Home/Away Performance needs to be tweaked

[Le Grande Orange]

Originally Posted by Malleus Dei

What other factors besides tailoring are there, LGO? Better familiarity with the field? Sleeping at home? Travel fatigue? Regular eating habits? Less carousing? The cheers of the home crowd? Given that the statistical differences are negligible, the major factor has to be how much the team is tailored to the field...

Why? Why are all the other factors you just mentioned automatically given lower weight than tailoring the team?

[Malleus Dei]

Originally Posted by gmo

Walks and strikeouts show the greatest differences (along with runs) among the stats

Exactly, and those are the only stats that show any significant difference. You can pretty well attribute the K's and W's to the home team's better familiarity/comfort with that particular batting box and with the associated batting background and sun/shadow effects of the field (been there/done that/have the jersey).

But - here's the important part - there just aren't any other significant statistical differences between home and away teams, and batting last can't account for the difference. You saw all that for yourself. And the increased W's/decreased K's rate simply can't account for the W-L% difference.

You are left with the effects of sleeping at home, travel fatigue, regular/irregular eating habits, different levels of carousing, the effects of the cheers of the home crowd...and, gee, how well the team is tailored to the ballpark (and vice versa).

Again, note the results in Montoya's league. They tailored their OOTP teams to their home fields and GOT HISTORICAL HOME FIELD ADVANTAGE RATES.

[Le Grande Orange]

So, as I mentioned earlier, I just went through the win-loss results at Retrosheet for four stadiums: the Astrodome, Atlanta Fulton-Country Stadium, Olympic Stadium, and Riverfront Stadium. All four of these parks had symmetric outfield fences, foul territories which are roughly similar, and field dimensions which are very similar. Given this, tailoring the team to its home park would be of little value as their are no distinctive features to tailor for.

I used only the results of these four clubs against each other; this removes any chance of teams in asymmetric parks from skewing the results. That leaves circular, symmetric ballpark teams playing other circular, symmetric ballpark teams for only those years when the aforementioned ballparks were in use for a full season.

The final result:

2166 games played.

Home team record: 1141-1025 .5268
Away team record: 1025-1141 .4732

The home winning percentage was .05355 points higher than the away winning percentage. The home team won 5.36% more games at home than the overall winning percentage suggests it should have won. The home team won 11.32% more games than it lost.

And all of this in games between clubs playing in nearly identical parks with no distinctive features to tailor a team for. So what factors are at work to explain these results?

[Malleus Dei]

Originally Posted by Le Grande Orange

Why? Why are all the other factors you just mentioned automatically given lower weight than tailoring the team?

1. Short Answer: Because, even when taken together, all the other factors are pretty darned trivial. Did you ever play the game at any level?

2. Long Answer: Because studies (add gmo's to the list) show that the only really significant statistical difference between home and away teams is in K's and W's, which is most logically explained by batter's box/background familiarity (Occam's Razor). There is also a very slight defensive improvement noted - again, almost certainly due to familiarity. If there was another cause besides tailoring - travel fatigue, home field, etc. - then ITS EFFECTS WOULD SHOW UP CLEARLY IN THE STATS. The home team would perform better than its opponents do and, darn it, it just doesn't do that. You therefore have to explain a .040 W-L difference without resulting to improved/decreased performance in any way. The *only* way that can occur is for the home team to somehow be more effective than the away team is, because both of them have essentially the same stats.

And how can the home team be more effective than the away team with the same stats? How can it get more bang for its essentially identical statistical buck? There's only way: it has to be better tailored to its home field than the away team is.

[Le Grande Orange]

Originally Posted by Malleus Dei

You are left with the effects of sleeping at home, travel fatigue, regular/irregular eating habits, different levels of carousing, the effects of the cheers of the home crowd...and, gee, how well the team is tailored to the ballpark (and vice versa).

Yes, and all contributing to the final results. The only debate is the relative strength of each of these factors. You place tailoring as the pre-eminent cause; I would submit the widescale results from MLB over a great many years and clubs demonstrates that tailoring cannot be the pre-eminent cause.

An equal cause, perhaps, but not a pre-eminent one.

[gmo]

Originally Posted by Malleus Dei

Exactly, and those are the only stats that show any significant difference. You can pretty well attribute the K's and W's to the home team's better familiarity/comfort with that particular batting box and with the associated batting background and sun/shadow effects of the field (been there/done that/have the jersey).

But - here's the important part - there just aren't any other significant statistical differences between home and away teams, and batting last can't account for the difference. You saw all that for yourself. And the increased W's/decreased K's rate simply can't account for the W-L% difference.

You are left with the effects of sleeping at home, travel fatigue, regular/irregular eating habits, different levels of carousing, the effects of the cheers of the home crowd...and, gee, how well the team is tailored to the ballpark (and vice versa).

Again, note the results in Montoya's league. They tailored their OOTP teams to their home fields and GOT HISTORICAL HOME FIELD ADVANTAGE RATES.

