THE BOOK--Playing The Percentages In Baseball

Well, if you want conspiracy theories, maybe MLB has decided to take a cue from the NBA, where it seems to be generally accepted that the refs favor the hometeam. Perhaps umps have started squeezing the strike zone on the road team pitchers and expanding it for home team pitchers. That should be testable in theory, though I’d be really surprised if something like this was happening.

I have done some research on how HFA changed over time and also some research on how the age distribution has changed over time. Below are the HFAs for each decade starting with 1901-10

0.1042
0.074
0.0928
0.0976
0.0922
0.0784
0.0796
0.0752
0.08
0.07

The last decade is actually 1991-2002. The .07 Means that home teams had a .535 winning percentage.

I also found the percentage of all players with 400+ PAs who were 20-24 for each decade. Here are the percentages over the same decades as HFA (except the last is 1991-2000)

16.11
21.04
15.95
16.05
18.29
14.60
21.77
21.88
15.34
13.07

I know it is only 10 observations, but the correlation is -.18. So the greater the % of young guys, the lower the HFA, although it is not statistically significant.

My HFA article from the Baseball Research Journal is at

http://www.geocities.com/cyrilmorong@sbcglobal.net/HomeRoad.htm

My article on player aging from Beyond the Boxscore. I don’t have the data there on aging that I present here. There is a graph, but part of it is cutoff since BTB made some format changes

http://www.beyondtheboxscore.com/story/2006/9/8/144018/2113

How much higher than 2 SD is it?  There are lots of weird things that could happen just by chance in the first two months of the season ... maybe this is just this year’s.

Yeah, I agree that if we start out with the question of “what is weird so far this year” we are going to find some things that are weird by chance alone, by definition.

There were around 800 games as of May 29.  According to the binomial, that is around .018 for one SD (a little more, given the fluctuating wp of each game). 

Assuming that .535 is the “true” mean home wp, .577 is 2.35 SD from the mean.  Not alarming, but, as I said, I find it somewhat troubling given the changes in the way baseballs are supposedly stored this year.

Yes, and the “theories” presented by Olney are troubling in two respects:  One, the fact that there are NOT more young players in baseball this year.  Two, the fact that these “experts” are just making up crap about young players having unusually good home/road splits, in order to fit the observation. 

If HFA were less this year, there is little doubt in my mind that the “experts” could easily have used the “reverse” argument.  “Given all the young players in baseball this year, they just don’t have the same comfort and familiarity level at home that the veterans have.”

How does ESPN (and Olney) even print this crap?

Once the game begins, don’t the umps have control of all the balls?

I wonder about the relationship between these two ‘weird things’ so far this year. As that one major league scout proposed, it does make sense that a weaker run-scoring environment would have an increased home field advantage.

So 2 questions that should be pretty simple to answer.

1. How does HWP change in response to runs scored per inning?

2. Is the increase in home field advantage this year showing up as an increased advantage in runs scored per inning or is that remaining fairly constant?

I don’t think that home wp has much, if any, relationship to rpg.  And run scoring is barely down in both leagues combined.

I have no idea where the balls come from once the game starts.  Obviously the ball boy gets them from somewhere and gives them to the umps.  Where the ball boy gets them from, I have no idea.

For whatever it’s worth, note that over the full 1978 season, the major league home team record jumped quirkishly to .573, essentially the same as it is now for 2008.  In 1977 and 1979 the HFA was perfectly normal—the ‘78 jump was presumably just an odd random variation, though if the sabermetric community had been then what it is now, there presumably would have been much speculation over explanations.

I alwsys like to think in terms of the null hypothesis (nothing ever goes on the change the HFA), and what the distribution would look like if that were true.  We would expect the HFA to be greater than .555 or less than .515 (2 sigma) once per 20 years by chance alone.  There would even be a 50% chance or so for one season to be greater than .560 (2.5 sigma), again, by chance alone.

Anyway, barring some information that we know about, picking out the first third of a season “after the fact” is cherry picking or curve fitting, or whatever it is called.  It is worth looking into, but not much more than that.

hmm maybe this is why the braves home winning pct is so good and their road so poorly lol

OK, here are the actual w%, pythagorean, and BaseRuns/out pythagorean (1.83 exponent) for home teams in 2008, 2007, and 2003-2007:

5-year: .542, .539, .539
2007: .542, .537, .537
2008: .577, .566, .560

The highest HFA in those 5 years was 2004: .550, .541, .541.

