clock menu more-arrow no yes

Filed under:

The Playoff Familiarity Effect Is a Myth

We’ve got good news for Clayton Kershaw, Jake Arrieta, and the NLCS’s other regular-rest repeat starters: Fatigue matters. Facing the same lineup again doesn’t.

AP Images/Ringer illustration
AP Images/Ringer illustration

Baseball’s playoffs are weird in more than one way. In October, the weather, the quality of competition, and the schedule all differ dramatically from regular-season conditions. The difference in the schedule is the easiest to see: For the first six months of the year, opponents play two-, three-, and four-game sets. In the postseason, they play one-and-done wild-card games and best-of-five/seven series, which means that hitters often face the same starting pitchers multiple times in quick succession. Earlier this week, a reader named Charlie wondered via email whether that unique aspect of the playoffs might have hidden implications:

In this case, both teams would suffer the same fallout from any such effect, since each of the Game 6 (and, if necessary, Game 7) starters will have made a previous start in the series. Still, it would be useful for teams to know whether an in-series familiarity effect exists. Not only might it help dictate starting assignments — a team with two roughly equivalent options for a given game would be more likely to rely on the one its adversary hasn’t already seen in the series — but it might affect how quickly a manager would make a move to the bullpen.

Extensive research has revealed a powerful intragame familiarity effect, even after controlling for the pitcher’s fatigue: The more times a batter faces a pitcher in the same game, the better he tends to do. During the 2016 regular season, the average OPS allowed by starting pitchers rose from .725 to .753 to .792 in their first three trips through the order. Increasing awareness of this “times through the order penalty” is at least partially responsible for the fact that this year’s playoff managers are exploiting October’s off-day-filled schedule by pulling starters earlier than ever. Barring a burned-out bullpen, there’s rarely any reason for a postseason team to stick with a nonelite starter the third time through the order instead of calling for a fresh (and less familiar) arm.

We know a lot less about intergame familiarity effects. The few public attempts to establish whether batters have a leg up on pitchers the more they see them over the course of their careers have been far from conclusive, and they haven’t focused on the rare case of consecutive starts. But we can tackle that topic today, with research help from Dan Hirsch, proprietor of the baseball stats and analysis site The Baseball Gauge.

Dan identified every pitcher who started more than once in a postseason series, from Cy Young in the 1903 World Series through Thursday night’s NLCS Game 5, when Kenta Maeda and Jon Lester returned to the mound after starting Game 1. Before looking for a familiarity effect, we separated starters who pitched on three days’ rest from those who pitched on four or five days’ rest, for two reasons. First, pitchers on three days’ rest are more likely to feel the effects of fatigue: 80 percent of regular-season starts in 2016 came on either four (48 percent) or five (32 percent) days’ rest, while less than 1 percent came on three days’ rest. Second, the pitchers who are asked to start on three days’ rest in October tend to be better-caliber arms: Collectively, they allowed a .649 regular-season OPS in the years when they were asked to make postseason starts on short rest, while the four- and five-day guys both checked in at .670.

With those distinct samples separated, we can compare like to like: the three days’ rest guys in their first and second starts in the series, and the four- and five-days’ rest guys in their first and second starts in the series.

The first takeaway is that postseason pitching is hard: Due to the quality of competition, starters in both groups allow a higher OPS in both their first and second postseason starts than they do during the regular season, despite October’s less offense-friendly weather.

The more relevant takeaway, though, is that while the pitchers on three days’ rest decline significantly in their second starts of the series (and go a lot less deep into games), the pitchers on regular rest don’t. The regular-rest starters recorded almost the same ERA in their first and second starts, and their collective OPS allowed actually improved in their second go-arounds. Even though the more talented starters in the short-rest group perform better in their first start of a series than the guys in the regular-rest group do, the difference disappears in their respective second starts. Unless there’s a familiarity effect that applies on three days’ rest but wears off a day or later (which seems far-fetched), that strongly suggests that across postseason starts, fatigue is a more powerful force than familiarity.

Because baseball has changed just a bit since 1903 — that year, Young threw 341 2/3 innings in his team’s 141 games — we can also rerun the analysis for the wild-card era only.

In this sample, we see that the starters who pitch on three days’ rest suffer a steeper decline, which makes sense in an era when pitchers aren’t conditioned to do that. Again, though, the regular resters show no ill effects from facing familiar opponents.

Ideally, we would control for ballpark, home/road status, the composition of the opposing lineup in each start, how many days of rest the pitcher was on when he made his first start in the series, and whether he made any relief appearances between starts, but over samples this large, such factors should wash out. If there were a significant familiarity effect, we would see something here, but no penalty appears. That’s the case even though we might have expected some selection bias toward pitchers who did especially well in their first outing; pitchers who bomb in their first start in the series might be less likely to receive a second one, while pitchers who exceed their typical performance in their first start might be more likely to receive a second one, in which they would tend to regress. Odds are, though, that that’s not a significant factor, because the pitchers who make multiple starts in one series are likely to be top-of-the-rotation types whose standing wouldn’t be swayed by one great or ugly outing.

In other words, not every way in which the playoffs are weird matters much from a predictive standpoint. So, sorry, Charlie: There’s no reason to think that Kershaw and Hill or Hendricks and Arrieta, all of whom will be on regular rest, will pitch any worse this weekend than they normally would against a good team.