NL central race based on runs scored vs. allowed...

[DivineBovine]

From baseball prospectus, using a formula called "pythagenport", what the standings would look like through yesterday if WL reflected runs scored vs. runs allowed:

 

RS RA W L GB

Cubs (65-60) 584 532 68 57 0.0

Brewers (65-62) 606 615 63 64 6.0

Cards (61-62) 554 613 56 67 11.0

 

 

Based on this formula, the Brewers should be 6 back and the Cards have no business in the race at all. Are the Cubbies really cursed because it's bad luck or are they just not clutch?

[killthegoat06]

Shouldn't this be in Baseball Discussions? Plus you'd probably get more replies, because I know some people don't even check the Rivalries Forum.

[soccer10k]

According to that the Diamondbacks should be in last place in the NL West.

[CubinNY]

Pythagorean method is not an exact science, nor is it a way to determine "luck". According to Rob Neyer it is a way of looking at whether teams under preformed or over preformed based on its runs scored and runs allowed.

 

At the end of the season most teams are about +/- 4 wins from their pythagorean record. All three teams are within measurement error.

 

Therefore, there is no need to lament bad luck or curses.

[Banedon]

Therefore, there is no need to lament bad luck or curses.

 

Welllll....there may be a need....but just not because of pythagenport... :P

[Soul]

If the stats haven't put us in our rightful place by August 24th, then I'd say there's a pretty good chance they simply won't be a good predictor in this case.

[CubinNY]

If the stats haven't put us in our rightful place by August 24th, then I'd say there's a pretty good chance they simply won't be a good predictor in this case.

But PM does have the Cubs in their "rightful place" and have been a good predictor.

 

Again, this is not an exact science. Each of the three teams are within measurment error (+/- 4 wins).

 

EDIT: Cards are +5, but that is close enough.

[Soul]

If the stats haven't put us in our rightful place by August 24th, then I'd say there's a pretty good chance they simply won't be a good predictor in this case.

But PM does have the Cubs in their "rightful place" and have been a good predictor.

 

Again, this is not an exact science. Each of the three teams are within measurment error (+/- 4 wins).

 

EDIT: Cards are +5, but that is close enough.

 

Just throwing it out there, but don't you think +/- 4 wins is a pretty big margin of error? That's an 8 game swing, the difference between a pretty good club and an also-ran, or the difference between a mediocre club and a poor one.

[CubinNY]

If the stats haven't put us in our rightful place by August 24th, then I'd say there's a pretty good chance they simply won't be a good predictor in this case.

But PM does have the Cubs in their "rightful place" and have been a good predictor.

 

Again, this is not an exact science. Each of the three teams are within measurment error (+/- 4 wins).

 

EDIT: Cards are +5, but that is close enough.

 

Just throwing it out there, but don't you think +/- 4 wins is a pretty big margin of error? That's an 8 game swing, the difference between a pretty good club and an also-ran, or the difference between a mediocre club and a poor one.

 

Not in a 162 game season. PM is not really used to judge a good team or bad team. It is simply a measurement model. It's a model that has shown to closely approximate true values. But with any measurement model random fluctuations in true values make measurement error inherent in the model.

 

It's like weather forcasting or stock speculation or anything else that uses a model to make predictions.

 

I haven't looked up PM data in the last few years so the measurement error might be a little different, but I suspect it will always be pretty close to what it was.

[Guest]

I can't look it up right now, but I've seen a pretty significant study that says pythagorean record is a better predictor of future record than actual w/l.

[CubinNY]

I can't look it up right now, but I've seen a pretty significant study that says pythagorean record is a better predictor of future record than actual w/l.

I guess that's good news for the Cubs unless the Brewers score in double digits for a few games.

[Ender]

Cubs have been lucky with RS this year which is what you are really seeing here. Looking at W3/L3 they should be 63-63, Brewers should be 63-64. The two teams are pretty even overall.

[Rob]

There was an interesting article at the Hardball Times a few days ago regarding the difference between teams and their runs scored and runs allowed.

 

http://www.hardballtimes.com/main/article/no-mirage-in-arizona/

 

Essentially, a bad fifth starter makes it easier to outperform a pythagorean record, while a good one makes it much harder.

[CubinNY]

There was an interesting article at the Hardball Times a few days ago regarding the difference between teams and their runs scored and runs allowed.

 

http://www.hardballtimes.com/main/article/no-mirage-in-arizona/

 

Essentially, a bad fifth starter makes it easier to outperform a pythagorean record, while a good one makes it much harder.

 

I'm as numbers oriented as anyone. Heck I'm a scientist. That's why qoutes like the below drive me up the freaking wall.

 

Differences between Real W/L and Pythag W/L are dismissed as mere chance. Heck, in B-ref’s expanded schedules, the site flatly categorizes any variation as luck.

 

No scientist worth his salt would make such a statement. There is measurement error in any model. And variation is not the same as luck. One does not call everything that cannot be explained by a model luck.

 

I'm glad that the author took the time to invistigate when and why the model can breakdown.

 

The point is to see how well a model fits reality, not the other way around.

Edited August 25, 2007 by CubinNY

[Mizzou]

The Rangers-Orioles game the other day is a good reason why these things don't hold a lot of water.

[BigSlick]

There was an interesting article at the Hardball Times a few days ago regarding the difference between teams and their runs scored and runs allowed.

 

http://www.hardballtimes.com/main/article/no-mirage-in-arizona/

 

Essentially, a bad fifth starter makes it easier to outperform a pythagorean record, while a good one makes it much harder.

 

I'm as numbers oriented as anyone. Heck I'm a scientist. That's why qoutes like the below drive me up the freaking wall.

