Baseball Home Field Advantage

[tweek]

So I'm trying to figure out the best way to include home field advantage in my calculations.

Obviously the most rudimentary would be to add some % win probability to the home team, the same for all home teams regardless of anything else.

Slightly more advanced may be to calculate a historic home field advantage for each park, with the rationale that each park favors the home team differently. But this wouldn't really account for who is playing.

What I really want to do is somehow use who is playing, and perhaps where they are playing, to get a better estimate of home TEAM advantage on a game-by-game basis.

So my questions are:

- What really causes the home team to win more than the road team? Is it simply having the last at bat if needed? Is it the fact that a ballclub can tailor their lineup to their home park?
- If not, what individual, team, or park stats can be extracted to get the best estimate of home team advantage for any particular game?

[wintermute]

Since 1992 the home team has won 53.85 percent of it's games.

In my model I increase the home team's probability by 2 percent and reduce the away team's probability by 2 percent. You still have to take park effects into account.

[tweek]

Since 1992 the home team has won 53.85 percent of it's games.

In my model I increase the home team's probability by 2 percent and reduce the away team's probability by 2 percent. You still have to take park effects into account.

Wouldn't you want to add 4% to the home team and subtract 4% from the away team? Presuming you're assuming a 4% edge. I think you would use +/- 2% in an "odds" type calculation, but since it's straight win percentage, I think you'd want to use 4%. If you start at the 50/50 case, you use +/- 4% to get to 54/46, which is what you'd get historically... I think this would be right anyway.

[coldhardfacts]

It depends on the team. The home field advantage for the Phillies, at least this year, is not the same as it is for the Dodgers.

[curious]

I'm not meaning to be rude, but I think using league stats for any purpose is unwise.

Different teams play Home/Away very differently.

What you should do is look at each team and see how that team plays Home / Away.

Then you have to consider a few other factors. Do they play a team that is not really a Visitor? The Washington Nationals playing at Camden Yards comes to mind. Or the Yankees playing as visitor vs the Mets. Or Oakland as visitor vs the Giants.

Some teams get a Home advantage because their is something about their park. Take an open air park someplace that gets REALLY hot. Like either Florida team or Arizona, or any of the Texas teams. It is a day game, a team from the northeast which is on a long road trip is the Visiting team. Will the old dudes be able to take the heat as well as the young dudes who play every day in this kind of heat?

Baseball is a tough mental game. There is something to be said for having 30,000 screaming fans helping you. There is also something to be said for a team that draws 3,000 fans. OUCH. I would bet against them.

So, you should also look at attendance at home, there might be something there.

A site that I use is SportsDatabase.com. You can run queries to get a team's home record for a given line.

I like to look at Home, Day/Night, and a line range of current line +- 20.

[Justin7]

Welcome back, Curious!

[waiverwire]

Can't say I agree with any of the more 'advanced' approaches recommended here. Either just adjust all teams the same, or adjust component stats of individual players before trying to figure out projected runs for and against for each team. Most research has showed that home teams benefit in walk, strikeouts, and triples. Not much difference between home vs. away for the other component stats.

[tweek]

Can't say I agree with any of the more 'advanced' approaches recommended here. Either just adjust all teams the same, or adjust component stats of individual players before trying to figure out projected runs for and against for each team. Most research has showed that home teams benefit in walk, strikeouts, and triples. Not much difference between home vs. away for the other component stats.

Interesting. Good to know about walks, K's, and triples. Seems somewhat "random" though... any thoughts as to why those three things?

[waiverwire]

Interesting. Good to know about walks, K's, and triples. Seems somewhat "random" though... any thoughts as to why those three things?

Not really. I guess triples could be due to outfielders familiarity with the quirks of the park. One cynical view of the walks and strikeouts could be that umpires favor the home team. Another possibility might be something to do with hitters being familiarity with the view/backdrop behind the pitcher.

[Pokerjoe]

I also would suggest being very wary of team-specific HFA. They necessarily involve small sample sizes (in football, REALLY small sizes.)

