Notice: Trying to get property 'display_name' of non-object in /var/www/html/wp-content/plugins/wordpress-seo/src/generators/schema/article.php on line 52
keyboard_arrow_uptop

In every sport and at every level, the home team wins more games than the visiting team. While this is true in baseball, it's less the case than in other sports. Throughout baseball history, the home team has won approximately 54 percent of the games played. Nearly every aspect of the game has changed drastically over the last century, but home-field advantage has barely changed at all. Consider the home-field advantage in each decade since 1901:


Decade    Home Win%
1901-09   .533
1910-19   .540
1920-29   .543
1930-39   .553
1940-49   .544
1950-59   .539
1960-69   .540
1970-79   .538
1980-89   .541
1990-99   .535
2000-09   .542

Although small decreases in home-field advantage have occurred at times, any incidence still represented a very small change relative to the large changes in nearly every other baseball statistic over the last century.

I will argue that this should surprise us as analysts more than it does. Nearly every study of psychology with respect to baseball has come up revealing either small effects or no effect. We all know that players are human, but the numbers do not seem to indicate many obvious psychological aspects. Hundreds of researchers have tried to discover clutch hitting, but few have found any evidence of its being a repeatable skill. Researchers have tried to look at synergy effects and if certain players increase a team’s chance of winning, and none have found any. Some have attempted to look for the impact of veterans on playoff races, and have not found significant effects there either. We have attempted all kinds of ways to splice the data to reveal a large psychological effect within baseball to show that baseball players don’t behave like statistical models, and there seems to be little evidence of any strong, detectable effects, even if we know they exist and occasionally can discover smaller ones.

The mantra that we go by as analysts is that talent trumps everything, and that most of the stories of heroism and mental fortitude are narratives written by the winners. However, home-field advantage is perhaps the most obvious area where we see something resembling a psychological effect, or at least an effect that is not captured by our typical models of baseball players and ballgames. It is clear that something about being the home team trumps talent in a way that is mathematically equivalent to benching an average player on the road team.

The reason why home teams have an advantage is less clear. Do they feel more at home and thus perform better? Does the crowd excite the home team or distract the road team? Do they know how the stadium plays better? Are they simply more comfortable when at home? All of these explanations seem possible, and we could give a number of plausible explanations. Finding out why the home team wins more often could open the doors to measuring at least one area where psychology has an impact in baseball. Everybody knows that baseball players are not computers, let alone Strat cards. The mainstream media certainly dramatizes the purported impact of psychological effects on players, but there must clearly be some effect if the home team wins more often.

This article will be the first part of a series of articles discussing home-field advantage. This series will ask what, when, where, when, and how, and hopefully will figure out why home teams win more often. The first step is to figure out what exactly it is that home teams are doing better. If we can look at which statistics show the biggest home-field advantages, we can figure out what exactly the home teams are doing to their opponents to consistently win more often than they lose.

Innings

Home-field advantage may come from the effects of comfort and familiarity, or it may come from the psychological effect of the crowd support. If the effects primarily come from comfort and familiarity, one would expect that home teams would have an even more distinct advantage in early innings. If the effects are primarily psychological, one would expect that home teams would have an even more distinct advantage in late innings when the crowd may put them on top.

The percent difference in runs scored by the home team as compared with the away team by inning is listed below. I ignored the ninth and any extra innings, because the home team stops scoring when they take the lead in extra innings or in the ninth inning, and the home team frequently does not bat in the ninth inning at all.


Inning  % of More
        Runs at Home
  1      16.2%
  2       9.3%
  3      10.1%
  4       6.0%
  5       7.8%
  6       8.1%
  7       8.7%
  8       6.5%

The home teams are gaining a lot of their advantage in the first inning, and enjoy a greater home-field advantage in the second and third innings than in the fourth through eighth innings. This implies that home-field advantage is primarily about comfort and adjusting to surroundings. Consider this as we discuss individual statistics below.

The Four True Outcomes

There are multiple people who play a role in the outcome of a plate appearance. First, the pitcher throws to the batter, and the batter either does or does not swing. If he does swing, he may or may not hit the ball. If he hits the ball, it may or may not be a fair ball in the field of play. It has been well established that pitchers have little control over what happens when the ball is in the field of play—pitchers primarily control walks, strikeouts, and home runs, and the rest is mostly a product of the defense and the hitter. In this section, I will discuss what happens when the ball is not in the field of play to see how much of home-field advantage is the responsibility of pitching. All data is from cumulative major league totals from 1998-2008, unless specified otherwise.

Strikeouts and Walks: Batters struck out in 16.35 percent of their plate appearances at home from 1998-2008, and in 17.30 percent of their plate appearances on the road. Batters received unintentional walks during 8.38 percent of their plate appearances while at home, and 7.80 percent of their plate appearances while on the road. Both of these numbers are statistically significant at the 99.9 percent level. This immediately indicates that a home-field advantage occurs within the strike zone. Somewhere between when the pitcher starts his windup and when the batter swings, home teams already gain an advantage over road teams.

Home Runs: Batters hit home runs in 3.22 percent of their at-bats at home, and in just 3.07 percent of their at-bats on the road. This is also significant at the 99.9 percent level. Home teams are able to hit for more power than road teams.

I selected strikeouts, walks, and home runs as the initial tests, because they indicate that the pitcher is involved. The significance of home runs, something that pitchers control less than walks and strikeouts, indicates the hitter must be doing something better at home as well. However, without any more information, it is difficult to tell if the walks and strikeouts are only the products of home effects on hitters. For instance, it may be that given enough time to prepare himself to pitch, the pitcher can overcome any disadvantage of being on the road, but the hitter has an advantage at reacting to the baseball when at home. In other words, it’s possible that travel or staying in a hotel dulls hitters’ reaction times, but do not hamper the pitcher. That’s why I now introduce the fourth true outcome—the neglected one on which fielders also have no effect.