 

Differences between Real W/L and Pythag W/L are dismissed as mere chance. Heck, in B-ref’s expanded schedules, the site flatly categorizes any variation as luck.

 

No scientist worth his salt would make such a statement. There is measurement error in any model. And variation is not the same as luck. One does not call everything that cannot be explained by a model luck.

 

I'm glad that the author took the time to invistigate when and why the model can breakdown.

 

The point is to see how well a model fits reality, not the other way around.

 

Aren't we arguing semantics here? Some people would call variation "luck" others would not.

[seanimal]

I'm as numbers oriented as anyone. Heck I'm a scientist. That's why qoutes like the below drive me up the freaking wall.

 

Differences between Real W/L and Pythag W/L are dismissed as mere chance. Heck, in B-ref’s expanded schedules, the site flatly categorizes any variation as luck.

 

No scientist worth his salt would make such a statement. There is measurement error in any model. And variation is not the same as luck. One does not call everything that cannot be explained by a model luck.

 

Really, it's neither here nor there, but isn't that exactly what the authors are stating, that the typical response in these situations is that it's luck?

 

I'm glad that the author took the time to invistigate when and why the model can breakdown.

 

The point is to see how well a model fits reality, not the other way around.

 

Very true, and an excellent article.

[CubinNY]

There was an interesting article at the Hardball Times a few days ago regarding the difference between teams and their runs scored and runs allowed.

 

http://www.hardballtimes.com/main/article/no-mirage-in-arizona/

 

Essentially, a bad fifth starter makes it easier to outperform a pythagorean record, while a good one makes it much harder.

 

I'm as numbers oriented as anyone. Heck I'm a scientist. That's why qoutes like the below drive me up the freaking wall.

 

Differences between Real W/L and Pythag W/L are dismissed as mere chance. Heck, in B-ref’s expanded schedules, the site flatly categorizes any variation as luck.

 

No scientist worth his salt would make such a statement. There is measurement error in any model. And variation is not the same as luck. One does not call everything that cannot be explained by a model luck.

 

I'm glad that the author took the time to invistigate when and why the model can breakdown.

 

The point is to see how well a model fits reality, not the other way around.

 

Aren't we arguing semantics here? Some people would call variation "luck" others would not.

 

Some might, I wouldn't unless I could rule out all other variables (like this article has done).

 

Either way, PM is a pretty accurate molar model of Ws and Ls. It breaksdown at the local level of analysis (game by game).

[jjgman21]

Pythagorean method is not an exact science, nor is it a way to determine "luck". According to Rob Neyer it is a way of looking at whether teams under preformed or over preformed based on its runs scored and runs allowed.

 

At the end of the season most teams are about +/- 4 wins from their pythagorean record. All three teams are within measurement error.

 

Therefore, there is no need to lament bad luck or curses.

 

isn't outperforming the expected record based on runs scored/allowed pretty close to the definition of luck? usually when that term is thrown about around here it is to summarize phenomena that can't be proven through various statistical methods. so based on your assertion that it is not luck, does the inverse apply? is there any explanation for this phenomena besides luck?

 

the Cards in the standings posted above are already outside the measurement error. if you consider the rate that all three teams are going, the Cubs and Cards will both be well outside of the measurement error upon completing 162 games, the Brewers on the cusp.

 

 

edit - I don't think I saw there was a second page of posts before writing this, but I'm still not clear on what you were saying here.

[CubinNY]

Pythagorean method is not an exact science, nor is it a way to determine "luck". According to Rob Neyer it is a way of looking at whether teams under preformed or over preformed based on its runs scored and runs allowed.

 

At the end of the season most teams are about +/- 4 wins from their pythagorean record. All three teams are within measurement error.

 

Therefore, there is no need to lament bad luck or curses.

 

isn't outperforming the expected record based on runs scored/allowed pretty close to the definition of luck? usually when that term is thrown about around here it is to summarize phenomena that can't be proven through various statistical methods. so based on your assertion that it is not luck, does the inverse apply? is there any explanation for this phenomena besides luck?

 

the Cards in the standings posted above are already outside the measurement error. if you consider the rate that all three teams are going, the Cubs and Cards will both be well outside of the measurement error upon completing 162 games, the Brewers on the cusp.

 

 

edit - I don't think I saw there was a second page of posts before writing this, but I'm still not clear on what you were saying here.

 

Lets say a team over-preforms and that over-preformance is far beyond measurement error. From the standpoint of science, one could to say, "the team over-prefermed the PR and I don't know why. It may be due to luck or it may be due some variable that PM hasn't accounted for."

 

I would think the closer the team is to the measurement error (+/-4) the more likely the results are to luck.

 

Be that as it may, enough games have been played that I suspect the Cardinals, the Cubs, and Brewers will finsih pretty close to their PR. It makes me optimisitic about the Cubs chances.

[Ender]

In Clay Davenport's latest chat he mentioned that at around 140 games into the season real records are statistically more accurate than pythagorean records. He didn't give any numbers to back up that statement but he's generally a pretty knowledgable guy when it comes to numbers so I assume he is correct.

[SouthSideRyan]

The Rangers-Orioles game the other day is a good reason why these things don't hold a lot of water.

 

Yeah those 27 run losses every week make it pretty much worthless.

[Rob]

In Clay Davenport's latest chat he mentioned that at around 140 games into the season real records are statistically more accurate than pythagorean records. He didn't give any numbers to back up that statement but he's generally a pretty knowledgable guy when it comes to numbers so I assume he is correct.

 

Actually, he was comparing the standard playoff odds report with the PECOTA version. Typically, the PECOTA one works by virtue of a weighted mean between real records and the PECOTA predicted record. All he was saying was that, this late into the season, the actual record dominates most of that number.