Advantage home team

[Ambermonk]

Originally Posted by Markus Heinsohn

Interesting thread OOTP does reflect the fact that hitters fare better at home and worse away. The question is if the reason is that pitchers are more comfortable with their mound at their home park, or hitters with the hitting background and vice-versa. It's probably a combination of the two, and OOTP reflects that...

Thanks for the info Markus, it is appreciated.

[Childress]

Originally Posted by Markus Heinsohn

The question is if the reason is that pitchers are more comfortable with their mound at their home park, or hitters with the hitting background and vice-versa. It's probably a combination of the two, and OOTP reflects that...

What's the difference? You get the same output whether you tweak the hitting or the pitching ratings in a balanced fashion. It becomes a zero sum adjustment. In it's present state OOTP reflects HFA reality pretty well.

But then there's the subject of the L/R split as applied to hitting production, which I also find a bit muted in the game. But that's for another thread....

[professor ape]

So after going through the 2007 data looking for the impact of the length of road trips it is pretty clear that the length of home stands and road trips don't account for the .540 winning percentage for home teams.

There are a number of potential factors driving the home filed advantage. Some are easier to quantify than others. Some of the first things to come to mind are:

- Advantage of batting last in the ninth so you know what you have to do strategy-wise to win the game. For example, if the game is tied going into the ninth then the road team wants to score at least one run but they can't assume that one run is enough. At the same time the home team goes into the bottom of the ninth of a tied game knowing that playing for one run wins the game.
- Fielding advantages of playing in the home park. This would be interesting to break down and I think that it is a significant factor. However, like most fielding questions, it would be challenging to measure this. I think that the differences in outfield fence configurations, wind patterns and the way turf effects bounces have a real impact on defensive performance between home and road teams.
- Hitters may have somewhat of an advantage batting where they have a familiar hitting background.
- The home crowd can provide that boost of support to provide confidence in the home team.

While all of these are real factors there is one area where there are some simple measurements to look at. Pitching. The chart below shows some core pitching stats from 2007.

	IP	K	BB	K/9	BB/9	K/BB	ERA
Home	744	563	265	6.8	3.2	2.1	4.29
Road	703	510	271	6.5	3.5	1.9	4.65

The most obvious is ERA where the home pitchers have an ERA that is 8% lower than road pitchers. The problem here is that this stat is a cumulation of the efforts of the teams. Side note, the expected pyth winning percentage of the home team based on the ERA split above is .533.

The two things that a pitcher has the most control over are strikeouts and walks. Home pitchers have a 4.3% higher K/9 rate than road pitchers, an 8.2% lower BB/9 rate and a 12.9% higher K/BB ratio. So why are there these differences?

As noted earlier, there could be some advantage for home hitters being familiar with the batting background. This could help them judge balls and strikes a bit better at home. However, I think that there is a more important factor, the pitcher's mound.

Every grounds crew shapes their mound a little differently meaning that no two are exactly alike. It doesn't take much to throw a pitcher's mechanics off enough to make a few strikes turn into balls. The fact that the home/road walk rates appear to be impacted more than the strikeout rates supports the idea that there is a home-mound advantage for pitchers. If the mound is a significant factor in a pitcher's control then you would expect to see walks jump up.

I would be intersted in doing a deeper analysis on the expected impact of just the walk and strikeout rate differences on run production. I have a feeling that these factors make up a good chunk of the home/road run production splits and therefore a significant portion of the home field advantage.

Thoughts?

[Ambermonk]

Originally Posted by professor ape

So after going through the 2007 data looking for the impact of the length of road trips it is pretty clear that the length of home stands and road trips don't account for the .540 winning percentage for home teams.

There are a number of potential factors driving the home filed advantage. Some are easier to quantify than others. Some of the first things to come to mind are:

- Advantage of batting last in the ninth so you know what you have to do strategy-wise to win the game. For example, if the game is tied going into the ninth then the road team wants to score at least one run but they can't assume that one run is enough. At the same time the home team goes into the bottom of the ninth of a tied game knowing that playing for one run wins the game.
- Fielding advantages of playing in the home park. This would be interesting to break down and I think that it is a significant factor. However, like most fielding questions, it would be challenging to measure this. I think that the differences in outfield fence configurations, wind patterns and the way turf effects bounces have a real impact on defensive performance between home and road teams.
- Hitters may have somewhat of an advantage batting where they have a familiar hitting background.
- The home crowd can provide that boost of support to provide confidence in the home team.

