September 1, 2009
Ahead in the Count
Home-Field Advantage, Part Four
In the first three articles of this series, we have studied what home-field advantage affects, who it affects most, and where it shows up most. We have found that home-field advantage affects nearly every aspect of a team's performance, including pitching, defense, baserunning, and offense. We found that the Rockies are the only team that has statistically significant home-field advantage, and that most other teams are bound to win about eight percent more games at home than on the road in the long-run. We also found that home-field advantage was larger in interleague games than intraleague games, larger in interdivision games than in intradivision games, and even within divisions, it was larger the further apart the teams played. This suggested that travel might be playing a significant effect in home-field advantage. Further evidence of this came from the fact that interleague games within teams in equivalent divisions (e.g. East vs. East) had smaller home-field advantages than interleague games where longer travel distances were involved.
Seeing as how distance traveled has such a clear effect on home-field advantage, it leads us to wonder if the process of traveling has an effect-perhaps through jetlag or some other medium. Many readers believed that home teams would have especially large home field advantages if the first game of the series, perhaps because they were adjusting after travel. In fact, this is not the case at all:
Home-Field Win% by Series Game:
The middle game(s) of the series are the one that show the most pronounced home-field advantage, and by a statistically significant amount. The last game of the series does show the least home-field advantage, which is an indicator that adjusting to surroundings may be a part of home-field advantage. However, the fact that the first game has a smaller home-field advantage than the middle game(s) is shocking, and it makes little sense taken at face value. I still do not have a definite answer as to why this is true, and I welcome comments. However, I will present some clues in this article to determine what is happening.
The term "middle game(s)" is obviously lumping together a few different things, so let's start by looking at home-field advantage by series length. In between 1998-2008, there have been series that have varied from one game long to five games long. Here is the home-field advantage in each series length:
One-Game Series: 56.0%
Two-game series have a statistically significantly smaller home-field advantage, and three-game series have a statistically significantly larger home-field advantages (although that is probably only because interleague series make up 14 percent of all three-game series instead of only 10 percent of all series). The fact that two-game series show less home-field advantage could be a clue, however.
Breaking things down game-by-game within a series, we can start to see the trend a little better. Although it does not seem to be a significant clue, I will also include the average home score/away score because it may help readers discover something that I have not. Keep in mind that the home team does not bat if they are winning in bottom of the ninth so they will have smaller run totals than you would otherwise expect.
Game 1 of 1: 56.0% (4.89-4.89)
Game 1 of 2: 51.3% (4.79-4.88)
Game 1 of 3: 54.3% (4.90-4.69)
Game 1 of 4: 53.0% (4.86-4.73)
Game 1 of 5: 45.5% (5.23-5.77)
None of these samples are large enough to really generate statistical significance for any individual series game, but it seems pretty clear that the effect is on the day before getaway day. Given that this is true, it made me wonder if there was something about travel days and offdays that played a role. Not only might it give teams the option to settle themselves earlier in a different city, but also to reset their rotation.
I checked whether home teams had a larger home-field advantage if they had an offday before the first game of the series. This effect was positive, though not significant: 54.1 percent vs. 53.6 percent. However, there was a statistically significant effect of having an offday on the road before the game; away teams won 47.6 percent of games when they had an offday the day before the first game of the series, but won just 45.4 percent of games when they had no offday the day before game one. What this means is that there is probably some travel effect on the first game, but some other factor playing a role towards the beginning of the series making it easier for away teams to win.
I checked the effect of having an offday two days before game two of a three-game series. Interestingly, this helped home teams by a statistically significant amount: they won 56.8 percent of the time in game two of a three-game series if they had an offday before the series begin, and only 54.0 percent if they did not. The run totals demonstrate this further, as the average home team outscored their opponent in game two of a three-game series by a score of 4.95-4.60 if they had an offday before the series, and only 4.87-4.80 if they did not have an offday before the series.
Away teams, however, lost more often when they had an offday two games before the series, but by a small amount: 55.5 percent for game two of three with an offday two days before, and 54.9 percent for game two of three with no offday two days before. This effect does seem quite strong when you look at run totals though, as they were outscored 4.85-4.61 in game two of three if they had an off-day before the series, and only 4.94-4.79 if they did not.
For four-game series, there is a huge effect in the third game of the series. This is apparently primarily if the away team had an offday before the series begin. Home teams won 58.2 percent of game threes if they had an offday before a four-game series, but only 54.3 percent of third games if they did not have an offday before a four-game series. That is not quite statistically significant. They outscored their opponents 4.90-4.41 in game three of four if they had an offday before the series, and only 4.75-4.73 if they did not.
