59 replies to this topic

[Eric Van]

Just thought I'd post this here rather than bury it in the MLB forum.

There is no accurate strength-of-schedule adjustment anywhere on the web; BP's (as I've noted before) underestimates the effects of competitive balance profundly (by looking at only opponents' SOS rather than schedule-adjusted opponents SOS -- and in fact you need to repeat (nest) the adjustment about 20 times).

Here are adjusted Pythagorean standings based on actual RS and RA through the end of Sunday. The Karma adjustment is positive for bad karma and negative for good karma (that is, it's the change in the number of wins if they had played to their Pyth percentage, just as the SoS is the change in wins if they had had an average schedule).

Adjusted Pyth Standings, 6/14

Team	Act	SoS	Karma	Adj
Bos	.603	2.9	-1.4	.628
TB	.523	1.5	4.5	.616
LAN	.656	-1.4	-1.6	.609
NYA	.571	2.7	-1.4	.591
Det	.540	1.1	.1	.559
Tor	.523	1.0	1.1	.556
Min	.492	1.8	2.0	.551
Tex	.565	1.0	-2.3	.543
Phi	.590	-1.5	-2.0	.533
Cle	.446	2.8	2.8	.532
LAA	.525	1.4	-1.7	.520
Col	.492	-1.2	2.6	.515
Mil	.540	-1.5	-.8	.502
SF	.548	-1.9	-1.2	.498
StL	.531	-1.6	-.9	.492
ChN	.500	-1.6	.9	.487
NYN	.525	-1.2	-1.1	.487
Pit	.476	-2.3	2.4	.478
ChA	.469	1.4	-.9	.477
KC	.452	1.4	.2	.477
Oak	.435	1.5	.5	.469
Sea	.476	1.1	-2.1	.461
Fla	.492	-0.7	-1.4	.460
Cin	.500	-2.1	-.4	.459
Bal	.429	1.9	-.6	.449
Atl	.484	-1.9	-1.0	.438
Htn	.475	-1.1	-2.0	.424
Ari	.422	-1.5	1.2	.416
SD	.452	0.0	-3.8	.390
Was	.262	-0.6	7.0	.367

And here is the same thing based on EqRS and EqRA, that is, the team's cumulative hitting and pitching stats and the number of wins you'd expect if they had tranlated into RS andd RA with neutral karma. I think the SoS adjustments in this version are the best to use; there's no reason to think that the NL has a real advantage in turning offensive and defensive events into more RS and fewer RA than the AL does despite being the inferior league talent-wise.

Adjusted EqA Pyth Standings, 6/14

Team	Act	SoS	Karma	Adj
TB	.523	2.1	6.2	.651
Bos	.603	3.7	-2.3	.625
LAN	.656	-1.5	-1.9	.603
NYA	.571	3.2	-1.7	.595
Tor	.523	1.2	2.0	.572
NYN	.525	-1.2	3.2	.558
Min	.492	2.1	2.0	.555
LAA	.525	1.6	-.8	.538
Cle	.446	3.2	2.0	.526
Tex	.565	0.9	-3.3	.526
Det	.540	1.1	-2.6	.516
Phi	.590	-0.6	-4.3	.510
KC	.452	1.3	2.2	.509
StL	.531	-1.9	.4	.507
Col	.492	-1.6	1.8	.495
ChN	.500	-2.1	1.2	.485
Sea	.476	1.0	-.5	.483
ChA	.469	1.3	-1.2	.470
Mil	.540	-1.9	-2.9	.463
Atl	.484	-1.7	.2	.460
Bal	.429	2.3	-.4	.458
Ari	.422	-2.3	4.5	.457
Fla	.492	-0.3	-2.2	.454
Oak	.435	1.7	-1.1	.446
Pit	.476	-2.1	.1	.445
Cin	.500	-2.1	-2.1	.432
SF	.548	-1.4	-5.9	.431
Was	.262	-0.6	9.8	.413
Htn	.475	-1.9	-2.0	.412
SD	.452	-0.6	-3.4	.388

You see that four of the five best teams in MLB are in the AL East. Last year, it was the top four.

Average SoS adjustment by division:

AL East: 2.5
Al Central: 1.8
Al West: 1.3
Nl East: -0.9
NL West: -1.5
NL Central: -2.0

To answer the question "how would team X fare in division Y," take the team's own SoS and add it to the division adjustment. For example, the Sox would be 44-19 or 43-20 if they were in the NL Central.

The overall difference between the leagues, 3.4 wins after a bit more than a third of a season, is consistent with last year, when it was 10.3 wins over the full season.

Edited by Eric Van, 16 June 2009 - 05:58 AM.

[sibpin]

EV, do you have SoS information for past seasons? Could you store them in the Wiki? I remember you posted this chart last season but can't find it.

[Eric Van]

EV, do you have SoS information for past seasons? Could you store them in the Wiki? I remember you posted this chart last season but can't find it.

I do have it from last year, and from early in the decade, but not from a bunch of intervening years -- that's a good project for the winter (and I'd use my planned forthcoming revision of BaseRuns / Contextual Runs instead of Equivalent Runs).

I'm crazy busy right now but I'll get it into the wiki in July.

[Pumpsie]

That's the good news. The bad news is that we will continue to have one of the toughest schedules for the remainder of the year because we happen to play in the AL East. That's why the Yankees, Rays and Jays are all in the Top 7 as well.

[trekfan55]

That's the good news. The bad news is that we will continue to have one of the toughest schedules for the remainder of the year because we happen to play in the AL East. That's why the Yankees, Rays and Jays are all in the Top 7 as well.

Foulkie from the Game thread:

Marlins, Braves,@Nationals, @Braves, @Orioles, Mariners, A's, and Royals.

Not one AL East team (except the Orioles) plus 4 series against the National League (No Mets or Phillies). So at least the next 7 series are a lot easier than what the Sox had before.

[Toe Nash]

That's the good news. The bad news is that we will continue to have one of the toughest schedules for the remainder of the year because we happen to play in the AL East. That's why the Yankees, Rays and Jays are all in the Top 7 as well.

