One Last Week For All the Marbles: ZiPS Projects the Postseason Home Stretch (9/28 Update)

9/28 Late Morning Update

`

The original data and methodology are below.
==
We’ve reached the final week of the 2021 regular season, and for fans of high-intensity, stretch-drive baseball — a group I think we can refer to as “everyone” — there’s still quite a lot to play for. Only five of the 10 playoff spots are claimed, with two of those five teams in a battle for a division title. And since there are just a handful of games left to play, we can move the ZiPS projections from the macro to the micro. In April, it’s always hard to project specific pitcher matchups, but with a week left to go in the season, it’s a more reasonable task of extrapolation. As a result, that allows me to adapt the ZiPS model into a game-by-game projection of the final week of the season for the relevant teams.

I’ve focused on three of the playoff spots, the two AL wild cards, and the NL East, along with the division versus wild card battle in the NL West. The Astros can still technically lose the division to the Mariners (one-in-about-1,800) or the Athletics (one-in-about-2,150), and the Cardinals could still have an epic collapse in which they lose six, the Reds win six, and they lose the tiebreaker (one-in-about-3,300). These could also become mathematical impossibilities quickly; if they become plausible rather than proverbial lottery tickets, I’ll update with the data.

Let’s start with the easy races.

NL East

The Braves enter the final week with a 2 1/2-game lead in the division but three games remaining against the Phillies. Their schedules are similar in strength, with Atlanta getting home games and Philadelphia on the road, something that’s largely canceled out by the former getting the slightly harder opponent (the Mets versus the Marlins). The edge comes from the cushion.

With the edge in the standings, ZiPS projects just over a four-in-five chance that the Braves will not have to play the Rockies in a makeup game on Monday. Overall, the Braves win the division 87.7% of the time without the makeup game, and the Phillies stick the Braves in at least a 1 1/2-game hole 1.0% of the time.

In 8.3% of scenarios, the Braves have a half-game lead after Sunday, requiring the Rockies game. If they win, they’re in, removing the need for an extra game against the Phillies. Kyle Freeland looks to pitch on Monday for Colorado, and ZiPS projects Atlanta as worse against lefties, so this is a closer projection than you’d expect, with the Braves beating the Rockies 54.6% of the time. In 3.0% of the simulations, the Braves go into Monday with a half-game deficit, meaning they would have to win to force a Phillies tiebreaker.

That leaves us with 5.4% odds of a second tiebreaker. This game would always be played in Philadelphia; with a two-game deficit in the head-to-head record, the Braves would have to sweep the Phillies to get the home game in a tiebreaker, in which case the Phillies would be mathematically eliminated, mooting the issue. In an Aaron Nola versus Ian Anderson matchup, the Phillies eke out a minor advantage in the projections. Add it all up, and ZiPS comes out with the Braves taking the NL East 94.8% of the time, with the Phillies winning it in the remaining 5.2% of the projections.

NL West

This is the simplest of the three scenarios, with no rainouts to make up.

ZiPS still projects the Dodgers as the better team, but with a two-game lead, the Giants win the division without needing a tiebreaker 87.3% of the time. That leaves an 8.8% projected chance of there needing to be yet another battle in the storied history of the two franchises, though the game sadly won’t be at the Polo Grounds, as history ought to demand. The final projection leaves San Francisco at 91.9% to take the division and Los Angeles at 8.1%.

AL East

Now we get to the brutal mêlée: five teams playing for two spots.

The three AL East teams are the clear favorites, though the Mariners and Athletics still have a lot of scenarios remaining without a strong whiff of implausibility. There’s too much to describe, so here are the overall final projections:

This was a bit complicated due to, as my colleague Jay Jaffe likes to talk about, Team Entropy. With five teams very close to each other, a tiebreaker is required in the projections 28.1% of the time, and this could even be more than just a three-team deadlock; ZiPS projects a four-team dance 2.9% of the time and all five teams finishing within a game of each other about once in every 670 simulations. I followed MLB rules for tiebreakers and made up my own for the five-team scenarios by hybridizing the two- and three-team tiebreakers. Suffice to say, this was a rather obnoxious bit of modeling that took up most of my weekend! As this post will be updated, I’ll add those games as need be; if I listed every permutation, the playoffs would be starting before you finished reading.

Just to make following the complexities of this race a little easier, I projected how a team’s playoff probabilities change with a forced result in each of the 24 games.

This chart is sortable, so you can reshuffle it for each of the five teams! Obviously, these will change on a daily basis based on the actual results.

Just for fun, here are the games ranked by the average absolute change in playoff projections for all five teams — to put it simply, the games that have the most impact on who makes the playoffs and who doesn’t:

As for the popcorn, you’re on your own!