Hit By Pitches: In every 1000 plate appearances, home hitters were hit by 9.41 pitches and road hitters were hit by 9.09 pitches. This is a statistically significant difference (p=.016). While some hitters may try to dodge pitches out of the strike zone more often, very few players can actually lean into the ball better. The fact that hitters at home are hit more than hitters on the road indicates that some of the home-field advantage is definitely on the pitcher. Getting hit by pitches is a persistent skill for batters, but it primarily derives from hugging the plate and not dodging inside pitches. Unless hitters hug the plate much more at home—and that this effect somehow dwarfs the extra reaction time at home to dodge pitches—chances are that this effect comes down to pitchers’ aim being worse on the road. While some of this may be umpiring making close calls as hit by pitches or more consistently awarding the base to a home batsman instead of ruling he made no effort to avoid the pitch, the 3.5 percent increase in hit by pitches is probably representative of a real, home effect-generated difference.

Balls in Play

When a ball is hit into play, defense becomes a factor. Some success on balls in play is the effect of how well the hitter hit the ball, but much of it is a product of how the defense plays as well. Looking at balls in play will be the first clue as to how defense plays a role home-field advantage.

Batting Average on Balls in Play: Home teams had a .301 BABIP, and road teams had a .295 BABIP during 1998-2008. Furthermore, when you factor in errors, home teams reach base a .316 clip on balls in play (or fail to reach base on 68.4 percent of balls in play), and road teams do so at a .310 clip (and thus fail to reach on 69.0 percent of balls in play). When you factor in double plays, defenses at home are able to record .720 outs per ball in play, and defenses on the road are only able to record .713 outs per ball in play. All of these differences are statistically significant at the 99.9 percent level. It is clear that batters are able to reach safely at home more often. It is not clear how much of the difference in outs on balls in play is hitting and how much is fielding, so it is useful to look at some indicators of defense to provide some clues.

Double Plays: Hitters ground into double plays in 2.92 percent of their balls in play at home, and 2.94% of their balls in play on the road. This is not a statistically significant difference. Initially, that sounds like the difference is hitting, until you realize that home teams have higher OBP than road teams (.342 vs. .330), so they have more opportunities to bat with a man on first base. When I approximate opportunities by the number of singles, walks, hit by pitches, and times reached on error that home and away teams have, the difference suddenly becomes statistically significant. The ratio of double plays to the sum of singles, walks, hit by pitches, and reach on errors at home is 7.62 percent, and the ratio on the road is 7.96 percent. Although some of this may be the hitters’ fault for swinging at bad pitches or making weak contact, this does seem to imply that defense is better at home. As some of this may be umpires calling away-team baserunners out on double plays more often, it's useful to check on other aspects of fielding too.

Reaching on Errors: Official scorers are often accused of giving the home team credit for hits more often, while recording road players’ close calls as errors. This may or may not be true, but it is highly doubtful that the reverse is true. Therefore, the fact that home teams reach on error during 1.46 percent of their balls in play and road teams do on 1.42 percent of their balls in play implies that players on the road are fielding more poorly. The ratio of infield hits to the sum of infield hits and times on reached on error is 70.3 percent for home teams and 69.8 percent for road teams. This is also not a statistically significant difference, which further suggests that official scorers are probably not all that biased. To me, this suggests that chances are that defenses are worse on the road.

Doubles and Triples: Another potential source of home-field advantage is that outfielders at home know their parks better—they know how the ball bounces off the wall, etc. Many triples are the result of balls that get away from outfielders. Therefore, it should not be surprising that the rate of triples per ball in play is 7.59 percent for home teams and only 6.44 percent for road teams. That is a huge and statistically significant difference. In fact, the rate of doubles on balls in play for home teams is 6.81 percent, and for away teams is 6.74 percent, which is not even a statistically significant difference (though it is weakly significant). The reason is that so many extra-base hits get away from outfielders and turn into triples; the ratio of doubles to triples is 9.0 at home, and 10.5 on the road.

Shutouts

Pitchers pitch complete=game shutouts far more often at home than on the road. Per every 100 games started, the hometown hurler will throw 1.68 complete-game shutouts and the road team hurler will throw 1.18. Even though home teams only win 54 percent of games, home teams record 59 percent of the complete-game shutouts.

Unearned Runs

It may seem like much of this could still be psychological, but I believe that a large portion of it is not. Some of the reasons for that will be clearer in subsequent articles in this series, but one indicator will be the look at unearned runs. If psychology is playing a significant role, then chances are that it will cause pitchers to be more prone to frustration and let problems escalate. One way to get some insight about this is to check the ratio of unearned runs to the times that hitters reach base on errors. If pitchers are less likely to be collected on the road, they will probably let more of these runners score. In fact, the ratio for home teams is 0.98 and 0.99 for road teams. This is a very small difference. In fact, from 1998-2008, the home team had a higher ratio than the road team six times out of eleven years. It does not seem like psychology is playing a role—road pitchers are not allowing crises to escalate any more; they are just finding themselves in more crises.

Stolen-Base Percentage

Home teams are more successful in their stolen-base attempts, nabbing bags successfully in 71.2 percent of their attempts, while road teams are successful in 69.4 percent of theirs. This is a strongly statistically significant difference. Whether that is due to catchers reacting slower, baserunners running faster, or umpires giving close calls the home crowd is tough to know. For now, it seems that there is a clear tendency to steal successfully while at home.

