the season 50,000 times; the playoff odds represent the percentage of simulations in which each team makes the playoffs.''
CAMERON: "Our playoff odds are essentially based on running 10,000 simulations over the remainder of the season (including the actual schedule remaining for each team) and reporting the aggregate results of those simulations. The inputs are based on the projected future performances of players based on the ZIPS and Steamer forecasting systems, which we host on FanGraphs.''
Then we asked if the playoff odds should be considered "good data" or a just-for-fun tool:
MILLER: "I definitely consider it good data. Baseball is incredibly unpredictable, of course, but using consistent ways of measuring and simulating team performance is far better than trusting either my gut or the incomplete snapshot that standings alone provide. Playoff odds improve standings by considering strength of schedule, by considering rather than ignoring the "luck" a team might have had (winning a huge proportion of a team's close games, for instance, is mostly not a repeatable skill; same with hitting exceptionally well with runners in scoring position), and by remembering that a great player who is having a poor year (like Robinson Cano, for instance) is usually a pretty good bet to play better going forward. Personally, I rarely even look at standings; playoff odds are my preferred way of tracking the progress of a season.
CAMERON: "Historically, the algorithms have proved to be pretty good tools. Attempting to forecast future events is always tricky, and there are always unexpected outcomes, but on the whole, I think our playoff odds stand up pretty well overall.''
Finally, we asked both about the discrepancy surrounding the Mets' chances:
MILLER: "Generally, the answer to our disagreements will come down to our projections disagreeing about a couple of key players, or disagreeing about a couple of key players on a division opponent's roster. (There are other factors, such as different ways of regressing the data, that affect the outcomes, but those are usually smaller and less interesting to the layman.) In the Mets' case, we don't actually disagree that much -- we project them to win about 1.8 more games going forward -- but because of their current standings, those two wins would be especially important to a wild-card push. Eyeballing it, I don't see immediately which players we disagree with, but it doesn't take much disagreement to add up to two wins.''
CAMERON: "In terms of the differences, our projections like the Nationals a lot more than the Mets, while BP's sees the two teams as being roughly equal going forward. Our projections [as of Tuesday] call for the Nationals to win 57.7% of their remaining games, compared to 50.4% for the Mets, while BP has it at 54.1% and 53.3%, respectively. Because we use different forecasting systems [they use PECOTA], there are going to be some areas of disagreement, and it seems like their forecasts see the two teams as closer in talent than we do.''