Pythagoras Frowned
Somewhere, about 3.5 clicks under the rainbow, there is a mystical place called Stat Heaven. It is a benevolent paradise ruled by kindly, all-knowing computers, where bloggers find respite from the toil of their mother's basements. It is a place where baseball is color-blind and Barry Bonds hasn't been pushed out of the sport (while Andy Pettitte and other alleged users play on unmolested). From Stat Heaven, the mighty and wise Pythagoras of Samos looks down at the state of the playoff race and frowns.
Division Leaders
Tampa Bay Rays 77-48 (.616 Winning %)
Chicago White Sox 72-53 (.576)
Los Angeles Angels of Marketing Gimmick 76-48 (.613)
Wild Card Leaders
Boston Red Sox 73-53 (.579)
Minnesota Twins 71-54 (.568)
New York Yankees 66-59 (.528)
One way to evaluate team performance is Pythagorean record, which is more accurate in predicting future success than actual record is. Pythagorean record is calculated based on total runs scored and runs allowed. The theory goes that good teams will amass many runs for themselves while keeping opponents' from scoring; bad teams give up more runs than they score. By this standard, the playoff scene would be quite different.
Pythagorean Playoffs
Red Sox 75-51 (.594)
White Sox 73-52 (.581)
Angels 68-56 (.548)
Rays 71-54 (.566)
Twins 68-57 (.545)
NYY 67-58 (.536)
Based on Pythagorean expectation, the Red Sox would have two more wins than they currently do, while the Rays would have six fewer wins, placing the Sox atop the division. LAA and Tampa are both outperforming their Pythagorean expectation - the Angels are a whopping 8 games above it. Chicago would have a more secure hold on their division, gaining a game while the Twins lose three games. The Yankees only gain a game, and would still be playoff long-shots.
Too much of a divergence between Pythagorean and actual record can mean that a team is benefiting unduly from luck, and is due for a fall. An example of this is the 2007 Seattle Mariners. They went 88-74 that year, at one point threatening the NYY in the Wild Card race; however, they were outperforming their Pythagorean expectancy. The team's final Pythagorean record was 79-83, a signs that the team was lucky, as opposed to talented. This year, the Mariners have collapsed, with a 46-79 record (2nd worst in baseball). Pythagorean record can also reveal teams victimized by bad luck. The 2006 Indians had a 78-84 record, 11 games worse than their Pythagorean record (89-93). The next year they went 96-66 and were a win away from making the World Series.
It's possible to outperform Pythagorean expectation, particularly by having a strong bullpen to preserve 1-run leads. The Rays and Angels have good pens, as did the 07 Mariners. By contrast, the 2006 Indians had an abominable bullpen, which accounted for many of their losses, and much of their success the following year came from improved pen performances (namely the Rafaels Betancourt and Perez).
Sadly, we do not live in the Pythagorean world - we live in the one where the Red Sox are in 2nd place in their division, rather than in possession of the best record in the AL. Still, these numbers have some implications for the playoffs and beyond. The Angels and Rays look a lot less intimidating by Pythagorean record, while the Sox look a little better. Heading into next year, we can expect some major regression from the Angels, although probably not enough to keep them from the playoffs in their weak division. The Rays may regress, since much of their success is based on their strong bullpen (Balfour and Howell are having career years), but their core of young talent should keep them a 90+ wins team.
One final observation: as Theo's reconstruction of the team has gone forward, the team has been unlucky. From 2003-2006, the Sox consistently beat their Pythagorean expectation, by as many as 5 games in '05 and '06. The 2007 and 2008 squads, which are largely composed of Theo's desired players, have both played worse than their Pythagorean records (5 games in 07, 2 this year). This is probably nothing more than random variation and luck. Still, it's frustrating to see our increasingly home-grown teams undershoot their projected records.
I'm curious what you guys think of all this. Do numbers lie, or are the Red Sox really the best team in the league? Are the Angels and Rays this good, or are they benefiting from luck or unsustainable performances? Should Theo petition MLB to make Pythagorean record the sole criterion for getting into the playoffs? Would you enjoy Stat Heaven? Answer all of these in the comments below.
* I used an exponent of 1.83 in the equation, which is what Baseball-Reference employs. Numbers were cracked last night, before B-R had updated, so my W-L figures should be the same as their's.
0 recs |
18 comments
Comments
theory
The Pythagorean calculations have always been constrained by the limited applicability. While runs scored can be reasonably be expected to be divided evenly per game, runs allowed cannot. A team with three aces and two holes in the rotation may give up the same runs/game as a team with five decent starters. But will they have the same winning percentages? It’s not clear.