Batted Balls

A little bit about home-field advantage can be inferred from batted-ball statistics. Looking at data from 2005-2008, we can see that home teams hit more line drives (19.0 vs. 18.6 percent) and fewer ground balls (45.1 vs. 45.5 percent) than away teams; they hit about the same number of fly balls. The differences in line drives and ground balls are both statistically significant. As ground-ball rate is something that pitchers control strongly, this also suggests not simply that pitchers enjoy this effect of home-field advantage, but that they affect it. The batter also plays a large role as well; this is clear from the statistically significant difference in popups per fly ball at home vs. away (21.8 vs. 22.2 percent), as infield-fly rate is something that batters affect more than pitchers do.

Hitters do a little bit better on each type of batted ball as well. Look at the slash stats for each type of batted ball:


Batted Ball Type        Home              Away
Ground balls       .248/.248/ .270   .239/.239/.259
Fly balls          .224/.218/ .597   .218/.213/.578
Line Drives        .733/.731/1.105   .732/.729/.997

It is quite clear that home-field advantage affects each type of batted ball. It does not seem to affect the batting average on line drives very much, but it certainly seems to affect the ISO on them. There are more triples than doubles, and there are a good deal more home runs as well (2.5 vs. 2.0 percent). The home-run rate on fly balls is also a bit higher in for home teams (9.2 vs. 8.9 percent), and the rate of triples on fly balls is higher too (1.13 vs. 0.95 percent). Ground balls find their way through holes more, leading to more singles (22.6 percent vs. 21.9 percent of ground balls), more doubles (2.02 percent vs. 1.89 percent), and more triples (.112 percent vs. .086 percent). The relatively larger jumps in doubles and triples on ground balls indicate that a lot of home-field advantage is knowing how to play the bounces in one’s home stadium.

Putting Things Together and Moving Forward

It appears that home teams do pretty much everything better than road teams. They hit the ball more, they hit the ball harder, they throw more strikeouts, and they surrender fewer walks. Home teams prevent triples more frequently, and record double plays more frequently. They record more outs on balls in play, and they make fewer errors.

The extra home-field advantage in the first three innings strongly suggests comfort and familiarity are significant factors in home-field advantage. The large effects within the strike zone indicate that the mound and the batter’s eye are things that pitchers and hitters are more comfortable with at home. The extra triples surrendered on the road strongly suggest an impact of stadium familiarity in home-field advantage. The extra stolen bases indicate reaction time may also be playing a role. All of these numbers together indicate a large significance of mental aspects that are not quite emotional, but affect things like eyesight, reaction time, and learning about the home turf. That pretty much answers the question of what home teams are doing better.

Next week, I will begin to answer the question of who has a home field advantage, and I think that the answer may surprise you. Now that we know what home teams are doing better, our next stop is to check differences between teams in terms of how much better they perform at home. After that, we will begin to discuss the where, when, and even broach the questions of how and why home-field advantage exists.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
formersd
8/11
One item that I know is significant in basketball is the impact of home field advantage on officiating. I suspect the same thing happens in baseball. If the home team gets a few more favorable ball/strike calls, that would be quite significant given the impact the count can have on expected batter/pitcher performance.
briankopec
8/11
Has anyone taken the Questec data and tried to figure out if this is true? A quick 2 second google search was fruitless.
swartzm
8/11
I'm sure that umpiring has some effect, but it wouldn't make a team surrender more of their extra base hits on balls in play as triples, for example. Officiating may certainly play some role though.
briankopec
8/11
Some possibilities that have nothing to do with psychology:

1. GM's select players best able to take advantage of their home environment.

2. The effect of twilight and shadows on the pitcher/batter confrontation. (The effect should lessen as the game goes on with the advantage going to the team hitting in the bottom of every inning.)

3. Perhaps I missed this being discussed....but obviously home teams win more often because they have the strategic advantage of hitting last.
bsolow
8/11
With respect to criticism 1, it seems a stretch to suggest that environmental factors are responsible for four-true-outcomes results. I suppose that it could explain some portion of home runs (Yankees maximize lefties to take advantage of RF porch), and through a secondary effect, explain some portion of hit-by-pitches, but it seems a stretch to say that some players are uniquely more capable of walking/not striking out in their home park.

2 -- I don't know enough about sunset times, but isn't it possible that the twilight/shadows doesn't take effect until during the game, thereby making the effect random (and likely equal between the two teams)? I will say, though, more experience at home with your particular backdrop may make it easier to deal with the shadows at your home park even if the time spent with shadows is equal across teams.

With regard to #3 -- on the other hand, the second table indicates that home teams also score more runs in each inning considered (excluding the ninth which is skewed by home teams not batting as often/the inning being truncated by a walk-off anything). Therefore, even if you remove the strategic advantage of hitting last, home teams will outperform, unless you're suggesting that there's some advantage to hitting in the bottom of the 2nd instead of the top of the 2nd.
brucegilsen
8/16
A seminal little study by Bill James in the 1983 abstract compared Mets control pitchers (less than 5k/9 inings) with power pitchers (more than 5k/9 innings) from 1980-1982. Every power pitcher had a better home era, all but one by at least 0.68. Power pitchers home vs road era was 3.48 / 4.46 whereas control pitchers home vs road era was 3.89 / 4.07.

His point was that the Bamberger-ish control pitcher type, so successful in Baltimore, was wrong for Shea.