I am not sure what you mean by "significant" in the differences. It seems you mean more than half that of the BB/K numbers looking back at the numbers I showed. {EDIT - I am guessing you mean "meaningful" because I would suspsect the differences in the other stats would probably pass tests of statistical significance, if not in the sample over 2001-2003 alone, then in more complete stats I expect would show similar differences.} You say that the BB/K differences are from some sort of a comfort thing, but you dismiss that for the other stats because... the differences are smaller in those stats? Tailoring to the park gives you the differences in wins and the differences in other stats like AVG, but those statistical differences outside BB & K are not meaningful, even though you say the BB/K differences do not explain the differences in results? Why would BB/K differences be related to comfort but differences in AVG not be? I simply do not follow.

The league linked by f.montoya is nice, but the numbers are for 20 teams over about two-thirds of one season. Looking at some unique field teams, Boston & New York in the AL show no clear HFA, but LA in the NL does. Are those results of poor & smart GMs or just variations in small samples? It is hard for me to draw any sort of conclusions about the relative importance of tailoring from that little data even showing a 55.5% home team winning percentage.

Back to some previous points - How do you explain HFA in other sports, particularly those with uniform playing dimensions like basketball? By your reasoning how does HFA work for teams that play in neutral parks (and thus by extension how does that influence the teams that do)?

[Le Grande Orange]

Originally Posted by Malleus Dei

And how can the home team be more effective than the away team with the same stats? How can it get more bang for its essentially identical statistical buck? There's only way: it has to be better tailored to its home field than the away team is.

And how can a home team be tailored to its park to produce better results when there is no distinctive feature to tailor for?

If anything, the numbers I just posted demonstate that all other causes contribute to the home field advantage up to .527. That means the tailoring aspect would only contribute the remaining .013 of the previously stated .540 home field record. In other words, tailoring accounts for approximately one-third of the reason for the home team doing better and all other causes contribute the other two-thirds.

[Malleus Dei]

Originally Posted by Le Grande Orange

the Astrodome, Atlanta Fulton-Country Stadium, Olympic Stadium, and Riverfront Stadium. All four of these parks had symmetric outfield fences, foul territories which are roughly similar, and field dimensions which are very similar.

Pardon me, but you're making me laugh. Have you actually been in these four stadiums? I have. Do you realize that you are desperately trying to lump together domed stadiums with turf in with open stadiums with grass? That you have lumped in a stadium known to be one of the worst for hitters in all of baseball with one known as "The Launching Pad"? LOLOLOLOLOLOL.

Sorry. Wrong number.

[Malleus Dei]

Originally Posted by Le Grande Orange

Yes, and all contributing to the final results. The only debate is the relative strength of each of these factors. You place tailoring as the pre-eminent cause; I would submit the widescale results from MLB over a great many years and clubs demonstrates that tailoring cannot be the pre-eminent cause.

An equal cause, perhaps, but not a pre-eminent one.

Nonsense, and I showed you why. If the cause was anything BUT tailoring to any significant degree, then the home teams would have better stats. They don't, but they still win more games. The only way they can do that is to be more efficient - i.e, be better tailored.

[Le Grande Orange]

Originally Posted by Malleus Dei

Pardon me, but you're making me laugh. Have you actually been in these four stadiums? I have. Do you realize that you are desperately trying to lump together domed stadiums with turf in with open stadiums with grass?

Hmm, I based the picks on the great similarity in dimensions, as you yourself used the example of the Yankees tailoring for a short left field fence. Now you're saying dimensions don't count? Which is it?

[Le Grande Orange]

Originally Posted by Malleus Dei

If the cause was anything BUT tailoring to any significant degree, then the home teams would have better stats.

Why? Why would they do better? If all other things are equal, the results should be equal.

[Malleus Dei]

Originally Posted by Le Grande Orange

And how can a home team be tailored to its park to produce better results when there is no distinctive feature to tailor for?

There are always going to be distinctive features. Look at the park effects.

[Malleus Dei]

Originally Posted by Le Grande Orange

Hmm, I based the picks on the great similarity in dimensions, as you yourself used the example of the Yankees tailoring for a short left field fence. Now you're saying dimensions don't count? Which is it?

Everything counts. Turf versus grass counts (speed). Dimensions count (power/pitching).

You're simply not getting it and I'm tired of trying to explain it to you. Good night.

[Malleus Dei]

Originally Posted by gmo

How do you explain HFA in other sports, particularly those with uniform playing dimensions like basketball? By your reasoning how does HFA work for teams that play in neutral parks (and thus by extension how does that influence the teams that do)?

I don't. I'm a baseball guy.

Good night, and have fun. See you all after Christmas.

[Malleus Dei]

And do grab a copy of WEAVER ON STRATEGY and see what Earl writes about it.

[Le Grande Orange]

Park effects? How can the same physical ballpark with the same physical dimensions go from a hitter's park one year to a pitcher's park the very next year? How does the Astrodome go from having a simple park factor of 0.854 in 1988 to a park factor of 1.056 in 1989? That's a heck of a change.