For the five year period, if you just used Tango’s szERA (ERA estimator using only K and BB), the home team has a .520 advantage. This year it is .515. The 5 year home has .165 K/PA and .081 BB/PA against .175 K/PA and .076 BB/PA on the road. This year, it’s .170 / .086 and .176 / .082.

There has been a slight uptick in HR HFA which has a reasonably big impact on the numbers; the difference is 19 HR (one every 44 games), and if you take them out the BsR HFA shrinks to .531. Balls in play have also had a far above average split. The split for wOBA on balls in play is .299 against .293 for the five-year, but .300 against .285 for 2008.

3B/(2B+3B), home: .099 5-yr, .098 2008
3B/(2B+3B), road: .087 5-yr, .071 2008
(1B+ROE)/(AB+SF-K-XBH), home: .2587 5-yr, .2608 2008
(1B+ROE)/(AB+SF-K-XBH), road: .2534 5-yr, .2457 2008

So, I think the argument that it is merely chance holds the most weight. If I were to try to come up with a hypothesis that would superficially match the data as Olney et al might, it would be that defenses have been extra good at home this season. I don’t really think that that is worth discussing. I don’t think the 19 HR are worth discussing whatsoever.

The .531 in that second paragraph was some sort of typo that took over the sentence after the error had been made. It should be .558.

If you want to check Buster’s claim about young players partying on the road or whatnot, why not just find the OPS home/road splits for each age group, and see if they are actually the cause of the splits more than other players (and if it’s that they’re that much better than expected at home, or that much worse than expected on the road).

Sal, that seems like a good and logical idea, but I’d bet that the difference is so tiny that it barely even shows up. Could be worthwhile to give it a shot for someone who has the data readily available.

I’d guess that it would be a negligible difference if at all as well, but it is possible, and doing it by team (for instance) may show that some clubhouses are more apt to drunken revelry than others.

The thing is, we just don’t know unless we check. And then once we do check, we can all mock Buster Olney about it.

(If Olney persists in claiming they’re partying hard on the road, yet their performance is better, we should all calculate what the HFA would be like if they drank at home too!)

I have a hard time believing that if teams were using a different set of balls for home and away teams that the HR rate in the AL would be down.  Up or even I could understand.  Down makes no sense to me.

Can each team chose their own “climate” to keep the balls in or is there a standard MLB “climate” that all teams are suposed to use.

I imagine a baseball would behave quite differently if you set the climate chamber to emulate july in the Sahara desert or Januari in Alaska.

I also imagine that some teams would be advantaged by the former and some by the later depending om exactly how the teams are “constructed”.

Another thing that could matter is that some of the physical properties of the baseballs probably doesn’t “hold” very well once out of the climate chamber. For instance I would guess that a ball of significantly different temprature than the ambient temperature wouldn’t be able to hold that temperature very long. So I guess you need to consider when during the game balls are brought out.

http://tangotiger.net/wins.html

The very first number you see is: 0.4608

Subtract that from 1, and you get .5392

That is the historical HFA, from 1957-2006, excluding 1999.

If the report is .577, that is 38 points above the norm. 

I seem to remember that there was a recent season, perhaps the AL about 5 years ago, where there was almost no HFA, and that was for the entire season.

Anyway, 2.1 SD on one side occurs 2.5% of the time, right?  There’s been 50 seasons in that sample, so we’d expect one such season to really overshoot and one to undershoot. 

It just seems to me that Buster understands numbers just enough to be dangerous.

Could it have anything to do with better late game managing decisions from the home team with the last turn to bat?  Maybe not bunting excessively like in year’s past?

Ken, almost all of the HFA comes from better component performance (s,d,t,hr,bb,hp,etc,) at home than on the road.  There is virtually no advantage to batting last in the first place, so any change in “late game” or home team strategy is not going to have any noticeable effect on overall HFA.

I can’t really think of anything that is going to have a significant “all of a sudden” (as opposed to gradual - over the years) effect on HFA, other than teams monkeying around with the baseballs.  I am not suggesting that that is a strong possibility.  I don’t think it is.  I don’t think it is a good idea to let teams “monkey” with the baseballs unless there are stringent controls, but still, I think it is way more likely to be a fluke than anything conspiratorial by certain teams.