Clearly, when Frank Cashen took over, he went for high fastball strikeout pitchers, epitomized by Sid Fernandez, who had huge home/road splits as a Met.

Understanding your park and building a team to suit would seem to be able to lead to a home park advantage.
sbrousc
8/11
It would be interesting to see how well this effect holds up during the 2nd and 3rd games vs. the first game of an away series, to determine whether the road team is able to adjust. Although I wouldn't want to try to compile that data set...
swartzm
8/11
That's coming up! I have some initial results and they blew my mind-- that's all I'll say for now!
briankopec
8/12
I'd also like to know how home teams fair in the first game of a homestand. I theorize that away teams do better as the series goes on due to the rigors of travel. But if that is true, then you should see a difference when the home team is fresh off the road as well.
baserip4
8/12
My thoughts are along these same lines: how well do teams fare the day after they travel from one city to another (especially pitchers)? What about the distance of travel? Time zones? Especially with my first question, might it make sense to send your starting pitcher to a new city a day or two ahead of time? Great start to this series and I can't wait for more.
bravejason
8/13
Many years I've heard on radio someone talking about a study they did on the impact of travel on team performance. Sadly, I don't remember which direction the caused the team to play worse than expected. They found either that teams traveling west to east fared worse than expected or that teams traveling east to west fared worse than expected.
hangman
8/14
I'd also like to know how long read trips and long home stands skew the numbers. The home team always slept in their own beds last night, but the road team may have been on the road for two weeks.
morillos
8/11
Interesting stuff. A couple of questions that may be jumping the gun on the future columns.

1. The strikeout/walk data make me wonder if the home crowd may have a psychological impact on the umps. Do umps favor the home team, ever so slightly, in defining the strike zone? Could this then have a follow-on effect on all the batting data?

2. Travel wearininess. Not a psychological or stadium-comfort issue, but a real possibility for inducing a small but across the board decline in player effectiveness. I, for one, sleep like crap away from my own bed, not that I'm an athlete.

3. The big one. How much of the difference in win rate for home teams is simply the result of the advantage of batting last? Given two identical teams, the home team will win a certain amount more simply because they have more margin for error in late and extra innings. Gave up a run in the top of the tenth? Score two in the bottom and you're still good. Give up that solo dinger in the bottom of the frame, walk off. Run a playoff odds report-style million game simulation between two identical teams and see how often the home team wins. If home teams win 53% of the time just because of that factor, and 54% of the time overall, all the stats are only accounting for a tiny part of the advantage.
swartzm
8/11
The strike zone stuff certainly might be playing a role, but that effect seems pretty minor compared to some of the other stuff going on.

The travel thing is interesting, and I've done some initial work on it. The results are very, very interesting...more on that in another article.

The batting last thing only could have a minimal effect. The extra run scoring came in the first inning more than any other inning. Home teams win 54% of games, but only 52% of games that are tied going into the 9th inning and 52% of extra inning games. Chances are that a lot of home field advantage could be this strategy, but maybe not so much.
rowenbell
8/12
Matt, I think you're under-estimating the value of morillos' 3rd point. There is a fundamental structural advantage to being the home team. If both teams were of absolutely equal quality, the home team would still be expected to win >50% of games.

Tactically, the road team essentially always needs to play to maximize its number of runs scored, since it can't know how many runs it needs to win. Whereas in close late-inning situations, the home team knows how many runs it needs to win and can vary its tactics accordingly in order to maximize its probability of winning, even if doing so varies from the run-maximizing tactic.

Quantifying this theoretical issue is the first & fundamental question that needs to be answered when addressing HFA. Only after understanding the theoretical baseline HFA does it make sense to explore the ground you're exploring, in order to explain why the observed HFA differs from the theoretical baseline.
swartzm
8/12
I don't deny that it is an effect. If it weren't than half of the HFA would not still be left in a time game going into the 9th. But the home team outscores the road team by 16% in the first inning. That's clearly not tactical alone.
rowenbell
8/12
Agreed. Still, I'd like to measure the cart before putting it before the horse. (Or something like that.)
eliyahu
8/12
I wouldn't dismiss the structural advantage that easily. Baseball is the only sport where the home team has this advantage -- and it doesn't necessarily play out only in the ninth inning. The old adage of "playing for the tie on the road and the win at home" makes some sense and it impacts the way the game is played by both teams in the seventh, eighth and ninth innings (e.g. pitching changes, playing for one run, etc.) Purely a hypothesis, but I'll be stunned if this isn't the primary determinant in the home field advantage.
bsolow
8/12
The question is how you define home field advantage. It may have an effect in the team's winning percentage, but I think Matt's point in this article was that home teams perform better at home on the micro level. Given that we know teams play better when they're at home, we also know that the structural advantage isn't the sole determinant of home field advantage.

Additionally, I don't think the uniqueness of baseball having this structural feature is an argument in favor of it being a determinant of home field advantage. As Matt documents elsewhere in the comments (I believe it's not actually in the article), MLB has the lowest home field advantage of any of the 4 major sports. Considering what we know about strategy, that common 1-run strategies like bunting a runner over often reduce not just the expected number of runs but also the probability of scoring, this structural feature may cause managers to behave in a way that they grew up being told was optimal, but in fact is hurting the team.
swartzm
8/12
There certainly is a small element of structural advantage, but it cannot be the sole or primary determinant. Consider the following list of Home Win Probabilities by inning when the score is tied and there are no outs and no one on in the top of the inning:

1st: .540
2nd: .533
3rd: .533
4th: .525
5th: .529
6th: .522
7th: .520
8th: .521
9th: .522
XTR: .520

If you're hypothesis were true, this number would be getting bigger as the game was tight closer. Instead, getting to a point where the home team has this one-run strategy stuff in front of them should be a goal of a road team as their win percentage will increase from 46% to 48% if they can do so.
rowenbell
8/12
No, I don't think this HWP (Home Win Probability) data is inconsistent with the structural HFA hypothesis at all.