Well, they don't go on the road to the West Coast again, and they have played 18 of their 36 games against Tampa and NY. They have 5 more games remaining at home than they do away (52 v. 47). They're played the Mets and the Phillies but have yet to face the weaker members of the NL East. And they have 14 games remaining against Baltimore. They have fewer off days than is ideal, but due to good luck earlier in the year they don't have any postponed games they need to make up the rest of the year. The toughest stretch looks to be August 4-16, where they play @ TB, @ NY, home v DET and then go to Texas for three games before an off day.

So, yes, their schedule is tough being in their division, but the very worst is likely over.

[maufman]

Going into today's games, the AL leads the NL 46-40 in interleague play, but leads 50-36 on a Pythagorean basis.

EV, does your model's assessment of the two leagues' relative strength rest solely on the results of these 86 games, or do you have some other method for determining the relative strength of the two leagues? I ask because even the entire interleague slate probably is a SSS; using 86 games to evaluate the relative strength of 30 teams seems almost meaningless.

[Eric Van]

Going into today's games, the AL leads the NL 46-40 in interleague play, but leads 50-36 on a Pythagorean basis.

EV, does your model's assessment of the two leagues' relative strength rest solely on the results of these 86 games, or do you have some other method for determining the relative strength of the two leagues? I ask because even the entire interleague slate probably is a SSS; using 86 games to evaluate the relative strength of 30 teams seems almost meaningless.

It's just the game played this year. I think that's actually a lot of baseball, about 6600 PA. I'm not sure that the number of teams is relevant. If at the ASB one team was 6 games back, 14 by Pyth, you would be pretty sure about the difference between them. The only random factor is the pairings that have been played, but the methodology adjusts for that.

A single day's games are a SSS and it's still rare to see the NL have an edge. The results in the games so far, as I noted, are perfectly consistent with last year.

[DieHardSoxFan1]

I think the Sox' strength of schedule is rendered less impressive when you take into account the number of elite pitchers the team has managed to evade. To date, the team has missed Roy Halladay in both series with the Blue Jays; both Felix Hernandez and Erik Bedard in their lone matchup against Seattle; both Justin Verlander and Edwin Jackson in their only series against the Tigers; and Cole Hamels in their three-game set with the Phillies. Furthermore, they're missing Josh Johnson this series and Javier Vazquez this weekend.

Obviously abnormalities like this can happen, especially when you're half-way through June. But such a fortuitous abnormality -- missing 8 of the game's top 20 or so starting pitchers, including two of the very best (Halladay and Verlander) -- seems like it can skew the results. The question is to what degree.

Edited by DieHardSoxFan1, 16 June 2009 - 11:47 PM.

[ManilaSoxFAN]

EV, I can't tell if you have also included the Home-Away splits when measuring a team's strength of sched...or its opponentss SOS. Have you?

[Eric Van]

EV, I can't tell if you have also included the Home-Away splits when measuring a team's strength of sched...or its opponentss SOS. Have you?

No home / away splits. That would be a desirable tweak, made less necessary by the fact that at the end of the season, it's almost a complete wash.

To do it precisely right, you would need to have the head-to-head matchup data broken down by home and road, and that's not available in any table that I know of -- it would have to be captured by a ton of web queries and more sophisticated programming than I can muster.

You could get a really good approximation by calculating RS / RA home and away, and EqRS / EqRA, and then making a sophisticated adjustment based on that, but then again, lots of queries and data manipulation.

You could get a decent approximation by doing a standard adjustment on RS, RA, EqRS, and EqRA based on total home / road breakdown. It wouldn't know, e.g., if a team had played mostly good teams at home and weak ones on the road. But you could take the MLB average RS/9 innings home and away, calculate an expected ratio based on the H/A split, and use that as your adjustment.

I think I'll try that tonight -- with this same data, to see how much of an effect it has.

Thanks for making me think about this -- I had assumed that any adjustment at all would be too complex.

P.S. It's surprisingly minor, or at least was last year in MLB. A team that has played 64 games, 29 at home and 35 on the road (Sox were 29 and 34) can be expected to go 31.7 - 32.3 rather than 32.0 - 32.0. But it would be nice to have the adjustments, which will mostly be in the .000-.005 range, at least at this time of year.

[ManilaSoxFAN]

No home / away splits. That would be a desirable tweak, made less necessary by the fact that at the end of the season, it's almost a complete wash.

To do it precisely right, you would need to have the head-to-head matchup data broken down by home and road, and that's not available in any table that I know of -- it would have to be captured by a ton of web queries and more sophisticated programming than I can muster.

You could get a really good approximation by calculating RS / RA home and away, and EqRS / EqRA, and then making a sophisticated adjustment based on that, but then again, lots of queries and data manipulation.

You could get a decent approximation by doing a standard adjustment on RS, RA, EqRS, and EqRA based on total home / road breakdown. It wouldn't know, e.g., if a team had played mostly good teams at home and weak ones on the road. But you could take the MLB average RS/9 innings home and away, calculate an expected ratio based on the H/A split, and use that as your adjustment.

I think I'll try that tonight -- with this same data, to see how much of an effect it has.

Thanks for making me think about this -- I had assumed that any adjustment at all would be too complex.

P.S. It's surprisingly minor, or at least was last year in MLB. A team that has played 64 games, 29 at home and 35 on the road (Sox were 29 and 34) can be expected to go 31.7 - 32.3 rather than 32.0 - 32.0. But it would be nice to have the adjustments, which will mostly be in the .000-.005 range, at least at this time of year.

Yeah, EV. I mean the expectations/results would be different if team X played team A -- which had high RS/low RA, but which played all of its games vs sub-500 teams and at home...versus playing against team B -- which had Lo RS/Hi RA, but played all of its games vs Division leaders and on the road. Right?

Baseball offers a huge advantage for home teams (vis-a-vis other sports). I can't see how/why that should be underestimated when trying to quantify/analyze SOS.

[MentalDisabldLst]

actually i think the home/road splits in baseball are lower than for other sports - certainly basketball and football, and probably hockey too. IIRC the average home team in baseball wins .560 of their games.

[URI]

Baseball offers a huge advantage for home teams (vis-a-vis other sports). I can't see how/why that should be underestimated when trying to quantify/analyze SOS.