Baseball has larger samples, but smaller effects, so that you still have a problem.

There really isn't much value in looking at recent history at all, in fact. Just because a team has a better record on road than at home, for example (and it sometimes happens in sports) doesn't mean it didn't have an HFA, it most likely only means that the effect of the HFA got lost in other noise.

That isn't to say that Hawaii doesn't have a non-generic HFA in football and baskets. It's just to say that it's best to go with the generic, league-wide HFA unless you have real statistical evidence for a team-specific HFA.

In baseball, it's more likely to arise from extreme home field peculiarities COMBINED with a team created or evolved to FIT those peculiarities. Without the fit, park effects will specifically impact totals, but not sides.

[evo34]

"curious" is dead wrong. People (like curious) who try to over-evaluate the theoretical, team-specific reasons for HFA, will produce mispricings you can profit from. Pokerjoe is correct in saying that you need to use generic HFA unless faced with overwhelming evidence something else is better. One thing I have not fully investigated is the HFA for games 1/2/3/4 of a series. Baseball is the only sport where a road game may be played in the same city in which the previous game was played. I would speculate that HFA decreases slightly for later games in a series, but I have not done the research. If anyone has, please post.

[coldhardfacts]

"curious" is dead wrong. People (like curious) who try to over-evaluate the theoretical, team-specific reasons for HFA, will produce mispricings you can profit from. Pokerjoe is correct in saying that you need to use generic HFA unless faced with overwhelming evidence something else is better. One thing I have not fully investigated is the HFA for games 1/2/3/4 of a series. Baseball is the only sport where a road game may be played in the same city in which the previous game was played. I would speculate that HFA decreases slightly for later games in a series, but I have not done the research. If anyone has, please post.

Actually, Curious is dead RIGHT. If you think the relative home/road advantage/disadvantage for the Padres and the Diamondbacks should be weighted equally when handicapping, then you are the one who is mispricing the game.

[Pokerjoe]

Actually, Curious is dead RIGHT. If you think the relative home/road advantage/disadvantage for the Padres and the Diamondbacks should be weighted equally when handicapping, then you are the one who is mispricing the game.

We're not talking park effect on totals, right? We're talking sides only.

With that in mind, show us your statistical evidence that AZ and SD should have team-specific HFAs.

Here's why the burden of proof is on anyone suggesting team-specific HFA: the sample size of ALL teams is >>>>> the sample size of ONE team.

You're saying, reject the results from the much larger sample. Okay. Prove it.

Here's the test: use or develop a power rating system (it doesn't even have to be that good; this test isn't about the PR system).

Then, compare the PR with generic HFA ... to the PR with team specific HFA. ((Make sure the team-specific HFA is also "generated" (as opposed to guessed at/opinionated), because only a generated HFA can be replicated and tested.))

Compared how? Either ROI, or accuracy re the final line, or accuracy re final score. And then see which is more accurate. Don't backfit.

The point is, this is a math question, not a baseball knowledge or opinion question.

Mind you, in my post above, I did agree that A) a substantial park effect, combined with B) a home team built or evolved to fit that park effect, could C) lead to a team-specific HFA.

But this needs to be established statistically, not guessed at from the bleachers with a beer in one hand and 'baseball knowledge' in the other.

BTW, I fear this thread will degenerate into a "statistics is bullshit, man," "sabermetrics? what's that?" type thread. So, let's try not to go that way.

And BTW #2: I don't bet or follow MLB. My interest is more general. I don't have a horse in the race. I use mildly team-specific adjusted HFA's in CFB; I've never been able to find a way to reliably do that in CBK. Although I admit my standards might be unreasonably high.

[coldhardfacts]

We're not talking park effect on totals, right? We're talking sides only.

With that in mind, show us your statistical evidence that AZ and SD should have team-specific HFAs.

Here's why the burden of proof is on anyone suggesting team-specific HFA: the sample size of ALL teams is >>>>> the sample size of ONE team.