My interpretation of your data is that the structural HFA represents roughly half of the total HFA (52% versus 54%).

If the HFA were entirely structural, then I think we would expect to see a flat pattern of HWP by inning. This is because the HFA wouldn't really become relevant until the 9th inning, or thereabouts.

A declining pattern of HWP by inning is evidence that there is a non-structural component to HFA. The more innings that remain in the game, the more opportunities the home team has to exploit the non-structural aspects of its home-field advantage. Thus, the more innings that elapse with the game still tied, the fewer opportunities the home team has to get rescued by the non-structural HFA.
swartzm
8/12
I think the structural advantage comes into play in the 9th inning and is not the primary factor, but it sounds like you agree with me that most of HFA is probably not structural at all. It seems like whatever causes the 1st-3rd innings is the primary determinant and that the structural aspect is less than half. Based on your post, I think we pretty much agree.
rowenbell
8/12
Yes, but....had I been writing an article on this subject, I would have started by: acknowledging the structural aspect of HFA; attempting to quantify it; using the HWP data to argue that in addition to the structural aspect, there must also be a non-structural aspect to HFA (which honestly was news to me, I'd always assumed it was entirely structural); and -- then and only then -- going down the path that you're actually going down, namely trying to explain what the non-structural aspects of HFA are and why they arise.

As such, by this point my criticism is more pedagogical in nature than substantive. I'm looking forward to the remainder of the articles.

Thanks for engaging in the dialogue.
bravejason
8/13
I agree with Rowen Bell (5629) comment "...my criticism is more pedagogical in nature than substantive." with respect to not addressing the structural aspect of home field advantage in the article.

What you have is good, but I think it would have been better to include a comment about the structural aspect and then guide the reader to the point you really wanted to make. I spent the entire article wondering when you were going to come around to talking about the "obvious" answer to the presence of home field advantage.
ssimon
8/11
These sound very similar to the numbers one would expect if the home team got to play with a slightly smaller strike zone when batting as compared to the road team. This was an excellent article and I look forward to the rest in the series.
Ragnhild
8/11
I have always wondered if the strength of home field advantage is correlated with attendance...does a packed stadium give more of an edge to the home team than an empty one does in a measurable way?

Sounds like that may be coming in future installations, which I really look forward to!
dreaming
8/11
if comfort and familarity are an advantage for home teams, would the visiting teams performance improve in the 2nd or 3rd game of a series, or in later series in the same season?
bsolow
8/11
To piggyback off this idea, would players who play in a division for a long time have more success at divisional opponents' stadiums than at other stadiums?
swartzm
8/11
Both of these things are on the way later in the series.
bjfrankfort
8/11
What a great article.

With regard to comfort and familiarity as a component of the home-field advantage, is it possible to compare data from the first game a series to the subsequent games in a series? The advantage seen in the first three innings may be less in subsequent games. Along the same lines, what is the home winning percentage for the first game of a series compared to subsequent games in a series? If familiarity is a key component, you might find that home winning percentage is highest for game 1 and lower for the subsequent games.

Keep the series coming!
marioreturns66
8/11
Great stuff Matt. One thing to keep in mind: the HFA is actually much smaller in baseball than it is in the NBA or NFL, even though both are played on homogeneous surfaces. Both are anecdotally more momentum/crowd/emotion driven than baseball, and it's funny to think that pure psychology (which we tend to crap on) could have such a much bigger impact than actual tangible differences in the teams' home ballparks.
swartzm
8/11
Thanks. I looked at the NBA/NFL/NHL for recent years and got 60%, 58%, and 55%. I thought I remembered it being higher in the NFL. I guess in the NFL, there is the 3rd down effect where people scream and make it tough to hear. I guess that's a more obvious effect. I suppose it could also be that baseball is won by the better team less often-- the best odds you ever see for a baseball game are 70%, but it seems like some football games make it close to 90%. Maybe there is something at play there, but I'd need to think more about it. The differences in HFA in different sports certainly are something I would love to know more about.
markpadden
8/12
Yeah, the raw HFA numbers are misleading if you try to compare across sports. In the NBA, e.g., the better team wins much more frequently than it does in MLB. This is simply due to the way points/runs are scored and their frequency.

The interesting thing will be to isolate travel effects in MLB, since it is the only sport offering a decent sample of road games played in same city as previous game. To my knowledge, no one has looked at this analytically yet, so I am looking fwd. to the future installments.

I hope you will compare HFA not only vs. game number of the series, but also wrt number of days rest. Will be interesting to see if unrested road teams in game 1 fare more poorly that rested road teams in game 1, e.g.
jmercan
8/11
Great stuff. To me, a lot of the advantage can be traced to the pitcher's mound. Listening to pitchers, each mound is unique (whether it is pitch, the kind of dirt, whatever) and also differs from the bullpen mounds on which the starters warm-up. It seems reasonable that the difficulty in locating and commanding pitches stems from the visiting team's starter being unfamiliar and/or uncomfortable with the mound.