Home field advantage by sport:
NBA - 59.8%
NFL - 56.5%
NHL - 55.4%
MLB - 54.8%

[ManilaSoxFAN]

Home field advantage by sport:
NBA - 59.8%
NFL - 56.5%
NHL - 55.4%
MLB - 54.8%

I wasn't aware of those stats. Thanks.

The reason I said that home teams have a bigger edge in baseball is because of the structure of the game...regardless of other home turf advantages like the playing surface and it's unique environment, the fans, the ability to sleep in one's own bed, etc.

Getting the chance to score and win...without the opponent getting an opportunity to strike back. Never losing immediately because the opponent scored...always having the chance to respond and tie or win. It's like "sudden death" rules/advantages can only apply to the home team. That's the edge I was talking about. It's structural.

I'm not cognizant of such a structural (vs an emotional, psychological) advantage in the other sports you mentioned. But, in pro baseball, almost all teams win between 40% and 60% of their games...while in the other sports the gaps between the top and bottom teams are huge. Some NFL teams win/lose 1 or 2 games in an entire season!

Anyway, regardless of the instructive inputs, I would still maintain the validity of the concept that a baseball SoS analysis should incorporate the weight of home/away splits.

[Myt1]

I wasn't aware of those stats. Thanks.

The reason I said that home teams have a bigger edge in baseball is because of the structure of the game...regardless of other home turf advantages like the playing surface and it's unique environment, the fans, the ability to sleep in one's own bed, etc.

Getting the chance to score and win...without the opponent getting an opportunity to strike back. Never losing immediately because the opponent scored...always having the chance to respond and tie or win. It's like "sudden death" rules/advantages can only apply to the home team. That's the edge I was talking about. It's structural.

I'm not cognizant of such a structural (vs an emotional, psychological) advantage in the other sports you mentioned. But, in pro baseball, almost all teams win between 40% and 60% of their games...while in the other sports the gaps between the top and bottom teams are huge. Some NFL teams win/lose 1 or 2 games in an entire season!

Anyway, regardless of the instructive inputs, I would still maintain the validity of the concept that a baseball SoS analysis should incorporate the weight of home/away splits.

I'd say that the structural edge in hockey is decidedly more pronounced. Getting the last line-change can be a huge weapon for the right coach.

[Smiling Joe Hesketh]

I wasn't aware of those stats. Thanks.

The reason I said that home teams have a bigger edge in baseball is because of the structure of the game...regardless of other home turf advantages like the playing surface and it's unique environment, the fans, the ability to sleep in one's own bed, etc.

Getting the chance to score and win...without the opponent getting an opportunity to strike back. Never losing immediately because the opponent scored...always having the chance to respond and tie or win. It's like "sudden death" rules/advantages can only apply to the home team. That's the edge I was talking about. It's structural.

I'm not cognizant of such a structural (vs an emotional, psychological) advantage in the other sports you mentioned. But, in pro baseball, almost all teams win between 40% and 60% of their games...while in the other sports the gaps between the top and bottom teams are huge. Some NFL teams win/lose 1 or 2 games in an entire season!

Anyway, regardless of the instructive inputs, I would still maintain the validity of the concept that a baseball SoS analysis should incorporate the weight of home/away splits.

If there's a structural edge in baseball's HF advantage you'd expect it to show up in the results. So perhaps that facet is already accounted for in the current HF stats URI presented above. Structural advantages or no, baseball's HF advantage is clearly less than other sports' HF advantages; the difference between winning and losing seems to be much tighter in baseball compared to the other sports. We certainly have a large enough sample size of games and seasons to be pretty confident in the results.

Certainly the Sox are HR beasts with their 22-8 record at home this year.

Also, I would like to second DHSF's post that the Sox have had pretty good luck in missing some terrific pitchers so far this season, which probably mitigates their strength of opponents somewhat. Missing Verlander and Jackson in Detroit and getting to face Willis and Galarraga was a pretty good piece of luck. Not facing Halladay at all was another. They'll likely face some of these top pitchers in their next go-around through the league.

[Alcohol&Overcalls]

I'm not cognizant of such a structural (vs an emotional, psychological) advantage in the other sports you mentioned. But, in pro baseball, almost all teams win between 40% and 60% of their games...while in the other sports the gaps between the top and bottom teams are huge. Some NFL teams win/lose 1 or 2 games in an entire season!

Shouldn't this wash, presuming teams split evenly between home and road? While it may increase variance, there's probably enough historical data that it isn't incredibly important.

[ManilaSoxFAN]

If there's a structural edge in baseball's HF advantage you'd expect it to show up in the results. So perhaps that facet is already accounted for in the current HF stats URI presented above. Structural advantages or no, baseball's HF advantage is clearly less than other sports' HF advantages; the difference between winning and losing seems to be much tighter in baseball compared to the other sports. We certainly have a large enough sample size of games and seasons to be pretty confident in the results.

Certainly the Sox are HR beasts with their 22-8 record at home this year.

Also, I would like to second DHSF's post that the Sox have had pretty good luck in missing some terrific pitchers so far this season, which probably mitigates their strength of opponents somewhat. Missing Verlander and Jackson in Detroit and getting to face Willis and Galarraga was a pretty good piece of luck. Not facing Halladay at all was another. They'll likely face some of these top pitchers in their next go-around through the league.

I'm so glad to see you and URI take such interest in this topic. Thanks SJH.

Again, though, I retract any comparisons with other sports (where team W/L records are much, much more lopsided than in baseball) as far as W/L %'s come out. Yet I still think structural "last ups" edge is significant within the confines of the sport itself. The data bears this out, I believe.

In response to your comment: "If there's a structural edge in baseball's HF advantage you'd expect it to show up in the results" -- I think it clearly shows up in the standings column. Good teams almost always play ~.500 ball away, with high 500's (or even a bit above .600) records at home. Even mid/lower-tier teams tend to have winning records at home. In a sport where the best and worst teams play mostly within a 40/60 bandwidth, the home edge carries much weight.

Thus, if you are constructing a SoS analysis, you'd want to take into consideration the H/A splits. Right?

[ManilaSoxFAN]

Shouldn't this wash, presuming teams split evenly between home and road? While it may increase variance, there's probably enough historical data that it isn't incredibly important.

I thought EV was looking at SoS during the course of the season...not doing it for retroactive, end-of-season studies.

