March 3, 2005
The Morning After
If you're heading down to Florida or Arizona in the next month, you'll surely be taking in a few spring training games here or there. One thing you won't see is extra innings. Much like the occasional All-Star game, spring training games sometimes end while tied. This is typically fine with everyone because teams aren't there to win the games, they're there to determine their roster, work on the fundamentals, and get a nice base tan. No one wants to get hurt and no one wants to get too tired, so after nine innings are up, everyone goes back to the batting cages or the golf course and has a great time.
During the regular season, things are obviously different and it's clear that extra-inning games can be a drag on a team. Playing nine innings every day is hard enough, but throw in the occasional 13- or 14-inning affair and players can show signs of fatigue, particularly in the bullpen or behind the plate. Managers find themselves with fewer options than normal the next day because players are still tired from the previous day. Doubleheaders have similar effects on clubs. Free baseball is great for the fans, but it can be rough on their team.
Just how much the extra innings affect a team is difficult to gauge. Looking at winning percentage in the following game doesn't reveal much because often the same two teams who battled through the long yesterday are playing again. Teams that played extra innings or a doubleheader and then play the next day are going to have a collective winning percentage of .500 no matter how tired they are or how many extra greenies are in the coffee. Instead, we must look at situations in which one team played extra innings the night before and the other did not.
In 2004, there were 218 extra-inning games. That may seem like a nice sample size, but 105 of those games were only 10 innings. Let's push ahead anyway and see what we find. Looking at all teams in 2004, we can find 89 games in which one team played extra innings the day before and their opponent did not. In those 89 games, the tired team--for lack of a better term--won 38 games for a winning percentage of .427. While that may seem to validate the idea that tired teams don't do well the next day, it may simply be a case where those tired teams happened to be very bad in general.
To correct for this, I've weighted the overall winning percentage of our tired teams by how many games they played and how good they were in general. For example, the Diamondbacks played in four such games, so their winning percentage--which looks a lot like a batting average--is counted four times. Add up all the teams like this and the teams in these 89 games had a weighted winning percentage of .505, nearly 80 points higher than their actual winning percentage in those day-after-extra-innings-games. Looks like we could be onto something.
Before we get too excited, let's see how things shaped up over the past 10 years:
YEAR G W L WIN % eWIN % DIFF 1995 99 50 49 .498 .505 .007 1996 78 32 46 .498 .410 -.088 1997 110 55 55 .502 .500 -.002 1998 99 46 53 .513 .465 -.048 1999 80 35 45 .516 .438 -.078 2000 85 41 44 .485 .482 -.003 2001 66 42 24 .508 .636 .128 2002 70 34 36 .497 .486 -.011 2003 97 53 44 .503 .546 .043 2004 89 38 51 .505 .427 -.078In seven of the last 10 years, the tired teams have had a worse winning percentage than the fresh teams. However, throwing out years where the difference was almost zero--1995 (.007), 1997 (-.002), 2000 (-.003), and 2002 (-.011)--leaves four seasons in which the tired teams performed poorly and two in which they played very well. Overall, tired teams have a 426-447 record (.488) with a weighted winning percentage of just over .500. Tired teams perform just slightly worse than fresh teams.
Using all extra-inning games may not be the best way to determine if teams are in fact tired. As mentioned, in 2004 nearly half of all extra inning games were merely 10 innings. Removing those games from the numbers yields a much smaller sample, but much more dramatic results:
YEAR G W L WIN % eWIN % DIFF 1995 34 12 22 .516 .353 -.163 1996 43 17 26 .501 .395 -.106 1997 50 23 27 .497 .460 -.037 1998 57 28 29 .510 .491 -.019 1999 40 17 23 .531 .425 -.106 2000 48 22 26 .492 .458 -.034 2001 41 26 15 .505 .634 .129 2002 38 17 21 .476 .447 -.029 2003 49 24 25 .496 .490 -.006 2004 45 17 28 .511 .378 -.133These teams underperformed their weighted winning percentage of right around .500 to the tune of nearly 50 points, going 203-242 since 1995. Only once in the last 10 years did tired teams outperform their weighted winning percentage.
As mentioned above, the other big exhaustion factor is doubleheaders. Though they are rare these days, they still happen from time to time and are certainly tiring, even for those of us in the stands. Running the same numbers, here's what we get:
YEAR G W L WIN % eWIN % DIFF 1995 10 8 2 .500 .800 .300 1996 13 6 7 .493 .462 -.031 1997 19 10 9 .530 .526 -.004 1998 12 9 3 .511 .750 .239 1999 15 6 9 .501 .400 -.101 2000 12 7 5 .508 .583 .075 2001 4 3 1 .529 .750 .221 2002 3 2 1 .431 .667 .236 2003 8 4 4 .458 .500 .042 2004 25 6 19 .464 .240 -.224I was rather startled by the amazing jump in doubleheaders in 2004, but the reason isn't just because there were more doubleheaders (there were 38 as opposed an average of 25 for 2001-03), it was because there were more doubleheaders where one of the teams played someone else the next day. Looking at the list as a whole, it's certainly easy to point to 2004's nice large sample size and say that that validates the idea that tired teams don't do well. But the problem is that in most years, the sample size is just too small. Adding everything up, the tired teams went 61-60 while their weighted winning percentage is just below .500.
Though teams playing the day after a doubleheader don't appear to suffer adversely from the fatigue, those teams that played significant extra innings did. There are several possible reasons for this, mostly due to the ability to prepare for a doubleheader. Extra-inning games can run late into the night, taxing a bullpen that has little idea when the game will end. Doubleheaders usually last no later than regularly scheduled games and teams are able to prepare for them by sparing the bullpen in the days before and using another starting pitcher in the second game of the double dip.
The difference in performance in games after longer extra-inning games would turn a .500 team into a .450 team over a long period of time, costing a team about eight wins on the season. On average, teams only find themselves in such a disadvantageous situation about one or two times a season, so it's unlikely that it's a significant problem for any team. But the teams in spring training have at least one more small excuse to hit the bars a little earlier. And so do those of us in the stands.