I wonder if the differences are less pronounced for pitchers on a new team returning to their previous team's home park?
swartzm
8/12
That would definitely be interesting to test, but it seems kind of tricky. It would be tough to separate the effect of knowing the hitters better (or them knowing you better) from knowing the mound better. There also could be some emotional aspect to it too. It's definitely a good question, but tough to get at empirically in such a way to determine cause and effect.
hotstatrat
8/11
I hope you see what the relativity of home field advantages are to attendence, because my theory is that the roar of the crowd is a very significant factor. You might need to do this on a per stadium basis, because not all crowds cheer as much (Toronto was silent the last game I went to) and not all stadiums lend themselves to equal intensity (thinking of Fenway and Metrodome vs. Astrodome, etc.).

Note that home field advantage is stronger in sports with enclosed more completely surrounding arenas such in the NBA and NHL. Yet other than some grandfathered ice rinks, the playing fields in these sports are all identical.
wonkothesane1
8/11
Isn't there a rules advantage with home field? Walk Offs? 8 1/2 inning game strategy? Maybe I'm missing something, but it seems to me the rules are somewhat designed to give the home team a slight advantage, but the article doesn't address it.
marioreturns66
8/11
There really isn't much of an advantage there -- the only difference is that a road team up two in the top of the 9th isn't going to be very urgent to score more runs, whereas if they were batting in the bottom of the ninth, and their opponent had already scored 3 runs, they would obviously know they need to score to keep the game going. That's a very minimal difference, since most games are won in 8.5 innings anyway.
hyprvypr
8/11
Unquestionably players get more comfortable hitting and perhaps even fielding and pitching at home. Familiar environments, glares, backdrops, outfield angles, mound size(yes they vary from park to park), etc. all play a minute role on a home teams success. 4% sounds about right.
hotstatrat
8/11
I know "percentages" are per 100, but moving from 50% to 54%, I feel compelled to point out, is an improvement of 8% in percentage points. 4 is 8% of 50. If you compare the home team to the visiting team the percent of percentage points they are playing better is double. However, percentages, of course, are not real counting numbers, so I'm not so sure it makes sense to look at percent of percentage points in the first place.
swartzm
8/11
hyprvypr, I pretty much agree with your sentiment. I think the vast majority of home field advantage is exactly the things you mentioned and the research for the next few articles strengthens that case, I believe. Specifically, I think mound and batter's eye are huge effects, and playing the hops of the field really seem to be an effect too given the ground ball hit type differences. Thanks for the comment.
mikeyworm
8/11
Is it possible that some of these statistics are swayed by the fact that visitors face the home team's better relief pitchers more often since they are frequently ahead? For example, I am sure the home team's closer (and theoretical best reliever) pitches much more frequently than his visiting counterpart.
swartzm
8/11
I don't actually think the home team is more frequently ahead. The home team is more frequently ahead at the end of the inning, but in the middle of the inning, the road team is probably more frequently ahead-- certainly early in the game. For example, despite the clear tendency for pitchers to pitch worse, closers on the road get more saves. I would suspect that is even true if you ignore extra inning saves.
jgalt73
8/11
It would be interesting to see if familiarity with the ballpark by a frequent visitor (e.g., division rival) lessens homefield advantage.
tinseltown
8/12
Another idea for a study on familiarity that Bill James proposed is the differences between a player's first year at a new park versus subsequent years.
markpadden
8/12
I like the direction this is headed...

One thing to look at is whether team-specific HFA is correlated y2y (assuming no stadium change). I.e., is it more accurate to use a generic HFA for all teams, or to tailor it based on a team's historical HFA?

On a similar note, I hope you can extend the analysis at some point to look at individual player home/road splits and whether they have any predictive value (vs. using generic team home/road adjustments). I suspect past individual splits are mostly worthless for predicting future splits.
bucswin611
8/12
Matt is a top-three writer at BP already. Great stuff!
fgreenagel2
8/12
I'll tell Matt what I was told by a union guy during a summer job 11 years ago: "slow down, this is a union job...you'll make the rest of us look bad."
dsc250
8/12
Great article Matt. I've always thought batter's eye is a huge factor, and it seems like you're pointing to that being one of the major reasons.

Someone upthread suggested looking at pitchers who are with a team for a while and then leave -- do they come back and perform better at their old haunt? I'd say the same study with hitters would be worthwhile. Someone who played in a stadium for a long time and then switches to another intra-division team would be good to study. He'd have lots of at-bats in the old stadium, but as a player on the other team. Does he still perform better at his old home? If it's about the batter's eye, then I'd have to think he'd have some advantage still (although the mound and outfield wall for the home team would be a bit of a counterbalance).
GreenMonster49
8/12
I think the data for triples is off by a factor of 10. "[T]he rate of triples per ball in play is 7.59 percent for home teams and only 6.44 percent for road teams." From context, I think you mean 0.759% for home teams and 0.644% for road teams. That's still significant, but no longer huge.

I would love to see this broken out by park--I would expect that oddball stadiums like Fenway Park or Baker Bowl would help their home teams more than Generic Ball Park.
DavidHNix
8/12
1. Home team batters know the light, so they see better.

2. Home team pitchers know the mound, so the pitch better.

3. Home team fielders know the ground, so they catch better.

I hope the series will address whether the 54% home field advantage stays them same when both home and visitor are good teams (above .500).
markpadden
8/12
Stadium-specific effects are minimal. Travel and psychology are the primary drivers of HFA in any sport.
orindv
8/12
About the "first-inning effect" - any thought as to whether the starting pticher having to sit through his team's at-bat after pre-game warmup might be having an impact?

The increase for innings 2-3 would suggest it's not big, but it was the first thing that came to my mind when I saw how much larger the first inning number is than ever those two.