You're saying, reject the results from the much larger sample. Okay. Prove it.

Here's the test: use or develop a power rating system (it doesn't even have to be that good; this test isn't about the PR system).

Then, compare the PR with generic HFA ... to the PR with team specific HFA. ((Make sure the team-specific HFA is also "generated" (as opposed to guessed at/opinionated), because only a generated HFA can be replicated and tested.))

Compared how? Either ROI, or accuracy re the final line, or accuracy re final score. And then see which is more accurate. Don't backfit.

The point is, this is a math question, not a baseball knowledge or opinion question.

Mind you, in my post above, I did agree that A) a substantial park effect, combined with B) a home team built or evolved to fit that park effect, could C) lead to a team-specific HFA.

But this needs to be established statistically, not guessed at from the bleachers with a beer in one hand and 'baseball knowledge' in the other.

BTW, I fear this thread will degenerate into a "statistics is bullshit, man," "sabermetrics? what's that?" type thread. So, let's try not to go that way.

And BTW #2: I don't bet or follow MLB. My interest is more general. I don't have a horse in the race. I use mildly team-specific adjusted HFA's in CFB; I've never been able to find a way to reliably do that in CBK. Although I admit my standards might be unreasonably high.

Sir,

I DO follow and bet on MLB. And nothing helped my handicapping more than reading Money Ball, so I'm the last person who would tell you that statistics and sabermetrics are bullshit.

San Diego is 23-20 at home and 9-29 on the road. Arizona is 18-28 at home, 18-21 on the road. These are STATISTICS based on a fairly large number of games. All I'm saying is if you're handicapping methodology weighs San Diego's chances on the road equal with Arizona's chances on the road, and Arizona's chances at home equal with San Diego's chances at home, you're making a big mistake.

[Pancho sanza]

Sir,

I DO follow and bet on MLB. And nothing helped my handicapping more than reading Money Ball, so I'm the last person who would tell you that statistics and sabermetrics are bullshit.

San Diego is 23-20 at home and 9-29 on the road. Arizona is 18-28 at home, 18-21 on the road. These are STATISTICS based on a fairly large number of games. All I'm saying is if you're handicapping methodology weighs San Diego's chances on the road equal with Arizona's chances on the road, and Arizona's chances at home equal with San Diego's chances at home, you're making a big mistake.

2 words for you

sample size

[coldhardfacts]

2 words for you

sample size

So you're saying the home/road records of MLB teams halfway through the season relative to their overall records is nothing more than chance?

[curious]

"curious" is dead wrong. People (like curious) who try to over-evaluate the theoretical, team-specific reasons for HFA, will produce mispricings you can profit from. Pokerjoe is correct in saying that you need to use generic HFA unless faced with overwhelming evidence something else is better. One thing I have not fully investigated is the HFA for games 1/2/3/4 of a series. Baseball is the only sport where a road game may be played in the same city in which the previous game was played. I would speculate that HFA decreases slightly for later games in a series, but I have not done the research. If anyone has, please post.

Let's see crackhead, my record in MLB the previous two seasons, this season has too few trials to be meaningful because I just started betting MLB this season, is > 70%, my record in NCAAF, NCAAB, and NBA over the previous two seasons is also > 70%. And I play thousands of wagers in each sport in each season because I do sides, totals, halves, quarters, 1st inning, 1st five innings, etc.

If you had faded me in any sport the last two years you would be bankrupt.

Over-evaluate? What the **** does that even mean?

The other night I was on Texas at home vs the Rays for $6,000 to win $7500 because Texas as a home dog is a really good play Texas was +125. EVERYONE on the forum jumped on me for going against the Rays in that game. Texas won 5-2. I knew I was right about that game. If the game had been played in Florida I would not have been so eager to take Texas because as a road small dog they are 8-10. The Rays as a small favorite at home are 12-3. No way I take the Rangers on the road as a small dog vs the Rays. But, the Rangers at home with the same lines? Money in the bank.