The so-called Pythagorean rule is simply a hypothesis. If the hypothesis fails to describe the data adequately, it seems more reasonable to reject the hypothesis than to decide the hypothesis must be correct, and criticize the data.
Or to put it another way: the Pythagorean rule is good at giving an estimated number of wins. But it is equally valid to attribute the differrence between the estimated number of wins and the actual number of wins to the weakness of the model as to ‘luck’.
by RickD on Aug 20, 2008 1:54 PM EDT reply actions 0 recs
your last paragraph is confusing -- I believe most people would agree that the difference between Pythag and actual wins is mostly luck
Throughout much of a season, Pythagorean record is a better predictor of future winning percentage than actual record.
Bullpens do make a difference — the Angels bullpen, by comparing WPA to WPA/LI gets about four games worth of credit, leaving five games to be explained. The Rays bullpen explains about 4 of their 6 “extra” wins. (Of course, some of the difference betweeen WPA and WPA/LI isn’t a skill, either, but that’s a different story.)
my blog // calling BJ Upton lazy is lazy // past results do not guarantee future performance
by Sky Kalkman on Aug 20, 2008 2:15 PM EDT up reply actions 0 recs
why would most people believe that luck is the factor?
Unless you have some external reason to think that wins should match Pythagorean wins, the discrepancy is more likely to be due to the inadequacies of the model, than it is for the model to be judged as definitely true, and any discrepancy from it to be due to luck.
Let me rephrase this: I do mathematical modeling for a living. No model ever describes reality perfectly. It is presumptuous to say that variation from any model is due to luck, as opposed to simple inadequacies in the model.
by RickD on Aug 20, 2008 4:06 PM EDT up reply actions 0 recs
Interesting point.
Pythagorean record has a stronger correlation with future performance than actual record does, i.e. it has better predictive value. Obviously it’s not going to consistently match wins, but I think it’s fair to attribute major discrepancies (like a difference of 8+ games in 162*) to “luck.”
Also, the model has been tweaked repeatedly to be more accurate. If it had no predictive value, I’m sure people would’ve abandoned it, rather than trying to make it better.
*I’m not sure what counts as statistically significant in this sample size, so this is just a guess.
by 0157H7 on Aug 20, 2008 4:39 PM EDT up reply actions 0 recs
i agree.
but I think it’s fair that most people who understand Pythag think that the difference is luck. although most now realize that a properly leveraged bullpen makes a difference too. i’m not saying the luck explanation is right or wrong, just what i tend to see out on the ’nets.
my blog // calling BJ Upton lazy is lazy // past results do not guarantee future performance
by Sky Kalkman on Aug 20, 2008 8:33 PM EDT up reply actions 0 recs
can't we just blame hansen?
i mean seriously
by EWS1532 on Aug 20, 2008 5:07 PM EDT reply actions 0 recs
And Lugo, and Cora, and Lopez, and Buch, and Coco...
The list goes on…
Don't question my choice of teams, or I will have to go all troll on you.
by bs.uf15bosox9bears23 on Aug 20, 2008 9:05 PM EDT up reply actions 0 recs
Pythagorean
Is a great tool for Fantasy sports but real life has to many variables to account for, to me it makes it a poor indicator of future success. Do I look at and consider it, sure I do. But sometimes what you see is what you get and sometimes what the numbers say isn’t what you get. The truth is numbers don’t lie but liars have numbers.
Baseball is God's sport! All Truth Goes Through Three Stages 1.It is ridiculed 2.It is violently opposed 3.Finally, it is accepted as self-evident. kinesiologist
by E5 on Aug 20, 2008 6:47 PM EDT reply actions 0 recs
yes
We have the best team. Yes, your evaluation is spot on in that pythag records is a great predictor of future success or failure. And yes, you are correct in pointing out that you can outperform it by having a good bullpen.
But not just a good closer, you need excellent work from your guys working the 6th and 7th. This has never been the strongpoint of the Sox and that’s why they always under perform their pythag. Our 6 and 7th innning guys always cough up runs when we are behind, usually turning a 1 or 2 run game into a 4-5 run deficit. Then the offense will score a couple in the 8th or 9th and we still lose by 3, but making it close in the pythag. Teams like LAA and ’zona last year have/had guys who keep the games close, then if you scarper a run or two in the last couple of frames you either win or go extra ininngs.
