01 Sep 12
Originally posted by sh76What of the 1990 Reds?
There are always going to be deviations of course, but run differential has historically proven to be an excellent indicator of how well a team will do.
I admit to not fully understanding the math, but this paper shows that the Pythagorean method of predicting record works very well:
http://arxiv.org/pdf/math/0509698.pdf
James himself has also written to defend this mechanism and I have not seen his assumptions challenged in any serious way.
I think this team is easily a better team than that team.
01 Sep 12
Originally posted by sh76The Orioles obviously need some advanced mathematicians to inform them that they shouldn't drub the Yankees like they did last night.
There are always going to be deviations of course, but run differential has historically proven to be an excellent indicator of how well a team will do.
I admit to not fully understanding the math, but this paper shows that the Pythagorean method of predicting record works very well:
http://arxiv.org/pdf/math/0509698.pdf
James himself has also written to defend this mechanism and I have not seen his assumptions challenged in any serious way.
02 Sep 12
3 edits
Originally posted by whodeyIt probably is. The 1990 was not that good. IIRC, it went 35-15 early and played .500 ball for the rest of the season. They beat a better Pitt team in the NLCS and a much better A's team in the WS. That sometimes happens in baseball, anomalous as it is.
What of the 1990 Reds?
I think this team is easily a better team than that team.
That they might have been better than they 1990 obviously does not mean that this Reds team is a great team (it is probably not) or that it will win the WS this year (the odds are fairly long on that one too).
02 Sep 12
1 edit
Originally posted by no1marauderNo, but I know all about the Orioles' poor run differential. It will likely catch up to them in September. But even if it doesn't, anomalies happen. That does not impugn the validity of the general trend.
Did you bother to read the link I provided concerning the Orioles?
02 Sep 12
Originally posted by no1marauderOkay, I read it.
Did you bother to read the link I provided concerning the Orioles?
The O's good record is probably at least partially attributable to good luck.
If heart and bullpen and fortitude and leadership and all that jazz were really the difference between a good performance relative to your Pythagorean record and a bad one, there would be consistency from year to year. There is none. Last year, for example, the Reds played 4 games under their Pythagorean record. The difference is that last year they were probably a little unlucky and this year they're probably a little lucky.
02 Sep 12
Originally posted by sh76How many "anomalies" per year does a trend make? You offered no link supporting your claim that early season run differential is a reliable predictor of second half results though I've heard that claim before. However, the Orioles are a glaring exception. And the Cardinals run differential has been better than the Reds all season, so St. Louis should have taken charge of the division after the All-Star break. In fact, the opposite has occurred. I haven't followed other divisions as closely but it seems to me that this "general trend" isn't very strong in 2012.
No, but I know all about the Orioles poor run differential. It will likely catch up to them in September. But even if it doesn't, anomalies happen. That does not impugn the validity of the general trend.
02 Sep 12
1 edit
Originally posted by no1marauderOkay, let's look at the Reds. I'll give you each year and their play relative to their Pythgorean record.
How many "anomalies" per year does a trend make? You offered no link supporting your claim that early season run differential is a reliable predictor of second half results though I've heard that claim before. However, the Orioles are a glaring exception. And the Cardinals run differential has been better than the Reds all season, so St. Louis should hav isions as closely but it seems to me that this "general trend" isn't very strong in 2012.
2012: +5
2011: -4
2010: - 1
2009: +2
2008: +2
2007: -3
Total: +1 with no discernible trend
Let's try the Yankees:
2012: 0
2011: -4
2010: -2
2009: +8
2008: +2
2007: -3
total: +1 with no trend and wild swings
These are just teams I picked at random because they're the teams you and I follow, but I'm sure I could keep going and find similar results.
The two things that are shown are:
1) Pythagorean record is an excellent indicator of record in the long term.
2) There is no discernible trend from year to year. The fluctuations have all the marks of statistical noise.
02 Sep 12
Originally posted by sh76Yes, we were "unlucky" that our bullpen was 12th in the NL in runs allowed per 9 IP and 12th in percentage of saves blown (37😵. This year we have been "lucky" that our relief pitchers are 2nd best in RP9 and 5th best in BSVPCT (26😵.
Okay, I read it.
The O's good record is probably at least partially attributable to good luck.
If heart and bullpen and fortitude and leadership and all that jazz were really the difference between a good performance relative to your Pythagorean record and a bad one, there would be consistency from year to year. There is none. Last year, for example, the ...[text shortened]... t last year they were probably a little unlucky and this year they're probably a little lucky.
02 Sep 12
Originally posted by sh76The claim, I thought, was that it had predictive value. It's uninteresting that in the long run it roughly matches team's won-loss - the equation is designed to. Yes, run every team's Pythagorean and total the individual variations and by an amazing coincidence it will equal EXACTLY zero every year. It is your claim that that means anything?
Okay, let's look at the Reds. I'll give you each year and their play relative to their Pythgorean record.
2012: +5
2011: -4
2010: - 1
2009: +2
2008: +2
2007: -3
Total: +1 with no discernible trend
Let's try the Yankees:
2012: 0
2011: -4
2010: -2
2009: +8
2008: +2
2007: -3
total: +1 with no trend and wild swings
These are just te ...[text shortened]... cernible trend from year to year. The fluctuations have all the marks of statistical noise.
02 Sep 12
Originally posted by sh76And as for "wild swings" even in that small sample you had two instances where the deviation from the Pythgorean prediction from one year to the next exceeded 5% (2011-2012 Reds, 2009-2010 Yankees). I'm guessing that is statistically significant.
Okay, let's look at the Reds. I'll give you each year and their play relative to their Pythgorean record.
2012: +5
2011: -4
2010: - 1
2009: +2
2008: +2
2007: -3
Total: +1 with no discernible trend
Let's try the Yankees:
2012: 0
2011: -4
2010: -2
2009: +8
2008: +2
2007: -3
total: +1 with no trend and wild swings
These are just te ...[text shortened]... cernible trend from year to year. The fluctuations have all the marks of statistical noise.
02 Sep 12
Originally posted by no1marauderOf course it's means something. The equation itself draws only on run differential, which you downplayed a little earlier and has been shown to predict almost precisely, in the long term, a team's record. That shows that there's an excellent correlation between run differential and wins and losses. If your supposition that run differential is not closely related to record is correct, there would be no way to write an equation that would correlate run differential with record so closely.
The claim, I thought, was that it had predictive value. It's uninteresting that in the long run it roughly matches team's won-loss - the equation is designed to. Yes, run every team's Pythagorean and total the individual variations and by an amazing coincidence it will equal EXACTLY zero every year. It is your claim that that means anything?
Keep in mind that the 6 years in my sample all took place well after the Pythagorean equation was written. It's not like the sample was included in the historical data that was used to generate the equation.