A Quick Look at Our Playoff Odds

With the release of full ZiPS projections, our playoff odds are up and running. For the most part that means putting a number to things that we already know. The Dodgers are 97.7% likely to make the playoffs, which sounds about right. The NL Central is a four-way tossup with the Cubs out in the lead. The NL East has three teams each with around a one-in-three chance at it. That all tracks with intuition.

Indeed, for the most part, the standings are self-explanatory. That doesn’t mean that everything is obvious and intuitive, however. Let’s take a quick look at a few of the cases where a deeper dive is necessary.

It’s tempting to think of a team’s expected win total as just a sum of their WAR. After all, the W is right there in the acronym! As Dan notes every year, however, adding up WAR totals on a depth chart isn’t a great way to go about things. Rather than just do that blindly, however, we can look at teams whose projected wins diverge the most from their WAR.

To do that, we’ll need each team’s projected WAR totals. Thankfully, there’s a handy page that shows all that data. The Dodgers have the most projected WAR and the Orioles have the least.

With that data in hand, we can work out what win totals every team would have if you could perfectly project WAR onto wins. First, let’s figure out replacement level. There are 1120 projected wins across all the teams and 2,430 total wins available in a season. This leaves 1,310 wins as the amount that replacement level is worth. Spread that across the 30 teams, and that’s 43.66 wins per team.

By simply adding a team’s WAR and 43.66, we can get WAR-approximated wins. That looks like this:

Again, don’t take these as gospel. They’re just a starting point so that we can see where WAR and projected wins disagree. Next, let’s graph these WAR-approximated wins against the wins our playoff odds page projects:

For the most part, they track pretty well. That makes sense — WAR is a great context-neutral way of looking at how good people are, and for the most part, good players means more wins. If there were no correlation between projected WAR and team wins, or even a noisier correlation, WAR would be a worse statistic.

But they’re not perfectly correlated. Let’s look at the teams on the extremes to find out why. First, the five teams who project to outperform the sum of their WAR by the most:

It’s a mixed bag of teams. The Royals don’t have a notable bullpen or any obvious reason they should outperform, but they do have a pretty easy schedule. Even though they don’t get to play the Royals (bad luck!), the AL Central is pretty soft. And their lineup benefits from stacking up contact-centric players with speed; WAR might treat events as context-neutral, but in the context of having fast guys who get on base, putting the ball in play and limiting strikeouts is good. BaseRuns, which drives our projected win totals, knows that. Between the schedule and the lineup effects, I can see why the Royals do well.

The Red Sox and Rays are similar to each other. They have easy-ish schedules and OBP-centered lineups that work well together. The Red Sox, in particular, show the value of lineup construction; with more runners on base, it’s more valuable to reach base for each subsequent hitter. Having great lineups, like the Red Sox still mostly do even without Betts, is good in a way that WAR alone doesn’t capture.

The Rangers and Reds mainly stand out for their stalwart starting rotations. I’m not exactly sure on the logic here, but my guess is that considering the way we allocate playing time to starters, having an above-average fifth and sixth starter impacts win percentage in a way that WAR can’t quite capture. Our WAR numbers take into account run environments that pitchers create, but BaseRuns doesn’t. That difference could account for the gap. This is, to be clear, extremely off-the-cuff and speculative.

Looking at the teams that underperform the WAR-counting method most is also instructive:

The Angels have a few problems. First, their schedule is sneakily difficult. They get the Dodgers as a “bonus” interleague team, which stings. Their extra games (not every team in a given division plays an equal out-of-division schedule) relative to the Rangers are against the Rays, Yankees, and Twins. Overall, they just got unlucky on the schedule.

Additionally, there seems to be some of the same starter game going on. They project to give an uncomfortable amount of playing time to Julio Teheran and Patrick Sandoval. It’s a reverse Rangers situation; their lack of depth and balance in pitching seems to hurt them.

Moving on, the Phillies have a combination of a tough schedule (being the fourth-best team in your division hurts), an awkward lineup that’s half on-base guys and half power guys, and the same soft back of the rotation that hurts the Angels. The Pirates are in the same boat — fifth-best in the Central, and their fifth and sixth starters sting a little.

The Orioles play the hardest schedule in baseball, which is rough. A team with a .500 true talent level would project to win only 78.4 games against their schedule. That’s 2.6 wins right there, and we have the Orioles underperforming their WAR-tabulated wins by just more than 2 wins. That explains that.

That leaves only the Mets, and I have a theory about that one. The Mets have a really valuable bullpen. They get most of that value from two players, though: Edwin Diaz and Seth Lugo project for 3 WAR out of the team’s overall 5.1 from the bullpen. There’s significant extra value as well from Justin Wilson and Dellin Betances.

That sounds great, but there’s something feeding those projections: leverage. All four of those relievers, the bullpen’s top four on a rate basis, had average leverage indices higher than 1 in their appearances last year, and all project to do so again this year. In other words, all of them came into games in situations that were more important than average, and WAR for relief pitchers accounts for that.

When we switch to BaseRuns, however, that goes away. We divvy up playing time across the bullpen and then use the aggregate result as an input to the BaseRuns formula. There’s no concept of leverage — every pitcher’s input is weighed proportionally to their playing time. Thus the Mets are losing a lot of the bonuses we gave them when assessing their bullpen with WAR. That’s worth slightly over 1 WAR, which explains a lot of the difference.

So yes, counting up the WAR doesn’t work exactly perfectly. It works pretty well though — there’s a 96% r-squared between WAR-based win estimates and our Depth Charts estimates. And the biggest differences are small in the grand scheme of things. Three or four wins tops, accounting for strength of schedule and sequencing and everything? It doesn’t sound so bad.

So enjoy our playoff odds and projected win totals, or enjoy carping about them if they disrespect your favorite team. In a pinch, you could use WAR as a replacement for the win totals and you’d get most of the story. But that’s why we’re here — for the rest of the story. WAR can do a lot, but our playoff odds go the extra mile.