[RingoOSU]

Good teams almost always play ~.500 ball away, with high 500's (or even a bit above .600) records at home.

This is not true at all.
Looking over the last 4 years alone:
Rk Tm Year G W L W-L% RS RA pythW-L%
1 CHW 2005 81 52 29 .642 374 302 .597
2 CLE 2005 81 50 31 .617 423 341 .597
3 STL 2005 81 50 31 .617 397 311 .610
4 LAA 2008 81 50 31 .617 382 343 .549
5 DET 2006 81 49 32 .605 430 326 .624
6 NYM 2006 81 47 34 .580 439 384 .561
7 NYM 2007 81 47 34 .580 437 374 .571
8 NYY 2006 81 47 34 .580 480 413 .568
9 LAA 2005 81 46 35 .568 400 332 .584
10 SDP 2006 81 45 36 .556 416 342 .589
11 BOS 2007 81 45 36 .556 395 305 .616
12 CLE 2007 82 45 37 .549 386 338 .560
13 OAK 2006 81 44 37 .543 399 381 .521
14 PHI 2008 81 44 37 .543 387 342 .556
15 LAA 2006 81 44 37 .543 400 386 .516
16 PHI 2006 81 44 37 .543 421 392 .533
17 OAK 2005 81 43 38 .531 381 313 .589
18 DET 2007 81 43 38 .531 441 380 .568
19 CHC 2008 80 42 38 .525 401 332 .586

Playoff caliber teams do a lot better than just .500 on the road.

Edited by RingoOSU, 17 June 2009 - 10:48 AM.

[Smiling Joe Hesketh]

Yet I still think structural "last ups" edge is significant within the confines of the sport itself. The data bears this out, I believe.

I think you might be double-counting this here, to be honest. Any last ups advantage would already be reflected in the HF advantage numbers URI's already posted. Last ups, sleeping in one's own bed, familiarity with the home ballpark (we saw this happen last night with Hanley getting thrown out at second because he was surprised by how close the Wall really was to second base), etc; all show up in the final HF number.

Thus, if you are constructing a SoS analysis, you'd want to take into consideration the H/A splits. Right?

On a really large scale, probably. Through 64 games, the splits are probably small enough that they don't make much of a difference in the final analysis.

[ManilaSoxFAN]

This is not true at all.
Looking over the last 4 years alone:
Rk Tm Year G W L W-L% RS RA pythW-L%
1 CHW 2005 81 52 29 .642 374 302 .597
2 CLE 2005 81 50 31 .617 423 341 .597
3 STL 2005 81 50 31 .617 397 311 .610
4 LAA 2008 81 50 31 .617 382 343 .549
5 DET 2006 81 49 32 .605 430 326 .624
6 NYM 2006 81 47 34 .580 439 384 .561
7 NYM 2007 81 47 34 .580 437 374 .571
8 NYY 2006 81 47 34 .580 480 413 .568
9 LAA 2005 81 46 35 .568 400 332 .584
10 SDP 2006 81 45 36 .556 416 342 .589
11 BOS 2007 81 45 36 .556 395 305 .616
12 CLE 2007 82 45 37 .549 386 338 .560
13 OAK 2006 81 44 37 .543 399 381 .521
14 PHI 2008 81 44 37 .543 387 342 .556
15 LAA 2006 81 44 37 .543 400 386 .516
16 PHI 2006 81 44 37 .543 421 392 .533
17 OAK 2005 81 43 38 .531 381 313 .589
18 DET 2007 81 43 38 .531 441 380 .568
19 CHC 2008 80 42 38 .525 401 332 .586

Playoff caliber teams do a lot better than just .500 on the road.

Sorry, but where is the home-away splits on that spreadsheet?

Anyway, the point is that good teams don't typically lose much on the road (they can stay around/above .500 ball) and win a lot at home...but even just-decent (mid/lower) teams tend to win at home. Of course, this all depends on which team you are playing against.

We're talking about SoS scheds...and whether or not H/A splits should be taken in consideration when considering such.

[RingoOSU]

Yes, but you're the one just quoting "good teams are .500 on the road" like it's a solid number. It's not. The number varies from team to team and year to year. Some teams do lose over 81 games on a year, and yet play over .500 ball at home. Some don't. The number was given that home teams win 54% of the time compared to 50% for all games. If you have numbers that dispute this, post them.

[ManilaSoxFAN]

On a really large scale, probably. Through 64 games, the splits are probably small enough that they don't make much of a difference in the final analysis.

SJH, the structural advantage in baseball for a home team -- disregarding any comparisons to any other sport -- is meaningful...significant even. The data does, in fact, bear this out. Home teams win at a rate that is much higher than do the visitors.

Or are you saying that playing against the Yanks, Tor, TB, Det, Tex away is the equivalent to playing Bal, Cle, Oak and the Nats at home?

In SoS analyses, I don't see how the h/a split can be ignored. Has anyone made the argument that these splits SHOULD be omitted/ignored?

[sibpin]

Getting the chance to score and win...without the opponent getting an opportunity to strike back. Never losing immediately because the opponent scored...always having the chance to respond and tie or win. It's like "sudden death" rules/advantages can only apply to the home team. That's the edge I was talking about. It's structural.

Wouldn't that be an argument for less of a home field advantage in baseball than in other sports? Each team has 9 innings, goal is to score more runs in 9 innings than the opponent did in 9 innings. There is no way for a team to "take away" the other team's right to score in an equal number of innings -- when a team wins in a walk-off, it's because they scored 5 in 9 innings and the opponent only scored 4 in 9 innings.

Compare that to basketball, where my casual observation (and this study) suggests that the home team receives a disproportionate number of those "take aways" and thus a "structural" advantage.

SJH, the structural advantage in baseball for a home team -- disregarding any comparisons to any other sport -- is meaningful...significant even. The data does, in fact, bear this out. Home teams win at a rate that is much higher than do the visitors.

Home teams do win at a rate higher than visitors do, and you may believe that there is a structural advantage in baseball for the home team, but proving that there is a higher win rate for home teams does not necessarily mean it's the "structural advantage" you hypothesized. It could be any number of different things.

In SoS analyses, I don't see how the h/a split can be ignored. Has anyone made the argument that these splits SHOULD be omitted/ignored?