However, what is statistically significant at the 95 percent level is that if an away team has an offday before a four-game series, they lost game three 60.3 percent of the time in game three, but with no offday, they lost 53.8 percent of the time in game three. They were outscored 4.91-4.33 if they had the offday before the series, and were even at 4.76-4.76 if they did not. This effect was even more extreme with an offday before a Thursday-Sunday series. Away teams lost 68.1 percent of the 144 game threes of four-game series that took place on Saturdays, which is statistically significant even at the 99.9 percent level.
Looking at all of this, my belief is that there is not only a travel hangover effect, but a differing one based on distance traveled, in addition to a "time away from home" effect. It seems that home teams really tend to do better in those middle games with an offday before the series, and away teams tend to do worse in middle games with an offday before the series. Although traveling earlier might help away teams avoid being jetlagged or tired during the earlier part of the series, it seems that the longer the series goes, the more tired they get. In the middle game, I suspect they are still not fully recovered from travel, but may also still be suffering from more time away from home, thereby giving the away team a larger disadvantage in the middle of the series.
I also tried looking at actual travel distance. Recall that last week we found that intradivision games among teams who played closer together had a smaller home-field advantage. I decided to look at intradivision matchups for three-game and four-game series involving both closer than average distances and further than averages distances. I changed my definition to reflect a helpful comment about mileage and degrees latitude and longitude.
Three-Game Series, Same Division, Less than 597 miles away
Three-Game Series, Same Division, More than 597 miles away
Four-Game Series, Same Division, Less than 597 miles away
Four-Game Series, Same Division, More than 597 miles away
It should be pretty clear from these that this "day before getaway day" effect is only a significant effect for games played between teams who are close together. In fact, if we look at interdivision games, we will see similar effects for those interdivision games where teams happen to be closer than 597 miles (although this is not a large set):
Three-Game Series, Different Division, Same League, Less than 597 miles away
Three-Game Series, Different Division, Same League, More than 597 miles away
Although there are not many four-game series between teams less than 597 miles away but in different divisions, it is difficult to know whether a similar pattern would emerge whereby the "day before getaway day effect" is only significant for teams who play in nearby cities.
The fact that this effect only holds for teams that play closer together suggests that perhaps one of the effects depends on either mode of travel or travel distance. Perhaps something about the last game of the series gives the road team back an advantage, but they struggle through the first couple of games. However, it may be that much of the effect on the first day is mitigated if the team only traveled on a bus to the game, rather than dealing with jetlag in the first game if they traveled on a plane. However, by the second game, regardless of mode of travel, the effect of staying in a hotel tends to wear on players. It sounds a bit far-fetched, but there probably is something far-fetched that does explain this effect.
This could be a psychological issue as well, whereby road teams can generate the adrenaline to overcome some of home-field advantage in the first game and last game of the series, but they lose their steam in between. I do not know for sure, but these effects generally point to either an effect on road teams that helps them early or late in the series, or an effect on road teams that hurts them in the first couple games and hurts them in the last couple games, amplifying home-field advantage for games in between. In reality, there may be several effects in play. This combination of these effects-travel exhaustion that depends on distance traveled, and time away from home-might be leading to this bizarre outcome.
I wondered if day of the week in general may play a role, but the day of the week of home-field advantage mostly seems to reflect the home-field advantage by series length:
I suspected that the reason that the last game of the series may have smaller home-field advantage could be that it is more likely to be a day game, and therefore I suspected that road team players would be less likely to have a disadvantage in these games. However, it seems like Sunday games have larger home-field advantages than Thursday games, which would probably not be true if night games are a cause of home-field advantage.
A natural question that many have asked is whether long homestands or road trips have an effect on home-field advantage. This does not seem to be all that significant. Consider the home winning percentage by game in a row at home:
Now consider the away winning percentage by game in a row on the road:
It does seem that road teams may do a little better early in a road trip than late, but only by a small margin.
All in all, this is an article that probably asks more questions than it gives answers. The previous three articles in the series have come to conclusions about the topics they addressed, and many of the answers were not surprising. In this article, there is clearly a surprising result-that home-field advantage is much larger in the middle game(s) of the series-and my hypothesis as to why that might be is not necessarily correct. I do think it's most likely that the significance of these games is a result of an effect of home-field advantage that increases with the duration of the series and a second effect of home-field advantage that decreases with duration of the series, but it is not necessarily the jetlag/time away pairing of effects that I presented.
The fifth article of this series will discuss individual player differences, but after that, I may choose to discuss run totals, or another subset of this data, depending on reader questions. I look forward to hearing what you have to say.