While all of these are real factors there is one area where there are some simple measurements to look at. Pitching. The chart below shows some core pitching stats from 2007.



The most obvious is ERA where the home pitchers have an ERA that is 8% lower than road pitchers. The problem here is that this stat is a cumulation of the efforts of the teams. Side note, the expected pyth winning percentage of the home team based on the ERA split above is .533.

The two things that a pitcher has the most control over are strikeouts and walks. Home pitchers have a 4.3% higher K/9 rate than road pitchers, an 8.2% lower BB/9 rate and a 12.9% higher K/BB ratio. So why are there these differences?

As noted earlier, there could be some advantage for home hitters being familiar with the batting background. This could help them judge balls and strikes a bit better at home. However, I think that there is a more important factor, the pitcher's mound.

Every grounds crew shapes their mound a little differently meaning that no two are exactly alike. It doesn't take much to throw a pitcher's mechanics off enough to make a few strikes turn into balls. The fact that the home/road walk rates appear to be impacted more than the strikeout rates supports the idea that there is a home-mound advantage for pitchers. If the mound is a significant factor in a pitcher's control then you would expect to see walks jump up.

I would be intersted in doing a deeper analysis on the expected impact of just the walk and strikeout rate differences on run production. I have a feeling that these factors make up a good chunk of the home/road run production splits and therefore a significant portion of the home field advantage.

Thoughts?

Useful ideas. Good effert!

[Childress]

Nobody's conjectured on the influence of umpiring/refereeing on home/away ratios. This might be a bigger factor than we realize!

[Ambermonk]

Originally Posted by Childress

Nobody's conjectured on the influence of umpiring/refereeing on home/away ratios. This might be a bigger factor than we realize!

Oh no! Maybe this will inspire Markus to put in some news items about umpires being fired for gambling!

[Childress]

Thoughts?

Interesting stuff. But I think with the pitching mound thing you're heading down a blind alley. It doesn't account for other pro sports, for example.

The real question is why the HFA is LESS significant than in other sports. I believe it's above 60% in most of them. Especially college football.

ESPN has a poll up asking which sport has the greatest home field advantage. NBA leads with 46%, NFL in 2nd at 41% and NHL and MLB are tied at 6%. Before this poll, all I heard from friends was that the NBA had by far the greatest home field advantage.

[professor ape]

Originally Posted by Childress

Nobody's conjectured on the influence of umpiring/refereeing on home/away ratios. This might be a bigger factor than we realize!

That is true. My guess is that the impact is there but not as significant as in other sports like basketball where the home-ref advantage appears to be very significant. Other sports have fouls and penalties which are very subjective and controlled by referees. Baseball does not have this factor. If there is a home-umpire advantage then I would expect it to appear on ball-strike calls especially in clutch situations when the crowd is likely to be fired up. It's only human nature to be impacted by a roaring crowd.

I would be curious if the home-road K/BB differences that I discussed in an earlier post are greater in late inning clutch situations than at other points in the game. You could take it a step further and see if attendance is a factor. I don't see 5000 apathetic Marlins fans having as much of an impact on an umps calls as 50,000 rabid Yankee fans.

[Le Grande Orange]

Originally Posted by professor ape

So after going through the 2007 data looking for the impact of the length of road trips it is pretty clear that the length of home stands and road trips don't account for the .540 winning percentage for home teams.

But one season of data is a rather small sample when it comes to home field advantage. If you look at it on a year-by-year basis you can seem some dramatic swings in terms of the overall percentage.

Repeat the analysis looking at ten or twenty seasons' worth of data.

Note that average length of scheduled home stands and road trips are much shorter now compared to earlier years. In 1915, the average of the longest scheduled home stands was 22.1 games (with the range being 17-27), while the average of the longest scheduled road trips was 21.6 games (15-28 was the range); in 1965, these averages were 14.3 and 14.3, respectively; in 2005, the averages were 11.7 and 11.5, respectively.

[professor ape]

Originally Posted by Le Grande Orange

But one season of data is a rather small sample when it comes to home field advantage. If you look at it on a year-by-year basis you can seem some dramatic swings in terms of the overall percentage.

Repeat the analysis looking at ten or twenty seasons' worth of data.

Note that average length of scheduled home stands and road trips are much shorter now compared to earlier years. In 1915, the average of the longest scheduled home stands was 22.1 games, while the average of the longest scheduled road trip 21.6 games; in 1965, these averages were 14.3 and 14.3, respectively; in 2005, the averages were 11.7 and 11.5, respectively.