I agree that an appropriate SoS analysis should measure if "a team had played mostly good teams at home and weak ones on the road." However, I think a SoS analysis that takes H/A splits into account without knowing which teams are played on the home and on the road has the danger of clouding the difference between the field and the opponent and could be less informative than a SoS analysis that ignores H/A altogether.

Edited by sibpin, 17 June 2009 - 11:49 AM.

[RingoOSU]

Or are you saying that playing against the Yanks, Tor, TB, Det, Tex away is the equivalent to playing Bal, Cle, Oak and the Nats at home?

Are you saying that playing against those good teams at home is the equivalent to playing those bad teams on the road?

[ManilaSoxFAN]

Yes, but you're the one just quoting "good teams are .500 on the road" like it's a solid number. It's not. The number varies from team to team and year to year. Some teams do lose over 81 games on a year, and yet play over .500 ball at home. Some don't. The number was given that home teams win 54% of the time compared to 50% for all games. If you have numbers that dispute this, post them.

Sorry, but when did I say that good teams are .500 on the road? I said that they typically play around .500 (hence the ~ sign I used), while doing much better at home. This was in response to a challenge that baseball does/doesn't offer an edge to the home team (and within the context of measuring a given SoS). I was emphasizing the idea that playing/winning at home is (or should be) a significant factor when evaluating SoS.

Are you positing that there is no advantage for a home team? Are you questioning the value of looking at H/A splits when trying to evaluate a team's preformance-to-date within the context of an SoS analysis??

[RingoOSU]

Are you positing that there is no advantage for a home team? Are you questioning the value of looking at H/A splits when trying to evaluate a team's preformance-to-date within the context of an SoS analysis??

Yes.
That 54% number is really too small to make a difference. Teams are bad, or teams are good, that's the point of strength of schedule.

Edited by RingoOSU, 17 June 2009 - 11:53 AM.

[Alternate34]

Wouldn't that be an argument for less of a home field advantage in baseball than in other sports? Each team has 9 innings, goal is to score more runs in 9 innings than the opponent did in 9 innings. There is no way for a team to "take away" the other team's right to score in an equal number of innings -- when a team wins in a walk-off, it's because they scored 5 in 9 innings and the opponent only scored 4 in 9 innings.

The structural advantage would be the strategic advantage from knowing how many runs you would need to score to win in the bottom of the ninth and subbing, pinch hitting, bunting etc. accordingly. This isn't the only "structural advantage" that baseball teams get over other sports teams. Baseball teams home parks have different dimensions. Teams can be contructed to take advantage of that.

The data seems to show that other aspects of home field advantage not deemed structural (psychological, whatever else) have more of an impact.

As for the strength of schedule, it seems that during the season is when you would want to consider home field advantage because teams will have differences in the amount of home games played. At the end of the season, home field advantage would wash out.

[ManilaSoxFAN]

Wouldn't that be an argument for less of a home field advantage in baseball than in other sports? Each team has 9 innings, goal is to score more runs in 9 innings than the opponent did in 9 innings. There is no way for a team to "take away" the other team's right to score in an equal number of innings -- when a team wins in a walk-off, it's because they scored 5 in 9 innings and the opponent only scored 4 in 9 innings.

Compare that to basketball, where my casual observation (and this study) suggests that the home team receives a disproportionate number of those "take aways" and thus a "structural" advantage.



Home teams do win at a rate higher than visitors do, and you may believe that there is a structural advantage in baseball for the home team, but proving that there is a higher win rate for home teams does not necessarily mean it's the "structural advantage" you hypothesized. It could be any number of different things.



I agree that an appropriate SoS analysis should measure if "a team had played mostly good teams at home and weak ones on the road." However, I think a SoS analysis that takes H/A splits into account without knowing which teams are played on the home and on the road has the danger of clouding the difference between the field and the opponent and could be less informative than a SoS analysis that ignores H/A altogether.

Firstly, I have said that I retract any comparisons with other sports. Especially when such comparisons have no effect on an analysis of this one sport. Nonetheless, I maintain that the "last ups" part of baseball is a rather uniquely significant advantage over an equally talented opponent. Is this a problem for anyone? One team can have a walk-off win, the other never can't.

So, moving back to the actual topic...


Of course the H/A split is much less meaningful if it ignores WHO the opponent (good/bad team) was during any of the H/A series.

My whole point is that a good/bad team split is much less meaningful if it ignores WHERE the games were played -- Home or Away.

Ya need to mesh both measurements. THEN one can also add the pythagoreans and such (or vice versa).....

[Myt1]

SJH, the structural advantage in baseball for a home team -- disregarding any comparisons to any other sport -- is meaningful...significant even. The data does, in fact, bear this out. Home teams win at a rate that is much higher than do the visitors.

But you seperated out the structural advantage (i.e. batting last) from the other advantages in your first post on the matter. What data seperates what you call a structural advantage from everything else SJH refers to (sleeping at home, knowing the field, fans, etc.)? The fact that home teams win more games than visitors does not do this.

[smastroyin]

Just for the record, while I think Manila's thesis is wrong in general, for the Theo Epstein Red Sox he has a point, not about structural advantage or anything like that, but about the Red Sox being a great home team. It is interesting that even ex ante we all thought that Theo was doing a good job bringing in guys who could thrive in Fenway, something Duquette/Williams seemed to ignore completely.

2009 .733/.500
2008 .691/.481
2007 .630/.556
2006 .593/.469
2005 .667/.506
2004 .679/.531
2003 .654/.519

2002 .519/.630
2001 .506/.513
2000 .519/.531
1999 .605/.556
1998 .630/.506
1997 .481/.481
1996 .580/.469
1995 .583/.611

[sibpin]

Nonetheless, I maintain that the "last ups" part of baseball is a rather uniquely significant advantage over an equally talented opponent. Is this a problem for anyone? One team can have a walk-off win, the other never can't.

Yes, I continue to have a problem with your expression. The rules are that each team gets an equal number of outs (usually 27) to score runs. This is structurally fair. Walk-off wins only occur because the home team has not extinguished their 27 outs yet.