Feel free to fade me any time, I post my plays every day.

[curious]

We're not talking park effect on totals, right? We're talking sides only.

With that in mind, show us your statistical evidence that AZ and SD should have team-specific HFAs.

Here's why the burden of proof is on anyone suggesting team-specific HFA: the sample size of ALL teams is >>>>> the sample size of ONE team.

You're saying, reject the results from the much larger sample. Okay. Prove it.

Here's the test: use or develop a power rating system (it doesn't even have to be that good; this test isn't about the PR system).

Then, compare the PR with generic HFA ... to the PR with team specific HFA. ((Make sure the team-specific HFA is also "generated" (as opposed to guessed at/opinionated), because only a generated HFA can be replicated and tested.))

Compared how? Either ROI, or accuracy re the final line, or accuracy re final score. And then see which is more accurate. Don't backfit.

The point is, this is a math question, not a baseball knowledge or opinion question.

Mind you, in my post above, I did agree that A) a substantial park effect, combined with B) a home team built or evolved to fit that park effect, could C) lead to a team-specific HFA.

But this needs to be established statistically, not guessed at from the bleachers with a beer in one hand and 'baseball knowledge' in the other.

BTW, I fear this thread will degenerate into a "statistics is bullshit, man," "sabermetrics? what's that?" type thread. So, let's try not to go that way.

And BTW #2: I don't bet or follow MLB. My interest is more general. I don't have a horse in the race. I use mildly team-specific adjusted HFA's in CFB; I've never been able to find a way to reliably do that in CBK. Although I admit my standards might be unreasonably high.

Wow, what an asshole. Most of the people who are discussing this issue in this thread are not spouting opinion from "the bleachers with a beer in one hand and baseball 'knowledge' in the other". This is the most insulting, condescending horseshit I've seen posted in SBR in a while.

Look at some of my past posts on handicapping theory and see if you think I'm in that category.

If your post had not been so insulting and condescending I would have taken you up on this because I know I can rip you to shreds. I have done the math on this and knowing the stats for a specific team in a specific situation which is a combination of both site, line, and a few other things is critical. I think the concept of HFA is totally flawed. "Give the home team X points". That is nonsense.

But, take Texas anytime they are a small dog at home. Money in the bank.

[curious]

We're not talking park effect on totals, right? We're talking sides only.

With that in mind, show us your statistical evidence that AZ and SD should have team-specific HFAs.

Here's why the burden of proof is on anyone suggesting team-specific HFA: the sample size of ALL teams is >>>>> the sample size of ONE team.

You're saying, reject the results from the much larger sample. Okay. Prove it.

Here's the test: use or develop a power rating system (it doesn't even have to be that good; this test isn't about the PR system).

Then, compare the PR with generic HFA ... to the PR with team specific HFA. ((Make sure the team-specific HFA is also "generated" (as opposed to guessed at/opinionated), because only a generated HFA can be replicated and tested.))

Compared how? Either ROI, or accuracy re the final line, or accuracy re final score. And then see which is more accurate. Don't backfit.

The point is, this is a math question, not a baseball knowledge or opinion question.

Mind you, in my post above, I did agree that A) a substantial park effect, combined with B) a home team built or evolved to fit that park effect, could C) lead to a team-specific HFA.

But this needs to be established statistically, not guessed at from the bleachers with a beer in one hand and 'baseball knowledge' in the other.

BTW, I fear this thread will degenerate into a "statistics is bullshit, man," "sabermetrics? what's that?" type thread. So, let's try not to go that way.

And BTW #2: I don't bet or follow MLB. My interest is more general. I don't have a horse in the race. I use mildly team-specific adjusted HFA's in CFB; I've never been able to find a way to reliably do that in CBK. Although I admit my standards might be unreasonably high.

I have a better idea on how to "prove" this Mr. High and Mighty math expert. (I have an advanced degree in math so don't give me that I must think statistics is bullshit nonsense).