[gmo]

Are you saying that BB/K differences cannot explain the win/loss differences observed, that the differences in the other stats are trivial, so the win/loss differences must be largely explained by the tailoring? How would there be win/loss differences without some sort of statistical underpinning? You can find individual cases of teams doing better in terms of wins than their stats suggest, but I think the power of the baseball version of the Pythagorean theorem shows that more runs scored and fewer allowed will overall lead to more wins. Where do those differences in wins, which logically come from differences in runs, come from if not the stats? Are there just not enough stats with the ones I raised like BB/K & AVG/OBP/SLG?

Well, I hope this discussion does not die off even though I too will be checking out for a while soon. Even though this is not new ground, I think it is good to go through some old topics periodically. In this case, I am simply not following the logic MD has used to explain his beliefs, and his conclusions clash with my own feelings that look at the observed HFA (54% home win pct as opposed to 50%) as being perhaps a few percent from batting last, maybe a quarter but likely less from tailoring, and the remaining three-quarters-plus from the differences in stats like home teams walking 7+% percent more and hitting for a 3+% higher average (and we have not even gotten into pitchers, and the likes of greater comfort they may have pitching where they do half the time rather than perhaps a handful of times per season). It would be interesting if someone could show what the differences in the home/away splits like I showed should produce in terms of runs scored, which then is just a short step to expected wins and losses.

[mrbill]

Originally Posted by Malleus Dei

2. Long Answer: Because studies (add gmo's to the list) show that the only really significant statistical difference between home and away teams is in K's and W's, which is most logically explained by batter's box/background familiarity (Occam's Razor). There is also a very slight defensive improvement noted - again, almost certainly due to familiarity. If there was another cause besides tailoring - travel fatigue, home field, etc. - then ITS EFFECTS WOULD SHOW UP CLEARLY IN THE STATS. The home team would perform better than its opponents do and, darn it, it just doesn't do that. You therefore have to explain a .040 W-L difference without resulting to improved/decreased performance in any way. The *only* way that can occur is for the home team to somehow be more effective than the away team is, because both of them have essentially the same stats.

And how can the home team be more effective than the away team with the same stats? How can it get more bang for its essentially identical statistical buck? There's only way: it has to be better tailored to its home field than the away team is.

While I agree with MD on almost everything he's said, when I think about how OOTP should model the "long answer" part of this, I have to think there is a necessary "cheat code" argument to have.

Yes, you can tailor your team to your park, but any player will be familiar with the home batting box, get better rest, the quirks of the field's surface, regardless of where they call home. It's the "park effects" and "park familiarity" that contribute to the 54%, one is tangible, one is a lot less.

If I had to guess at a proper OOTP model for "park familiarity", I'd say we should test whether there should be a slight advantage to contact/avoiding Ks/fielding range/fielding % based on HOW LONG a player has played at his home field. So, a veteran on a particular team starts playing better at home than his ratings would suggest (albeit slightly), but if he's traded, he's back to just what his ratings say, until he 're-learns' his new home field.

The depth of this impact is clearly up for debate here, and in order to model it in OOTP some numbers would need to be researched up to support/debunk it.

Of course, the other "problem" with OOTP6, was that only park dimensions are unique across parks, and you can't get any effect from weather, prevailing winds, turf/grass, so there's a lot less to tweak your team for, providing a smaller potential home field advantage based on tweaking alone (compared to MLB).

That's probably the other suggestion for OOTP, rather than just adjusting dimensions, adjust other factors about the park in a way that they can be taken advantage of.

[QuestGAV]

Originally Posted by Malleus Dei

What other factors besides tailoring are there, LGO? Better familiarity with the field? Sleeping at home? Travel fatigue? Regular eating habits? Less carousing? The cheers of the home crowd?

Yes, yes, yes, yes, yes, and yes.

Given that the statistical differences are negligible,

They're not.

the major factor has to be how much the team is tailored to the field -

An assertion, not a fact.

How this translates to OOTP: a better GM AI which will tailor a team to its home field will replicate home historical field advantage.

This is based on the anecdotal evidence in Montoya's post above, where humans doing just that in OOTP may or may not have matched the actual home field advantage rate in the MLB.

Fixed that for you. You may believe certain things. They are not empirically supported and talking about them as facts gives your argument LESS weight.

[RonCo]

Home field advantage is always interesting to discuss because no one that I've seen can prove definintively what it is. To put it into the game, you have to actually code something, though.

Here's what I think is known:

1. Home teams walk a little more than visiting teams (this should be able to be quantified and coded)
2. Home teams strike out a little less than visiting teams (this should be able to be quantified and coded)
3. Home teams often triple a little more than visiting teams (this should be able to be quantified and coded)
4. Home teams play a little better defense than visiting teams (this shouldbe able to be quantified and coded)
5. Teams can, and do, attempt to tailor their personnel to their home park. I suspect this is a little harder to code, but could be done. Of course, online leagues are reliant upon us wet-ware folks to do the job.

At the end of the day, home field advantage is only a 3-6 win difference, depending on the quality of your team. This is not a lot of runs in the big picture. These five things would seem to be enough to do a decent job of coding a realistic home field advantage.

As I remember, one point that has been statistically proven to _not_ affect performance in any way is the overall distance of travel. A New York team visiting LA doesn't appear to play any differently agasint them than they would against a team closer to NY. This would lead me to think that jet lag/sleep/whatever isn't much of a factor.