If you are saying (as Alternate34 does) that a home team receives an emotional advantage from knowing "we only have to score 2 runs to beat them" whereas an away team never knows if the number of runs they scored is enough, I can see that point, but that is not structural. As Myt1 said, there is no way for us to separate this from sleeping at home, knowing the field, or the fans.

We know that a home field advantage exists in baseball. 4.8% over the history of MLB is statistically significant. And while we can find reasons for why that home field advantage exists, it's impossible for any person to identify beyond a reasonable doubt exactly how much each of those reasons contribute to home field advantage, which is why we should not discount one at expense of another.

[ManilaSoxFAN]

But you seperated out the structural advantage (i.e. batting last) from the other advantages in your first post on the matter. What data seperates what you call a structural advantage from everything else SJH refers to (sleeping at home, knowing the field, fans, etc.)? The fact that home teams win more games than visitors does not do this.

The "structural advantage" is a unique one...no? When else does the home team have the last say-so? That said, I never discounted all of the other factors. And I have repeatedly retracted any comparisons with other sports.

WHY a baseball team has an intrinsic advantage over the visitor is of little consequence. The fact that the home field advantage DOES exist is enough for my main point.

The key point (when I was having a discussion with EV about SoS scheds) is that his analysis did not weigh the H/A splits when measuring SoS and opponents records-to-date. Since we all agree that the home team typically has a natural advantage (born out by actual W/L results), I posited that the H/A split should be included...and not to just refer to the opponent's W/L record in a vacuum...when doing an SoS analysis. I'm not sure if I've heard an argument against that position.

[Hairps]

Just for the record, while I think Manila's thesis is wrong in general, for the Theo Epstein Red Sox he has a point, not about structural advantage or anything like that, but about the Red Sox being a great home team.

Unrelated to anything, and in an affront to all of my sabermetric sensibilities, I do the following bit of mental accounting throughout the year to keep track of how the Sox are doing in my head: winning 2/3 of home games and 1/2 of all away games gets a team to 95ish wins.

[ManilaSoxFAN]

Yes, I continue to have a problem with your expression. The rules are that each team gets an equal number of outs (usually 27) to score runs. This is structurally fair. Walk-off wins only occur because the home team has not extinguished their 27 outs yet.

If you are saying (as Alternate34 does) that a home team receives an emotional advantage from knowing "we only have to score 2 runs to beat them" whereas an away team never knows if the number of runs they scored is enough, I can see that point, but that is not structural. As Myt1 said, there is no way for us to separate this from sleeping at home, knowing the field, or the fans.

We know that a home field advantage exists in baseball. 4.8% over the history of MLB is statistically significant. And while we can find reasons for why that home field advantage exists, it's impossible for any person to identify beyond a reasonable doubt exactly how much each of those reasons contribute to home field advantage, which is why we should not discount one at expense of another.

The tactics and strategies are different when the score is the same but the H/A venue is the variable.

Anyway, and again, the reason for the advantage isn't the key point. The fact that the home team DOES, in fact, have an advantage is enough reason to incorporate that differentiation in a SoS analysis.

[Smiling Joe Hesketh]

The "structural advantage" is a unique one...no? When else does the home team have the last say-so? That said, I never discounted all of the other factors. And I have repeatedly retracted any comparisons with other sports.

WHY a baseball team has an intrinsic advantage over the visitor is of little consequence. The fact that the home field advantage DOES exist is enough for my main point.

The key point (when I was having a discussion with EV about SoS scheds) is that his analysis did not weigh the H/A splits when measuring SoS and opponents records-to-date. Since we all agree that the home team typically has a natural advantage (born out by actual W/L results), I posited that the H/A split should be included...and not to just refer to the opponent's W/L record in a vacuum...when doing an SoS analysis. I'm not sure if I've heard an argument against that position.

The last ups structural advantage is not unique to baseball. The home team in hockey gets the last change for shifts after a stoppage, which is probably even more of an advantage than last ups in baseball as it allows the home team to dictate on-ice matchups.

I don't think anyone is arguing that the home field advantage doesn't exist in baseball; in history there's a persistent, measurable advantage enjoyed by the home team. At the same time we are risking overstating such an advantage: it's only about 4.8%. And I would guess that the level of complication in adding such a H/A number to the SoS analysis when the difference really isn't all that large probably makes the analysis too much of a time-consuming effort.

Frankly the studly pitchers the Sox have missed so far this season likely have far more of an impact than any type of H/A analysis anyway. Roy Halladay wins everywhere, not just in Toronto.

[sibpin]

The "structural advantage" is a unique one...no? When else does the home team have the last say-so? That said, I never discounted all of the other factors. And I have repeatedly retracted any comparisons with other sports.

WHY a baseball team has an intrinsic advantage over the visitor is of little consequence. The fact that the home field advantage DOES exist is enough for my main point.

The key point (when I was having a discussion with EV about SoS scheds) is that his analysis did not weigh the H/A splits when measuring SoS and opponents records-to-date. Since we all agree that the home team typically has a natural advantage (born out by actual W/L results), I posited that the H/A split should be included...and not to just refer to the opponent's W/L record in a vacuum...when doing an SoS analysis. I'm not sure if I've heard an argument against that position.

Please stop saying that the "structural" advantage is a "significant" "big" or "huge" one then. It may be "unique" but we have no way of knowing how significant it is.

The argument against the position was made by EV himself, that SoS standings with a complete H/A split would be time consuming to calculate, with the additional argument that SoS standings with incomplete H/A split could be worse than SoS standings without a H/A split. I think we would all agree that standings adjusted for strength of schedule, home/away, day/night, left/right, rain delays, travel time/length, etc. would provide us with a much better idea of team performance than standings adjusted to account for fewer or none of those components.

[ManilaSoxFAN]

Unrelated to anything, and in an affront to all of my sabermetric sensibilities, I do the following bit of mental accounting throughout the year to keep track of how the Sox are doing in my head: winning 2/3 of home games and 1/2 of all away games gets a team to 95ish wins.

Unbeknownst to you (and lucky for you), that is a measurement which is an underlying issue here...at least for some SoSH personalities.

The fact that the Sox have built a team (or several teams) to win at around .500 on the road and .667 at home...for a roughly 95 win season...and that it is therefore a plausible baseline for in-season evaluation of succes/failure, over/under specifically for these Sox...is somehow a contentious issue. Strange.