I'll cap games using team specific home field stats as part of my capping formula. You cap games using league HFA in your formula. I'll bet $100 a game. You bet $100 a game. The ROI of each approach is measured at the end of the season.

That is a test that I would gladly accept. But for you to come in here and call us beer drinking idiots and you don't even bet on baseball, you don't know what you are talking about.

[curious]

So you're saying the home/road records of MLB teams halfway through the season relative to their overall records is nothing more than chance?

That is exactly what these idiots are saying. No way these idiots actually bet on baseball. Not for more than a few months anyway, before they go bankrupt.

I loved that one quote from the one idiot: The same size of MLB is >>>> the sample size of any one team.

True and totally meaningless. Because the HFA of all of MLB is an AVERAGE. And AVERAGEs are meaningless in baseball. Matchup is much more important than league averages. How will Team A fare playing Team B TODAY? No way that league averages will help you figure that out UNLESS you are comparing very specific data about how each team performs to the "average" team to predict how a given team will perform if they were playing an "average" team. You can then compare the two predictions to each other. Only thing is, this approach only works in truly team games like basketball. In baseball we have that nasty aberration called a starting pitcher. In basketball the defense can control how an offense performs to a degree, but not in any way close to how a starting pitcher controls how an offense performs.

These crackheads don't know what they are talking about.

[durito]

[Pancho sanza]

So you're saying the home/road records of MLB teams halfway through the season relative to their overall records is nothing more than chance?

Yes.

If AZ truly has a homefield disadvantage I would expect to see this year over year.

In fact in 2008 they had one of the biggest differentials home vs road, but opposite of what you see in 2009.

[Pancho sanza]

Let's see crackhead, my record in MLB the previous two seasons, this season has too few trials to be meaningful because I just started betting MLB this season, is > 70%, my record in NCAAF, NCAAB, and NBA over the previous two seasons is also > 70%. And I play thousands of wagers in each sport in each season because I do sides, totals, halves, quarters, 1st inning, 1st five innings, etc.

If you had faded me in any sport the last two years you would be bankrupt.

Over-evaluate? What the **** does that even mean?

The other night I was on Texas at home vs the Rays for $6,000 to win $7500 because Texas as a home dog is a really good play Texas was +125. EVERYONE on the forum jumped on me for going against the Rays in that game. Texas won 5-2. I knew I was right about that game. If the game had been played in Florida I would not have been so eager to take Texas because as a road small dog they are 8-10. The Rays as a small favorite at home are 12-3. No way I take the Rangers on the road as a small dog vs the Rays. But, the Rangers at home with the same lines? Money in the bank.

Feel free to fade me any time, I post my plays every day.

Ill bet you can't hit even 60 % against -110, make 500 picks from any of the sports you named.

I'll put up a 50k offer at matchbook in the xbets section, ill even give you +110

[curious]

Ill bet you can't hit even 60 % against -110, make 500 picks from any of the sports you named.

I'll put up a 50k offer at matchbook in the xbets section, ill even give you +110

Most of my plays were posted on SBR in my daily thread. In baskets I play the money line only, I don't play the spread.

You dont' know what you are talking about you just come in here shooting your mouth off.

You want to wager me that I can't hit 60%? You are an idiot. You'll be throwing your money away. I'll take the wager (and your money) though.

[curious]

Yes.

If AZ truly has a homefield disadvantage I would expect to see this year over year.

In fact in 2008 they had one of the biggest differentials home vs road, but opposite of what you see in 2009.

Year over year? What does that have to do with anything? You do know that they have this thing called free agency in baseball? And trades? And a minor leagues?

What happened last year has nothing to do with what will happen this year.

[pimike]

[Pancho sanza]

Most of my plays were posted on SBR in my daily thread. In baskets I play the money line only, I don't play the spread.

You dont' know what you are talking about you just come in here shooting your mouth off.

You want to wager me that I can't hit 60%? You are an idiot. You'll be throwing your money away. I'll take the wager (and your money) though.