I guess this is also realted to the idea of pitchers "cooling off" during a long batting inning, which is a topic for another study.
MountainCat
8/12
Referring to "large psychological effects" in baseball; "there seems to be little evidence of any strong, detectable effects, even if we know they exist" -- Err, uh, and just how do we know that they exist?? It seems to me that there are all sorts of prosaic explanations for the home team advantage -- teams tailored to the park, umpiring, knowledge of the park as it relates to positioning players, etc. Why do we have to believe there must be some psychological extract-of-moonbeam at play.
swartzm
8/12
I did not say that psychological effects were the cause of home field advantage. I said that without a clear theory as to why it is so strong and so persistent, it seems silly to dismiss that it could be psychology. In the article, I went on to explain how it seems that knowing the mound, batter's eye, and the bounces of the field seem to be the primary reasons for home field advantage.

But when I say we know psychology affects baseball, that is not really debatable. The question is the magnitude, and I agree that it is likely quite small compared to the effects of talent. But you only need to watch some tape of Chuck Knoblauch trying to make throws from second base to know it's possible. I think it was Adam LaRoche who missed a play on defense because of ADHD a few years ago. The question isn't that it exists; just the magnitude of the effect, which I think seems even smaller in light of the findings in this article.
gtgator
8/12
Maybe I'm being obtuse, and maybe this is addressed in later parts, but I'm not really sure any conclusions can be drawn from looking at stats that favor winners when trying to determine why winners won.

For example, striking out more batters and walking fewer batters increases your chance at winning games. Home teams won more games. Ergo, it would be expected that they struck out more batters/walked fewer. But was striking out more batters and walking fewer batters the cause (in part) of home teams winning more games or simply the effect of examining a data pool in which home teams won 54% of the games?

Or put another way, is it 1) home teams won 54% BECAUSE (in part) they struck out more/walked less; or 2) home teams struck out more/walked less BECAUSE they won 54% of the games examined? Which is the cause and which is the effect?

Reading this article, I cannot tell if it has been clearly established why the view point is 1) (which appears to be my reading of the article). Because unless I've missed it, I don't see how 2) has been eliminated as a possible conclusion which simply supports the notion that these are the stats that win games - and not the reasons why home teams win 54% of the time.
swartzm
8/12
I'm not sure I follow what you're saying. The argument that you're making is the type of argument that makes sense when retroactively looking at games, and saying "oh, my team wins 65% of games that this guy hits a home run," completely missing that you're selecting categories based on the outcome. In other words, it's a mistake people make when selecting on the dependent variable and then trying to crank out independent variables as a cause.

What I did here is look at the subsets of home and away games. That's exhaustive, and does not include selecting on the dependent variable. When the road team walks a home team hitter, he is already a road team player.

The cause and effect cannot go the other way simply because the future cannot cause the past.

The argument you are making is useful when a third variable may be causing both the first and second variables to be correlated, but that does not make sense here. If it were bad pitching that caused your star slugger to hit more homeruns when facing him and caused your team to win more games when facing him, that would be a third variable (bad pitching) affecting the latter two (homeruns by slugger, wins by team). However, that does not hold up when it comes to home field advantage because there is no variable that causes one team to be at home-- just the schedule maker, who presumably doesn't assign location to games based on who seems more likely to walk and strike out hitters on a random day nine months later.
gtgator
8/12
No, that's not the argument/question I am raising. I guess I am trying to say/ask is that when looking at a data set that is biased (i.e. home teams winning 54% of the time), how have you eliminated the inherent bias for the underlying variables that support such a result (such as more Ks, less BBs allowed, more DP turned, etc.)?

You looked at variables that have correlation to winning games, correct? Teams that win generally score more runs than they allow. They do that with higher OBP, higher SLG, allowing fewer opposing runners on base, allowing fewer runners to score, etc. As a result, it would be expected that teams that won more games would have performed better at those variables.

Let me use an example - hypothetically let's say teams with the letter "A" in their names win 54% of their games (sample size of 200,000 games). If I examined variables that help teams win games, I would expect teams with the letter "A" in their names would: strike out more batters, walk fewer batters, turn more DPs, have more XBH, get on base more, play better defense and so on. Why? I have a biased data set (one group winning 54% of the games) and, as such, all the resulting variables that win games should also be biased in favor of that group, especially over a large sample size. It would have nothing to do with how their team name was spelled.

Now you may have done this and I missed it, but you can eliminate bias in the variables despite examining a biased data set. For example, how many batters did home teams K in their wins vs. their losses and how did that compare to how many batters road teams K in their wins vs. their losses? If there was an advantage to home teams in terms of striking out opposing batters, this is where it would show up.

But simply looking at cumulative stats for all games and then trying to base the difference in the home teams' stats being better than the road teams' stats on some HFA factor seems like a stretch when that result would be expected simply because you looked at data where home teams won more games than road teams (and striking out more batters is a good thing in trying to win games).
swartzm
8/12
Each team plays 81 home games and 81 road games. That would eliminate the bias you describe. Since the data set is automatically composed of every team, your concern is not an issue. Your description is not necessary, because each team is in each data set 891 times (81 games, 11 seasons).

If I were to look at how each individual team did at home versus on the road and average team, I would have exactly what I have above with the sum of all teams put together, but everything would be divided by 30. The significance would not change.
gtgator
8/12
I'm still not getting my question across. I'll try a mathematical example. If this fails, I give up.

First, let's see if we agree on one point - every other variable being the same, can we agree that striking out more batters increases you chances of winning a game? Reason - fewer BIP, so fewer hits, less chance for runners to advance, etc.