[Smiling Joe Hesketh]

Unbeknownst to you (and lucky for you), that is a measurement which is an underlying issue here...at least for some SoSH personalities.

The fact that the Sox have built a team (or several teams) to win at around .500 on the road and .667 at home...for a roughly 95 win season...and that it is therefore a plausible baseline for in-season evaluation of succes/failure, over/under specifically for these Sox...is somehow a contentious issue. Strange.

I see what you did there.

IIRC your Metric has been discussed at substantial length in the past here on the main board. I would offer that conflating that issue with the SoS one in discussion here may well prove distracting.

There's already been numerous responses to why it's perhaps not very helpful to combine the two issues (SoS with H/A splits). I think that the two main issues at stake here at SoS so far, with some consideration given to some of the top pitchers the Sox have missed in their series. Getting the H/A split into it (for a seemingly small advantage either way) doesn't strike me as likely to make a huge difference in the final numbers.

This is EV's thread at its heart, so if he'd like to add in the H/A figures I'll defer to him. But my opinion is that the Metric is a separate issue from the one we're talking about here.

[URI]

Unbeknownst to you (and lucky for you), that is a measurement which is an underlying issue here...at least for some SoSH personalities.

The fact that the Sox have built a team (or several teams) to win at around .500 on the road and .667 at home...for a roughly 95 win season...and that it is therefore a plausible baseline for in-season evaluation of succes/failure, over/under specifically for these Sox...is somehow a contentious issue. Strange.

You really need to let it go, and not whine about how some people who were looking for deeper analysis didn't like a counting game that you named after yourself and posted over and over again, especially since the only year that held to your theory is 2005.

[ManilaSoxFAN]

OK, sibpin and SJH -- do your last posts indicate that we all agree that an advantage exists and that it doesn't matter why is exists? That it doesn't matter how baseball compares with different sports?

I hope so.

"I think we would all agree that standings adjusted for strength of schedule, home/away, day/night, left/right, rain delays, travel time/length, etc. would provide us with a much better idea of team performance than standings adjusted to account for fewer or none of those components."

Thanks, sibpin. That was my exact point to EV. Now...out of all of those factors, how big a role do "Opponent" and "H/A splits" play when looking at a given season or a given series? If you propose to add "lefty/righty" to the calcualtion, more power to you...it's all about having a better understanding.

[BroodsSexton]

The "structural advantage" is a unique one...no? When else does the home team have the last say-so? That said, I never discounted all of the other factors. And I have repeatedly retracted any comparisons with other sports.

I see your point, but in quantifying the structural advantage, you have to control for the structural disadvantage of potentially giving up the first run in the top of the first inning. And that's just going to be impossible.

[ManilaSoxFAN]

I see what you did there.

IIRC your Metric has been discussed at substantial length in the past here on the main board. I would offer that conflating that issue with the SoS one in discussion here may well prove distracting.

There's already been numerous responses to why it's perhaps not very helpful to combine the two issues (SoS with H/A splits). I think that the two main issues at stake here at SoS so far, with some consideration given to some of the top pitchers the Sox have missed in their series. Getting the H/A split into it (for a seemingly small advantage either way) doesn't strike me as likely to make a huge difference in the final numbers.

This is EV's thread at its heart, so if he'd like to add in the H/A figures I'll defer to him. But my opinion is that the Metric is a separate issue from the one we're talking about here.

Yeah, what I did there was hypnotize some guy over the internet so that he could come up with this wild concept that a 95 win, 54-27, 41-40 H/A split would be a good thumbnail baseline to monitor the Sox during a season.

IIRC, the metric of monitoring the team's performance vis-a-vis a hypothetical 95-Win season was in place and repeated for years (with positive feedback). You suddenly had something to say about it once you became a Dope. I would offer that it was you (and pals) who became a distraction. Notice how the same kinds of comparitive models have recently, independently shown up on the main board? Like even in this very thread?

Meanwhile, if this is indeed "EV's thread", and he and I were on the same wavelength, and you would like to defer to him, then why not just let us have our discussion?

If you think that the venue is unimportant in a SoS analysis, I can't help you.

[ManilaSoxFAN]

You really need to let it go, and not whine about how some people who were looking for deeper analysis didn't like a counting game that you named after yourself and posted over and over again, especially since the only year that held to your theory is 2005.

You really need to let it go, and not whine about how some posters post what they believe in. I'm not insulting people who think that a 95-win 54-27, 41-40 is a faulty baseline...
providing a rough measurement of the variations thereof...even though the baseline has repeatedly proven to be valid.

[Smiling Joe Hesketh]

Yeah, what I did there was hypnotize some guy over the internet so that he could come up with this wild concept that a 95 win, 54-27, 41-40 H/A split would be a good thumbnail baseline to monitor the Sox during a season.

IIRC, the metric of monitoring the team's performance vis-a-vis a hypothetical 95-Win season was in place and repeated for years (with positive feedback). You suddenly had something to say about it once you became a Dope. I would offer that it was you (and pals) who became a distraction. Notice how the same kinds of comparitive models have recently, independently shown up on the main board? Like even in this very thread?

Meanwhile, if this is indeed "EV's thread", and he and I were on the same wavelength, and you would like to defer to him, then why not just let us have our discussion?

If you think that the venue is unimportant in a SoS analysis, I can't help you.

Sigh. Must we put an emoticon on everything?

As always, any personal problems you have with myself or the modding team can be taken up in Backwash or via PM.

I was trying to gently point out that this is not a thread to discuss your Metric once again, that it would probably prove distracting to the larger topic. I thought I was being pretty evenhanded in noting that many other people pointed out some of the issues involved in trying to add a H/A split into the SoS discussion, particularly the small but consistent overall effect it has on results.

I think some of the questions you asked are very fair, but the responses have been as well. And as a Dope I'm nicely asking that you drop the H/A line of questioning for now, since we seem to have come to the end of fruitful discussion with it. If EV (who did the work at the beginning of the thread) wants to do the labor to add in the splits, he's welcome to do so.

[URI]

You really need to let it go, and not whine about how some posters post what they believe in. I'm not insulting people who think that a 95-win 54-27, 41-40 is a faulty baseline...