OK, 60 % over 500 picks is the equivalent of +80 net units against -110

you have to use closing lines, no lines that are 3-4 points off using a local.

Besides, this is the only way we can do it, if you tell me you got +3.5 on a game, I have no way of verifying it.

You tell me which books closing lines you want to use, we can grade your picks off those lines which we can both see at donbest.com?

Agree? If you agree I'll have the offer up at matchbook tonight and you can take all you want.

[curious]

OK, 60 % over 500 picks is the equivalent of +80 net units against -110

you have to use closing lines, no lines that are 3-4 points off using a local.

Besides, this is the only way we can do it, if you tell me you got +3.5 on a game, I have no way of verifying it.

You tell me which books closing lines you want to use, we can grade your picks off those lines which we can both see at donbest.com?

Agree? If you agree I'll have the offer up at matchbook tonight and you can take all you want.

I already did better than this and the posts are in sbr forum. so the thing you are saying i cannot do I already did.

i dont have a matchbook account nor do I want one.

[durito]

i dont have a matchbook account nor do I want one.

No offense, but that right there says it all. Every serious player in the world, whether they live in vegas or not, plays at matchbook. No one passes up that kind of $ savings.

[Pancho sanza]

I already did better than this and the posts are in sbr forum. so the thing you are saying i cannot do I already did.

i dont have a matchbook account nor do I want one.

OK, at least i dont have to waste anymore time with you.

I guess you just dont need the money.

[Pokerjoe]

Curious, I didn't insult you. I wasn't referring to you specifically when I said "beer drinker;" since when is "beer drinker" an insult? I'm a beer drinker, anyway.

All I suggested is that we resort to good statistical analysis, not bleacher-style "opinion" which is, ultimately, at the heart of most sports bets, and thus the heart of most losing.

Your language in response is juvenile. "Crackhead?" "Idiot?" "Asshole?" You humiliate yourself. Completely.

You also failed to present data supporting your position.

Someone else presented data. Of a trivial sample size. If a roulette wheel showed variance of that amount, would you bet on that variance's continuance?

Curious, you aren't interested in learning, you're interested in defending, for ego reasons, your position. Defending it, not with data, but with insults.

That's fine. You can finish the thread by yourself.

But it's interesting that you claim to have an advanced math degree. Personally, I've never known a real mathematician who insults people who are wrong. Instead, all the ones I know disprove wrongful work, and teach the better approach.

If you truly have a good math background, you won't have trouble presenting evidence supporting your position. In fact, if you truly have a math background, you understand that the whole point of math is that people don't have to use insults to win arguments, they can use ... math.

You can start with some evidence supporting your statement: "what happened last year has nothing to do with what will happen this year."

Really, I want to see the proof that there isn't any correlation between, say, one year's team's W/L record and the next year's. I mean, that quote is taken directly from your post above, so you can't say you didn't say it. And if you're right, you are the most revolutionary thinker in sports history. Just, prove it a little.

In the meantime, congrats on your bet sizes and win rate! I'm sure we're all impressed, and completely believe you.

Especially considering you are such a successful sportsbettor, in MLB nonetheless, yet don't use MB.

[marylandmafioso]

has anyone created a proven algorithm or applied a working computer model of their own

[Dazzez]

I have an advanced degree in math

Not only that but you're a ninja war hero, too.

[coldhardfacts]

Year over year? What does that have to do with anything? You do know that they have this thing called free agency in baseball? And trades? And a minor leagues?

What happened last year has nothing to do with what will happen this year.

It's amazing to me that most people here can actually disagree with this. I suppose the Rays and the Phils are the odds on favorites to go to the World Series this year.

I think alot of folks in this forum know alot more about statistics than they do about baseball.

[element1286]

It's amazing to me that most people here can actually disagree with this. I suppose the Rays and the Phils are the odds on favorites to go to the World Series this year.

I think alot of folks in this forum know alot more about statistics than they do about baseball.

If you don't know anything about statistics, you don't know anything about baseball.