Agreed about one year of data not being enough to make conclusions. Unfortunately the data based on length of homestand is not a basic filter on the stat sites so it took a couple of hours to sort through the data during a day off from work. I may do another year or two sometime over the next couple of weeks because this is an interesting topic but I won't be able to crank out 10-20 years any time soon.

Any suggestions on sites to copy data other than ESPN or similar sites?

[Cryomaniac]

Originally Posted by Ambermonk

Oh no! Maybe this will inspire Markus to put in some news items about umpires being fired for gambling!

Haha, why not ?

[Le Grande Orange]

Originally Posted by professor ape

Agreed about one year of data not being enough to make conclusions. Unfortunately the data based on length of homestand is not a basic filter on the stat sites so it took a couple of hours to sort through the data during a day off from work.

Yeah, I figured getting the data to analyze would be a signicant issue.

What about using Retrosheet's game logs? Those have the game-by-game results for each regular season.

[Matt Uk]

I think HFA is a mix of many things, some intangible. For instance in Football (Soccer) there is without doubt a different mental approach from most away teams than when they play at home. With the draw (tie) being an outcome in football many away teams approach a game with the intention of trying not to lose it, rather than trying to win the game. Also I would think the crowd influence both on the team and on occasions the pressure on the referee has an impact on decisions, although again marginal. Across all the European leagues away wins account for around 25% of results.

The interesting thing on the previous chart on the 2007 MLB road trips is that the first three games fell into the .4's percentage wise. The next three, or more likely the second series on the trip, came out higher. It does appear that travel, home comfort, has some effect but how you could ever get a true measure of it I don't know.

It's certainly a fascinating subject and lots of elements come into play. The mindset of players, familiarity of the home stadium, home support and many more.

[gmo]

As per the above I went into the Retrosheet game logs and examined the results of games ending after X innings and winning percentage after Z games played consecutively at home or away.

I looked at two periods: 1920-2007 and 1972-2007. Results are very similar for the two periods.

First overall home field advantage in all games separated by length in innings. Home teams average that 54% winning percentage overall, but only ~52% in extra inning games. I believe this is consistent with home teams performing better and batting last being of much less significance.

The HFA is about half a percent less for 1972-2007 (53.88%) compared to 1920-1971 (54.35%). Guessing, I think that is probably statistically significant, which may merit speculation. But for me here it is all about 54%.

1920-2007                            |1972-2007
                                     |
Game      Home   Away                |Game      Home   Away
ended     wins   wins                |ended     wins   wins
Inn  5:     71     98                |Inn  5:      5     27
Inn  6:    130    142                |Inn  6:     41     38
Inn  7:    159    137                |Inn  7:     37     38
Inn  8:    145    136                |Inn  8:     25     25
Inn  9:  57845  68767                |Inn  9:  32144  37813
Inn 10:   2765   3000                |Inn 10:   1470   1678
Inn 11:   1545   1686                |Inn 11:    850    906
Inn 12:    853    940                |Inn 12:    471    500
Inn 13:    475    497                |Inn 13:    260    266
Inn 14:    250    310                |Inn 14:    147    160
Inn 15:    140    145                |Inn 15:     79     69
Inn 16:     73     97                |Inn 16:     37     47
Inn 17:     43     46                |Inn 17:     28     23
Inn 18:     23     24                |Inn 18:     14     16
Inn 19+:    28     24                |Inn 19+:    15     14
 Total:  64545  76049 : HFA (0.5409) | Total:  35623  41620 : HFA (0.5388)
 upto9:  58350  69280 : HFA (0.5428) | upto9:  32252  37941 : HFA (0.5405)
 Extra:   6195   6769 : HFA (0.5221) | Extra:   3371   3679 : HFA (0.5218)

Next up homestands. No matter how many games into it, the home team winning percentage appears very close to that 54%. I did not run any statistics, but it seems abundantly obvious that it would be unlikely you could reject a null hypothesis that length of homestand affects home team winning percentage.