Well that’s my theory based on watching this organisation for 40 years…
by sydneysox on Aug 20, 2008 6:59 PM EDT reply actions 0 recs
actually, the guys working the 6th and seventh usually aren't used in situations that are much more important than average situations
teams that outperform their Pythag have closers and top setup guys who are significantly better than the rest of the bullpen (not simply good overall, but better than the rest) and back-end guy who are significantly worse than the others members of the pen. preventing a few runs at important times is worth the same as preventing two to three times as many runs at non-important times. and giving up tons of runs in a blowout matters zilch. (it’s the same logic as clutch hitting, except that leveraging a bullpen is a team skill.)
my blog // calling BJ Upton lazy is lazy // past results do not guarantee future performance
by Sky Kalkman on Aug 20, 2008 8:36 PM EDT up reply actions 0 recs
Anomoly
When we see a variation in “Pythags” record thereom, we attribute it to luck, but it is really an anomoly. We do not know the full mechanism that causes this fluctuation from the norm. There are alot of theorys going around, one of which is “luck”. I don’t firmly believe this, I agree with RickD and that we don’t actually know what is causing this deviation from the norm so we associate it with luck. This method isn’t very scientific.
On the other hand, the realization by us in this blog (and most likely others else where) that a good bullpen can change Pythags norm. This seems to be a nice scientific observation with some proof behind it.
So what am I really trying to say……. the closer role is stupid. It has been proven by a bunch of smart people like Bill James that the end of the game isn’t always the most crucial point for a relief pitcher. If managers put in relief pitchers in the situations they are supposed to be in I think there will be alot less deviation from Pythags Thereom.
As much as I love Tito and the awesome job he does with the team, I think most of us can agree he doesn’t always bring in the right releiver at the right time.
by drabidea on Aug 21, 2008 10:56 AM EDT up reply actions 0 recs
Who is the right reliever at the right time?
When whoever Tito puts in gives up runs, then it’s impossible to succeed at bullpen management. We have a bunch of guys who are often completely unable to find the strike zone or record outs. Aardsma, Delcarmen, Timlin, and recently Lopez have all been very hittable lately. I’d like to see more of Okajima in the next few games. Aside from the stinker against Toronto he’s been pretty good.
by 0157H7 on Aug 21, 2008 11:33 AM EDT up reply actions 0 recs
The right time
In the 7th inning of a 2-2 ball game with two on and one out. You put in Papelbon. That is a high leverage situation. Paps is the best at high leverage. He will shut out the inning and you take your chances for the 9th.
If you can’t win the 7th there is no hope for the 9th.
by drabidea on Aug 21, 2008 2:15 PM EDT up reply actions 0 recs
I like the idea of a Jamesian relief ace
but only if we can get or train someone reliable to pitch the 9th. A Joe Borowski / Todd Jones type, who throws lots of strikes and finishes games off reliably. In our current pen, I might trust Okajima in those situations, but everyone else is too unreliable. As it stands, having Paps as the closer is a good way to limit his overall innings.
Also, I don’t think Papelbon or Francona would go for it. Paps sees $$ in those saves totals, and would complain (and possibly perform worse) if moved from the role, and Francona is too enamored of the closer idea to move Paps out of that role. Francona will use setup men as relief aces – i.e. Okajima last year – but only if he has an established closer to finish games off.
by 0157H7 on Aug 21, 2008 11:22 PM EDT up reply actions 0 recs
Agreed
I don’t know how BIlly Beane seems to groom a “closer” every year who isn’t even his best relief pitcher. I agree with Francona and Papelbon not wanting to move from the closer role. I would say Timlin of yesteryear would be a great person for the Jamesian closer but this year I just can’t trust him enough either.
Until Paps and Tito can set aside their differences or we miraculously get better relief pitching I am not going to expect seeing a change in our “luck”.
by drabidea on Aug 22, 2008 9:12 AM EDT up reply actions 0 recs
Paps would still see a fair amount of action in the ninth inning
He’d pitch all the one-run saves, most of the two-run saves, but leave the 3-run leads to someone else. If you can’t trust Oki or someone else with a three run lead in the ninth, then why are they pitching with smaller leads in the seventh or eighth?
The Jamesian model does NOT want Todd Jones pitching in ANY close game, let along a close game in the ninth.
my blog // calling BJ Upton lazy is lazy // past results do not guarantee future performance
by Sky Kalkman on Aug 22, 2008 3:15 PM EDT up reply actions 0 recs
I wish I had been less sarcastic when presenting this info on DRaysBay...
Maybe more than one person would have taken my presentation seriously then…
Don't question my choice of teams, or I will have to go all troll on you.
by bs.uf15bosox9bears23 on Aug 20, 2008 9:06 PM EDT reply actions 0 recs

by 


















