Will a Compressed Playoff Schedule Have a Measurable Effect on the Outcome?

The delayed start to the 2022 season due to the lockout has had a lot of small consequences for the structure of the season, ranging from expanded rosters to my least favorite thing, the continued use of zombie runners in extra innings. The last (we hope) of these changes is a slight alteration to the playoff schedule, which the league sees as a necessity in order to keep the postseason from straying too far into November. On Monday, MLB announced that the three-game Wild Card Series will be played without any off-days, while an off-day will be trimmed from the Divisional Series (between Games 4 and 5); teams in the ALDS get one additional off-day, without travel, between Games 1 and 2. The Championship Series will lose an off-day between Games 5 and 6). The World Series is business as usual.

While I expected this configuration for the Wild Card round (it was already accounted for in the generalized ZiPS projections for postseason performance), there are some slight tweaks that need to be made to account for the changes to the Division and Championship Series with respect to pitching. When projecting the roster strength of a team for the purposes of postseason probabilities, ZiPS weighs pitchers at the top of the rotation more heavily. That’s because historically they have gotten a larger percentage of starter innings in the playoffs than during the regular season. But losing an extra day of rest could result in teams using the pitchers after their No. 3 starters more heavily, as well as more dilemmas involving bringing back a top starter on three days rest. There are also possible consequences for the bullpens. In other words, teams will need to be slightly deeper than normal this playoff season.

So, how do we account for that? To get a rough estimate — I’m not sure there’s a methodology that will let us do any better than that — of the potential effects of the compressed schedule, I went back into the ZiPS game-by-game postseason simulations and put together a new, quick simulation for starting pitcher usage. I used projections as of Tuesday morning.

I used the order on our depth charts to give every starting pitcher a number to represent their rotation rank and started each team’s postseason run with their No. 1 starter. I then went through the rotation in cardinal order, with the simulation defaulting to the top pitcher available with four days of rest. To try and emulate how top pitchers are used, for elimination games (where a team’s losses going in are equal to one in the Wild Card Series, two in the Division Series, and then three afterward) I instructed the simulation that there was a two-in-three chance that a No. 1 starter would be used on three days rest; if that pitcher wasn’t available, I told it there was a one-in-three chance that the No. 2 starter would go. If neither of the top two pitchers was available for an elimination game that wasn’t an elimination game for both teams, there was a coin-flip to see if it would be a bullpen game instead (bullpen games are less common in Division Series Game 5s and Championship/World Series Game 7s). I then ran the simulation both with and without the two off-days, taking into account the extra off-day American League teams get after Game 1 of the Division Series. I also assumed that there were two off-days between every round of the playoffs. That isn’t strictly true, but the conditional probabilities were getting a little complex already and this convenient little kludge holds true for both scenarios equally. This is, naturally, far tidier than real life is.

I then calculated the difference in a team’s FIP based on the projected mix of pitchers in each scenario. I repeated this exercise for every team that has at least a 5% chance of making the playoffs in ZiPS, which amounts to every team with at least those odds via the FanGraphs Playoff Odds, plus the Orioles. These FIP projections have already been put into park-neutral terms:

Overall, the differences are fairly small. Amusingly, the Orioles and Cardinals actually improve slightly due to the simple fact that ZiPS prefers most of the O’s rotation to Jordan Lyles, and likes Jordan Montgomery and José Quintana better than Miles Mikolas and Adam Wainwright.

Bullpens are a little trickier to simulate. I needed to look at how different depth chart rankings for relievers had resulted in different usage in the past, which meant I needed historical depth chart bullpens rankings. Rather than using a simple method — like assuming that teams strictly divided inning totals by depth chart ranking — I chose to make my life more difficult. After some exercises in dimensionality reduction, I used our depth chart rankings to construct a model depth chart based on saves, innings, FIP, and leverage when entering a game.

I kept it simple to avoid any overfitting wackiness and got an r-squared of 0.77 and an RMSE of 1.14 depth chart positions. I then went back as far as we have WPA data and went through each postseason series, separating them into two categories: games that were the third consecutive game without an off-day and games that weren’t. This got me an estimate for the mix of relievers used in either situation. I then applied this mix to the park-neutral FIP projections for current bullpens:

Using these changes, no individual team’s change in World Series championship probability moved by more than a tenth of a percentage point. In other words, having fewer days off, while exhausting for teams, isn’t likely to cause enough changes in team construction to alter any of our thoughts about the postseason.

But wait, there’s one more issue to worry about: jet lag resulting from long travel with an off-day. In the playoffs, this usually isn’t a problem, but this time, it could be. A team may have to fly from Los Angeles to New York without a day of rest. I made a further change in the simulation to reflect the results of a study on jet lag performance in baseball from 2017. I then applied the change to all games for which there were no off-days and a team’s location changed from the Pacific Time Zone to Eastern or Central. The updated model changed every team’s championship percentage by less than a tenth of a percentage point. This held true even in the face of a few the different scenarios; a Mariners vs. Yankees ALCS would have a travel issue while the Yankees playing one of their AL East rivals would not, but in either case, the effect was pretty muted.

So, what have we found? Not much, but it’s always good to check these things for an effect if possible, and knowing that an effect is small or non-existent remains useful. Score another victory for Betteridge’s law of headlines, the prime Michael Jordan of adages!