1920-2007                            |1972-2007
                                     |
Homestand   Home   Away              |Homestand   Home   Away
game #      wins   wins              |game #      wins   wins
home  1 :  10506   8614  (0.5495)    |home  1 :   6042   5138  (0.5404)
home  2 :  10034   8650  (0.5370)    |home  2 :   5963   5171  (0.5356)
home  3 :   9623   8189  (0.5403)    |home  3 :   5805   5022  (0.5362)
home  4 :   8431   7201  (0.5393)    |home  4 :   5067   4425  (0.5338)
home  5 :   7715   6707  (0.5349)    |home  5 :   4804   4109  (0.5390)
home  6 :   7181   6014  (0.5442)    |home  6 :   4486   3723  (0.5465)
home  7 :   5361   4673  (0.5343)    |home  7 :   3049   2648  (0.5352)
home  8 :   4191   3472  (0.5469)    |home  8 :   2163   1723  (0.5566)
home  9 :   3543   2958  (0.5450)    |home  9 :   1721   1444  (0.5438)
home 10 :   2747   2286  (0.5458)    |home 10 :   1113    931  (0.5445)
home 11 :   2062   1790  (0.5353)    |home 11 :    630    547  (0.5353)
home 12 :   1688   1408  (0.5452)    |home 12 :    391    357  (0.5227)
home 13 :   1348   1111  (0.5482)    |home 13 :    222    223  (0.4989)
home 14 :    986    873  (0.5304)    |home 14 :     86     96  (0.4725)
home 15 :    811    678  (0.5447)    |home 15 :     40     34  (0.5405)
home 16 :    640    576  (0.5263)    |home 16 :     18     14  (0.5625)
home 17 :    532    460  (0.5363)    |home 17 :     11     10  (0.5238)
home 18+:   1618   1441  (0.5289)    |home 18+:      9      8  (0.5294)

Now road trips. These appear quite similar to homestands. It sure looks like no matter how many games you are into the road trip the expected away team winning percentage is about 46%. Maybe the longest road trips from the longer 1920-2007 show some indication of being slightly less than that, but that looks pretty flimsy. Again without doing the statistics, I would figure the null hypothesis of length of road trip not mattering would not be rejected.

1920-2007                            |1972-2007
                                     |
Road trip   Away   Home              |Road trip   Away   Home
game #      wins   wins              |game #      wins   wins
away  1 :   8694  10420  (0.4548)    |away  1 :   5132   6046  (0.4591)
away  2 :   8531  10156  (0.4565)    |away  2 :   5131   5993  (0.4613)
away  3 :   8375   9386  (0.4715)    |away  3 :   5050   5723  (0.4688)
away  4 :   7200   8427  (0.4607)    |away  4 :   4388   5097  (0.4626)
away  5 :   6701   7695  (0.4655)    |away  5 :   4143   4786  (0.4640)
away  6 :   6045   7175  (0.4573)    |away  6 :   3723   4510  (0.4522)
away  7 :   4625   5473  (0.4580)    |away  7 :   2612   3080  (0.4589)
away  8 :   3520   4190  (0.4565)    |away  8 :   1787   2115  (0.4580)
away  9 :   2955   3629  (0.4488)    |away  9 :   1462   1726  (0.4586)
away 10 :   2346   2798  (0.4561)    |away 10 :    969   1150  (0.4573)
away 11 :   1759   2100  (0.4558)    |away 11 :    539    637  (0.4583)
away 12 :   1449   1635  (0.4698)    |away 12 :    334    384  (0.4652)
away 13 :   1092   1315  (0.4537)    |away 13 :    187    211  (0.4698)
away 14 :    883    976  (0.4750)    |away 14 :     86     91  (0.4859)
away 15 :    643    816  (0.4407)    |away 15 :     36     36  (0.5000)
away 16 :    527    668  (0.4410)    |away 16 :     18     15  (0.5455)
away 17 :    425    535  (0.4427)    |away 17 :      8      5  (0.6154)
away 18+:   1331   1623  (0.4506)    |away 18+:     18     15  (0.5455)

[gmo]

We can also look at runs scored from the Retrosheet game logs. I will only take innings 1-8 since from the 9th inning and beyond the home team often will not take all of the batting opportunities possible to them. Again we examine 1920-2007 and 1972-2007, which strongly resemble each other in their results.

First note that home teams score about 52.4% of the runs in innings 1-8. This percentage should not match the HFA winning percentage - for example, if home teams scored 75% of all runs they would win almost all games.

There is quite a bit of variation by inning in percentage of runs scored by the home team. The 1st inning especially and to a lesser degree the 3rd & 5th innings see home teams scoring a greater percentage of the runs compared to other innings. Those innings would seem to be those where the top of the order (better hitters) would more often bat. So perhaps better hitters have more pronounced home/away differences. Another explanation may be that away SPs struggle (relative to how they perform at home) early on, especially in the 1st inning.

