June 25, 2010
Ahead in the Count
Not So Home-Field Advantage
On September 26, 2007, the Cleveland Indians won a “home game” against the Seattle Mariners at the Mariners’ home ballpark, Safeco Field. The original game along with a three-game series had been snowed out in April that season. Two of the games were made up during the season in Cleveland on mutual off days. However, without a third mutual day off, the teams simply made up the game as part of a regularly scheduled series in Seattle. While other games had been up in the opponent's home ballpark, Major League Baseball decreed that the Indians would be the home team in this game. Thus, for the first time since 1913, a team batted first in its own park.
This weekend, it will happen again. While other economists are more focused on the implications of the G20 summit, this economist is focusing his skills on the baseball implications of the G20 summit—namely, that Toronto hosting all the major world leaders has forced the three-game series between the Phillies and Blue Jays to be relocated to Philadelphia, where the Phillies will bat first in all three games at Citizens Bank Park.
Many casual fans believe that the home-field advantage comes primarily from the host team batting last. In reality, in my five part series last year I explained that this was not the source of home-field advantage, as one-run defensive strategies are just as effective as one-run offensive strategies. Instead, there were a number of other things that I learned about home-field advantage that can help give us a clue about what to expect this weekend. In addition to looking back at some of that research, I have also reviewed the 222 games played at “neutral sites” that have recorded box scores on Retrosheet. This can allow us to isolate some more of the effects of home-field advantage and what might be different about this weekend’s series.
One of the first things we learned about home-field advantage is that it has actually been relatively consistent at 54 percent through time. Thus, looking at the 222 games played at neutral sites, we might get a better sense of how much of this is eaten up by changing the venue. Of the games that were not tied, the “home teams” won a solid but smaller 51.4 percent, which first indicated to me that maybe a lot of home-field advantage was taken away by removing these teams from their home parks. However, I then realized that I should check the quality of teams that played in those games. That indicated that the "home teams" should have only won 46.9 percent, meaning that the 4.5 percent increase in winning percentage is actually what we would have expected otherwise.
Much of what I learned about home-field advantage last year related to its relationship with two factors—familiarity and distance. This was confirmed by later work by Russell Carleton as well. The familiarity factor suggested that I look at the home games more closely, and I realized quickly that 43 of the neutral-site games were in 2003 and 2004 when the Expos played in San Juan, Puerto Rico, where they presumably got at somewhat used to the playing conditions at Hiram Bithorn Stadium. Another 58 games were play from 1929-32, when the Boston Braves regularly played Sunday home games at a park called Braves Field. Thus, I realized that perhaps removing these from my sample might provide some clarity and help get rid of some bias in the sample. However, this actually increased the home-field advantage to 59.5 percent, when the matchups suggested a 53.1 winning percentage. Of the remaining 119 games, 66 were played either in the home team's city or a nearby location. Removing those games, the home team only won 49.3 percent of the games, though the matchups suggested it should have won 47.2 percent. We are nowhere near learning anything from these sample sizes, but this certainly is an indication of what we would expect—distance does play a role. That’s certainly one strike against the Blue Jays, who had to travel from Toronto to meet the Phillies tonight in Philadelphia. Meanwhile, the Phillies remained in Philadelphia after concluding a three-game series with the Indians on Thursday afternoon at Citizens Bank Park.
One of my more interesting findings from last summer’s series was that the differential in runs between home and away teams was actually larger in early innings than late innings. This could be indication of a couple things, one of which is that familiarity may play a role as the visiting team takes longer to get adjusted to the mound and batter’s eye. Interestingly, my later extension of this work in the Society of American Baseball Research’s Baseball Research Journal revealed that this difference was actually particularly large in all games of a series, not just the first game. If this is about the starting pitcher adjusting to the mound, that makes some sense, but it is certainly surprising that visiting batters are struggling with the unfamiliar hitting background equally in all games of the series. Using all 222 neutral-site games to get a sufficiently large sample size, we do see this gradual decline in home-field advantage in runs scored by inning. Innings 1-3 in these 222 games saw the home team outscoring the away team by 7.9 percent, while this fell to 2.4 percent for innings 4-6, and 1.1 percent for innings 7 and 8. These effects are smaller than those found in last year's study, when the first eight innings had the home team outscoring the road team by 17.2, 8.5, 11.1, 6.5, 9.6, 7.5, 8.0, and 7.7 percent.
That first-inning differential is particularly interesting, because it could be caused by another factor. Instead of the issue being that the home pitcher was more familiar with the mound than the visiting pitcher, perhaps the home pitcher is just more prone to be more effective immediately after warming up. Maybe a home pitcher knows the exact moment that he is going to step on the mound and can warm up until a couple minutes beforehand, and maybe a visiting pitcher loses focus and cools off while in the dugout waiting for his team to finish batting in the top of the first inning before getting the ball. While first warning readers about the small sample size, I can now say that we do see some clear indications that the first-inning warm-up effect may be part of home-field advantage. In all 222 games, the home team outscored the road team 114-91 in the first inning, but was outscored 112-71 in the second inning. Removing the Boston Braves' Braves Field and Expos' San Juan games, the home team outscored the away team 62-32 in the first inning, while being outscored 51-39 in the second inning. In the 53 neutral-site neutral-city games, the home team outscored the road team 17-13 in the first inning, but was outscored 25-14 in the second inning. If the Blue Jays are going to gain any advantage, it is going to be from making Roy Halladay, Cole Hamels, and Jamie Moyer sit and watch their hitters face Jesse Litsch, Shaun Marcum, and Brett Cecil. Perhaps another small advantage for the Blue Jays could come from having a regular designated hitter on their roster, although I am not confident this effect is large enough to matter much.
The rest of home-field advantage clearly favors the Phillies, however. Even though the Phillies have had a relatively small home-field advantage in recent years, I showed in last year’s series that outside of perhaps the Rockies, there is little indication that any team has any sizable ability to win more than 8 percent more games than they lose on the road. This will not get any easier on Saturday and Sunday, either, as home-field advantage is present in all three games. In fact, Saturday may be the game that gives the Phillies the largest advantage, as "home" teams performed better by a statistically significant margin in the second game of a three-game series. After reading these three articles by Russell Carleton, I inquired if this finding regarding the second game of the series has to do with sleep cycles. He confirmed that he had this thought as well, which provides a solid psychological basis for a peculiar result that many wrote off as a statistical anomaly.
The reality is that short of maybe some first-inning adjustments and a smidge of designated hitter-related advantage, this is a home series for the Phillies and a road series for the Blue Jays. Expect the Phillies to have an eight percent increase in their probability of winning each game, versus if these games were played in Toronto. That is roughly similar to taking an ace away from one team and giving it to the other team. But I guess these teams have already done that, too.