Will the Compressed Schedule Make Depth Starters More Valuable?

There are plenty of rule changes coming to baseball in the 2020 season. We’ve written about many of them: the universal DH, the extra-innings rule, and expanded rosters, to name a few. Today, I thought I’d take a crack at something less immediately evident but still meaningful: the denser schedule.

In 2019, teams played 162 games in 186 days. That meant the schedule was 13% off days, give or take. Four of them were clustered around the All-Star break, but for the most part, they were spread out evenly. Off days are a welcome respite in a team’s schedule, a break from the grind. Sometimes they’re necessary for travel, of course, but mostly they’re meant just to be time for players to recover from the grueling march to October. This year, 60 games in 66 days means only 9% off days.

For most players, an off day is simply that. For pitchers, however, days of rest carry greater meaning. A day off is a day closer to starting again. Imagine a schedule with 80% off days — a game every five days. Your ace could pitch more or less every game, give or take a maintenance break here and there. Conversely, in a schedule with no off days at all, every pitcher would take the same number of turns in the rotation.

Given an off day, teams can, in theory, squeeze extra starts out of their ace. In practice, it doesn’t quite work that way. It also doesn’t not work that way, however; seven pitchers made 34 starts in 2019, more than a fifth of the games on the schedule. Teams are logical about giving their best starters extra turns when they can. An extra Justin Verlander start beats a Jose Urquidy start, no offense to Urquidy.

To model this effect, I built a simple model. First, I took a pre-COVID 2020 schedule. I used the Cardinals, but any team would do. I modeled out a simple rule: what if a team simply started the best fully rested pitcher every day? In other words, a six-day block with one off day would look like this:

Teams don’t act exactly like this in real life. They tend more towards a strict rotation, though not exclusively so. Bear with me, though. This is all going to be rather theoretical. We can’t exactly model teams’ behavior, and we can’t reasonably create a plan to handle injuries. So instead, let’s just roll with it. This isn’t a perfect model of team behavior, but it might get close. Here’s how a 162-game season looks in that world:

It’s not quite the same as the way the actual majors work, but it’s close. The best players get more starts. Fifth and fourth starters get fewer turns, fifth starters in particular. A team’s ace gets 50% more starts than its fifth starter, with a waterfall down from there. The general idea is straightforward: more off days means more time to get your best pitchers back to 100% rest.

We can add one wrinkle before moving on to the actual 2020 season: cancellations. Let’s add rainouts randomly distributed throughout the season. For this, we’ll need a few more abstractions, because rescheduling rainouts is hardly an exact science. I decided on the following: if a game is rained out, it’s always played the next day. If it was an off day, easy! They simply play the game. If there was a scheduled game, now it’s a doubleheader. Because this will frequently result in six games in five days, I’ve added a sixth starter, a minor leaguer or swingman who can pick up the slack if necessary:

A quick note on the rainouts: to get the effect right, I wanted to run a ton of simulations. Sprinkling something randomly throughout a season once doesn’t really tell you much — variance and all. I wanted to run a ton of simulations, so I gave every game a 4% chance of being rained out and then ran the season a million times. Why a million? 100,000 seemed too low and 10 million seemed too high. We’re very scientific here.

The results mostly make sense. When there are six games in five days, someone is getting squeezed — a start disappears into the aether, swallowed up by a starter the team hoped not to use. That start isn’t likely to cost the ace a turn in the rotation; he’s already going as often as he’s rested, more or less every five days. I can’t stress enough that this is only a rough model. It’s a guess, and a stylized one at that. That doesn’t mean it isn’t useful, only that we’re not trying to act exactly like a team here. The broad lessons still matter.

Before we go to a 60-game season, let’s recast everything in terms of percentage of total starts, like so:

As compared to a 20% across-the-board split, our top three starters get an extra 4.6% of the workload in total. Most of that comes at the expense of the fifth starter. This isn’t to say that fifth starters aren’t valuable — in a real season, injuries inevitably come up, which creates extra opportunities for fifth starters in a different way. From a pure scheduling standpoint, however, and assuming perfect health, more off days decrease the impact of the back of the rotation.

Next, let’s cut the season to 60 games in 66 days. As there’s no exact schedule yet, I’ve sprinkled the off days evenly throughout the season. What should we expect to find? To be honest, I wasn’t quite sure. Fewer days off as a percentage of games played might mean fewer chances to double-dip with a team’s best starters. On the other hand, I spaced the off days out more evenly than in the regular season. There’s no four-day All-Star break, no disproportionately sparse early season, none of that scheduling weirdness. It’s just a homogeneous block of games and breaks, games and breaks, with rainouts popping up at random.

With no rainouts, the 60-game schedule doesn’t have much to differentiate it from a full season:

Every starter is within .3% of their game allocation. The even spacing makes up for the fewer off days almost exactly.

Next, it’s time to make it rain. Make it rain 4% of the time at random, to be precise. Over a million simulations, our sixth starter gets some work in:

For the most part, these are small differences. On a 162-game scale, which will probably make the most sense in people’s heads, it’s an extra start for the fifth and sixth starters at the expense of one by the ace. That’s worth three runs on the outside, assuming a really good ace and really poor fifth/sixth starter. That’s over 162 games — it’s more like a run at most in this year’s abbreviated schedule.

What if we cranked the rain up a notch? I ran the simulation again with an 8% chance of rain:

Now the sixth starter is getting more work, but the top two pitchers carry the same workload as always. In fact, the biggest losers in this scenario are the third and fourth starters. That might not be an intuitive result, but think of it this way: when the sixth starter is forced into a game, someone needs to lose a turn.

There are enough off days, particularly when taking into account the rainouts, that the ones and twos usually don’t miss a turn — they can pitch as soon as they’re rested, and the off days help them get ready in fewer games. That leaves the third and fourth starters on the shelf to keep the top two starters on their regular schedule. The fifth starter loses some turns there too, but there’s a countervailing effect: sometimes a rainout turns a schedule that would normally allow the fifth starter to be skipped into one where he’s needed. The net effect is nearly no change.

Is the effect of the schedule particularly meaningful? On balance, I’d say no. Half a start, on average, for aces? A turn here or there moved from a fourth starter to a swingman? It’s all negligible. To me, though, that finding itself is significant. There are all kinds of reasons to think of this season as weird. The compressed rest schedule won’t massively add to that, though, by changing the way rotations work. In this small corner of the game, at least, it’s nearly business as usual.

Are there limits to this study? For sure! I’m ignoring injuries, which are a valuable reason to have good pitching depth in both 60- and 162-game seasons. I’m using a simplistic rescheduling rule rather than intelligently hunting for off days. I’m treating keeping pitchers on a set schedule as having no value — fifth starters in this universe sometimes go 10 days without starting, which isn’t something teams like to do. The list goes on.

Despite all that, I’m comfortable saying that the new rest schedule will look roughly like the old rest schedule when it comes to starting pitching. In a season where almost everything else seems to be up in the air, it’s nice to have a little bit of consistency.