If so, let's say that in the history of baseball, winning teams strike out 18% of opposing batters while losing teams strike out 16% of batters. We'll keep the home team wins 54% of games.

If I examined home/road performance in games, and home teams struck out 17.08% of opposing batters while road teams struck out 16.92%, is this a home field advantage or simply a winning team advantage? One cannot conclude either way.

However, if home teams struck out 19% of batters in wins and 17% in losses while road teams struck out 17% in wins and 15% in losses, THEN it is much easier to conclude that home teams strike out more hitters. And since striking out more hitters is conducive to winning, there is a HFA that helps explain why home teams win more often (at least in this factor).

But in just using cumulative stats biased in the winners favor, I don't know that you can show why home teams win more games. Because if you subbed "winning team" for "home team", "losing team" for "road team" and "winners' advantage" for "home-field advantage" in your article, it would read logically - and it would not matter who was home or away in those games.
swartzm
8/12
Your example is doing what you think it is correcting. You are looking at results and looking back to determine cause. That's selection bias. In my example, the opposite occurs. I look at what home teams do, win or lose, and what away teams do, win or lose, and study effect. What you are saying is the equivalent of saying that you cannot look at pitchers with good W-L records' ERAs because "of course they must be good-- they won more games." Since the winning occurs after the fact, it's not a problem. If the average team strikes out more batters in home games and fewer batters in road games, that's a home field advantage. I understand what you are saying, but you are selecting on the dependent variable in your attempt to do a typical correction to fix selecting on the dependent variable, which I am not doing in this article.

When the home team pitcher strikes out the away team batter to start the game, he does not know whether he is going to win the game. However, he has increased the odds of doing so by 2%. If he is less likely to strike out the first hitter in the bottom of the 1st when he is on the road, that is a home field advantage.

Ultimately, I see what you're doing but it does not apply for two distinct data sets that every team is part of 891 times each.
gtgator
8/12
Because I believe winning teams strike out more hitters than losing teams, here is how I interpret part of your last comment:

"If the average team strikes out more batters in {games it wins} and fewer batters in {games it loses}, that's a {winning team} advantage.

When the {winning team} pitcher strikes out the {losing team} batter to start the game, he does not know whether he is going to win the game. However, he has increased the odds of doing so by 2%. If he is less likely to strike out the first hitter in the bottom of the 1st when he is on the {losing team}, that is a {winning team} advantage."

As such, it is irrlevant who is home/away because you are examining an event that favors winning teams.

Regardless, I've said my peace. If you believe you've addressed my question/concern (or don't feel my question/concern is applicable), that's fine. Like I said, I may be obtuse.
swartzm
8/12
I think that you are misinterpreting statistics. What you are saying does not apply. I apologize that I am not explaining why better. Read Ben Solow's comment below, which I think is a better explanation than mine. The key to understand is that each team is a home team 50% and an away team 50% of the time. Whether they have won is not exogenously based on whether they are home. Something needs to happen between the first and last inning to make that happen. The cause of home field advantage can be revealed by the composition of plate appearances that generate the 54% winning percentage.
bsolow
8/12
The problem with this argument is that correlation is not the appropriate concept to be considering, but rather causation. Additionally, the flaw in your analogy is that while teams with the letter "A" in their names are exhaustive with their adjoint group, there's no overlap there, which allows for the possibility that a third variable (the letter A or something about the players on those teams) is causing both micro and macro performance to be higher. Matt's analysis works around this fact because all teams play equal numbers of home and road games, so the data can't be skewed by team-specific factors (i.e. if the Yankees played every game at home and the Royals every game on the road). Furthermore, there are equal number of home teams and road teams at every park in the majors, so you can't have park effects driving any of the causation either. What he has, in essence, is a natural experiment where you have identical samples (not even random samples that you can treat identically, but actual identical samples) and an easily observable treatment to those samples.

What it boils down to is that, yes, knowing that a team won 54% of its games at home would make you expect that it would perform better based on correlation, but that doesn't bias your results based on causation. We know that winning games cannot CAUSE your team to perform better in those games, because wins are a function of team performance, and not the reverse. We also know, for sure, that the two are correlated because of that functional relationship. We can rule out wins in a given game as an input to the function generating how your team performs in that game, but location of the game is an exogenous input (like talent, skill, etc.) into the performance function. The argument for the structural effects of batting second suggest that it's also an input directly into the wins function, which may present some problems in estimating the causal relationship, but doesn't mean that the causal relationship doesn't exist.

Consider this example: I'm from Iowa, which produces a lot of corn. Based on our corn exports, you would expect the state (correlation) to have fertile soil, which in fact it does. But that has no relevance to the question of whether or not local geologic factors cause the soil to be fertile, because corn exports don't increase soil quality.
Oleoay
8/13
First off, great stuff again Matt. I especially liked the increased chance of run scoring by inning and the increased triple frequency.

If your hypotheses are true, is it possible to see if the same effects are noticeable in all-star games? This might serve also as a control element for travel and crowd/psychology issues.

Second, there's a bit of a myth that groundskeepers tailor the field to suit their team, either by sculpting the foul lines for/against bunts, trimming the grass more or less depending on the defensive characteristoics of a team, and dampening the basepaths to prevent stolen bases. I imagine that might be hard to measure, but if that myth is true, it might be a factor to consider.

Other factors to consider include home team video rooms might be better organized and equipped, and with pregame coaching might prep home teams better. Also the home team might have access to camera angles, for positioning, pitch location, and sign stealing that visitors don't have. In any event, the home team advantage might be more than psychological or rules based..
Mariofan
8/13
Great work!