You should see the URI metric I invented. It shows that when clouds aren't in the sky, the sky appears to be blue.

But I'm not the one coming off as being persecuted, like you are here:

Unbeknownst to you (and lucky for you), that is a measurement which is an underlying issue here...at least for some SoSH personalities.

The fact that the Sox have built a team (or several teams) to win at around .500 on the road and .667 at home...for a roughly 95 win season...and that it is therefore a plausible baseline for in-season evaluation of succes/failure, over/under specifically for these Sox...is somehow a contentious issue. Strange.

and here:

IIRC, the metric of monitoring the team's performance vis-a-vis a hypothetical 95-Win season was in place and repeated for years (with positive feedback). You suddenly had something to say about it once you became a Dope. I would offer that it was you (and pals) who became a distraction. Notice how the same kinds of comparitive models have recently, independently shown up on the main board? Like even in this very thread?

providing a rough measurement of the variations thereof...even though the baseline has repeatedly proven to be valid.

Right. You were right in 2005 (95 wins, 67%/50% split) and wrong in every other year. Hairps wrote it as a throw away line in a thread. You used it to try and build a personality cult around the fact around the self evident goal of making the playoffs with 95 wins. Saying that the baseline is winning 2/3 at home and 1/2 on the road is just masturbating. You are trying to cum all over the board and then cry when people say that there are better ways of looking at it that don't involve your narcissism.

Either stop bitching about your stupid Metric* or actually contribute to the site (with actual content, not just acting like you or financially).

But I do love when you come out from whatever rock you're under and play with us for a few weeks. It's great fun.

*Not really a metric. Just alliteration to cleverly name counting.

[RingoOSU]

OK let me just help you with what the average 95 win team will do over the course of a season. I've taken all the 95 win teams in the last 10 years to get a sampling.
These teams are:
1999 Texas 51 Home 44 Road wins
2000 Atlanta 51 Home 44 Road wins
2000 St Louis 50 home 45 Road wins
2000 Chicago(AL) 46 Home 49 Road wins
2001 New York (AL) 51 Home 44 Road wins
2002 San Fransisco 50 Home 45 Road wins
2003 Boston 53 Home 42 Road wins
2005 Boston 54 Home 41 Road wins
2005 Los Angeles(AL) 49 Home 46 Road wins
2005 New York (AL) 53 Home 42 Road wins
2006 Detroit 46 Home 49 Road wins
2008 Boston 56 Home 39 Road wins

Total: 50.83 home 44.17 road or .627 vs .545. Now, I'm not sure what those 6 games really amount to over the course of the season, but success on the road is just as important as success at home. Someone better with numbers than I am can determine the margin of error and standard deviations, and the sample size is probably too small. I didn't have time to analyze all teams since 1961. However, your guesstimate seems close to Boston's performance during a 95 win season, but it's not like any of those teams really had success.

[Myt1]

Baseball offers a huge advantage for home teams (vis-a-vis other sports). I can't see how/why that should be underestimated when trying to quantify/analyze SOS.

The reason I said that home teams have a bigger edge in baseball is because of the structure of the game...regardless of other home turf advantages like the playing surface and it's unique environment, the fans, the ability to sleep in one's own bed, etc.

Getting the chance to score and win...without the opponent getting an opportunity to strike back. Never losing immediately because the opponent scored...always having the chance to respond and tie or win. It's like "sudden death" rules/advantages can only apply to the home team. That's the edge I was talking about. It's structural.

I'm not cognizant of such a structural (vs an emotional, psychological) advantage in the other sports you mentioned. But, in pro baseball, almost all teams win between 40% and 60% of their games...while in the other sports the gaps between the top and bottom teams are huge. Some NFL teams win/lose 1 or 2 games in an entire season!

SJH, the structural advantage in baseball for a home team -- disregarding any comparisons to any other sport -- is meaningful...significant even. The data does, in fact, bear this out. Home teams win at a rate that is much higher than do the visitors.

Anyway, and again, the reason for the advantage isn't the key point. The fact that the home team DOES, in fact, have an advantage is enough reason to incorporate that differentiation in a SoS analysis.

Unbeknownst to you (and lucky for you), that is a measurement which is an underlying issue here...at least for some SoSH personalities.

The fact that the Sox have built a team (or several teams) to win at around .500 on the road and .667 at home...for a roughly 95 win season...and that it is therefore a plausible baseline for in-season evaluation of succes/failure, over/under specifically for these Sox...is somehow a contentious issue. Strange.

I don't know about any of the other stuff you're complaining about. But complaining about a red herring when you're the one who reeled it in is bothersome.

The contentious issue is that, as evidence that the failure to include H/R corrections was a mortal and not merely a venial sin, you stated that baseball had a greater homefield advantage than other sports. This was shown to be wrong. Then you shared your belief that baseball had a sort of structural advantage that was unique, while other sports had merely emotional/home cooking type advantages. This was also show to be wrong. Finally, you claimed that the structural advantage in baseball was significant and that the data, in fact, bore this out. This was also shown to be wrong (or, more precisely, unproven by any data, including that you cited).

H/R corrections would be useful right now. EV agreed with you. I don't think anyone is seriously arguing that home teams do not win more games than away teams. But that's all it is. Your discussion of the structural issues and components of home field advantage are wrongheaded.

It's like "sudden death" rules/advantages can only apply to the home team.

This quasi-sudden death statement is the clearest indicator of this. It's not sudden-death when both teams get the opportunity for the same amount of chances. And, even if it were, other sports also have this structure: in hockey, the hometeam shoots last in a shootout. In fact, this is actually the exact opposite of sudden death.

The structural advantage occurs only in strategy for the last innings. The away team may have some minisculely greater incentive to play for 2 runs in the top of the 9th of a tie game than the home team would in the bottom. At the same time, defensive benefits of being the away team (e.g. knowing to play infield in, OF shallow or "no-doubles," or even 5 infielders) represent structural advantages that at least partially offset the structural advantage to being the home team.

Just pointing to the historical homefield advantage would have been sufficient to make your point. You sidetracked the discussion with an incorrect statement, sidetracked it some more with incorrect defenses of that incorrect statement, and then derailed it completely with the martyr act.

Edited by Myt1, 17 June 2009 - 02:48 PM.