1920-2007
         Away  (runs/    Home  (runs/     Percent of runs
         runs   inning)  runs   inning)   scored by home team
-------------------------------------------------
Inn  1:  73626 (0.5223)  86913 (0.6166)    54.14%
Inn  2:  59123 (0.4194)  65268 (0.4630)    52.47%
Inn  3:  66769 (0.4737)  74857 (0.5311)    52.86%
Inn  4:  68906 (0.4888)  74477 (0.5284)    51.94%
Inn  5:  67458 (0.4786)  74407 (0.5280)    52.45%
Inn  6:  70578 (0.5014)  76057 (0.5408)    51.87%
Inn  7:  67999 (0.4841)  73692 (0.5251)    52.01%
Inn  8:  67755 (0.4835)  72325 (0.5166)    51.63%
-------------------------------------------------
 1-8  : 542214          597996             52.45%

_______________________________________________________

1972-2007
         Away  (runs/    Home  (runs/     Percent of runs
         runs   inning)  runs   inning)   scored by home team
-------------------------------------------------
Inn  1:  40276 (0.5211)  47698 (0.6172)    54.22%
Inn  2:  33422 (0.4324)  36434 (0.4714)    52.16%
Inn  3:  37525 (0.4855)  41765 (0.5404)    52.67%
Inn  4:  38626 (0.4998)  41370 (0.5353)    51.72%
Inn  5:  37723 (0.4881)  41466 (0.5366)    52.36% 
Inn  6:  39076 (0.5058)  42168 (0.5462)    51.90%
Inn  7:  37395 (0.4846)  40287 (0.5224)    51.86%
Inn  8:  36580 (0.4746)  38988 (0.5060)    51.59%
-------------------------------------------------
 1-8  : 300623          330176             52.34%

[gmo]

Originally Posted by Markus Heinsohn

OOTP does reflect the fact that hitters fare better at home and worse away. The question is if the reason is that pitchers are more comfortable with their mound at their home park, or hitters with the hitting background and vice-versa. It's probably a combination of the two, and OOTP reflects that...

To me this is all that matters. We can speculate about why there is the difference, but for the game just simulating the results is the key. I am not sure how one may pull out any particular strong influence on the effect. It looks a lot like home players are just better to a certain degree pretty much uniformly across whatever circumstances.

I guess the last post above with the differences in runs scored home/away by inning points to something a little more going on, but generally it seems a small general "help" for home players and a small general "hurt" for away players can take care of the home field advantage. Back in OOTP6 the BB & SO home/away differences were smaller than in MLB at least in one test I ran, but considering the above quote I would figure that the home/away stat differences and thus HFA win% probably work out to be very similar to the historical MLB numbers.

[Curtis]

Originally Posted by Le Grande Orange

It would be interesting to see though if the length of a typical road trip or home stand has any effect on a team's performance. For example, do teams generally fair better on the first half of a 14-game road trip as compared to the second half.

professor ape and gmo seem to have covered these aspects. What would have interested me more was whether there was an effect caused by the first day a visiting team stays in a new city, or whether there was less effect for the first day if it followed a travel day.

In case I didn't make that clear, my guess would have been that visiting teams were more likely to lose the first game in a different stadium than subsequent games, and less likely to lose that first day if it followed a day off.

Based on the analyses performed by professor ape and gmo, it now looks to me as though there probably is no appreciable effect, though I would be grateful if someone could look into it.

Originally Posted by professor ape

You could take it a step further and see if attendance is a factor. I don't see 5000 apathetic Marlins fans having as much of an impact on an umps calls as 50,000 rabid Yankee fans.

You may be forgetting that those 5000 'apathetic' Marlins fans are actually rabid Mets fans.

If you do look at the attendance effect, don't forget that you'll need to factor out the correlation between team winning percentage and attendance. In doing so you might discover whether or not teams with higher winning percentages have higher (or lower) home field advantage, as was brought up above.

[DrArbiter]

Originally Posted by Kemp

But the questions are:
1. How much of that advantage is due to the fact that the home team bats last?

None, and it might even be an advantage to bat first, though it's so close to zero it's almost certainly a wash. See my recent paper in Journal of Quantitative Analysis in sports for a literature survey, data, and some theoretical calculations.

Interesting fact: the longer a game goes in terms of innings, the more likely the visiting team is to win. This isn't compelling evidence one way or the other -- it's probably just a selection bias, because, since there is a HFA, the fact that the game goes to extra innings suggests the visiting team is probably stronger. But, it's the opposite of what we'd expect to see if having last ups were decisive.