PDA

View Full Version : Baseball Prospectus is insane


Fenway
05-31-2007, 01:27 PM
As some of you know BP predicts how the season will turn out using PECOTA (Player Empirical Comparison and Optimization Test Algorithm) (http://en.wikipedia.org/wiki/PECOTA)

Their prediction for the White Sox is laughable

Chance of winning division .60477
Chance of wild card 1.20902
To make playoffs 1.81378


http://www.baseballprospectus.com/statistics/ps_oddspec.php

itsnotrequired
05-31-2007, 01:37 PM
As some of you know BP predicts how the season will turn out using PECOTA (Player Empirical Comparison and Optimization Test Algorithm) (http://en.wikipedia.org/wiki/PECOTA)

Their prediction for the White Sox is laughable

Chance of winning division .60477
Chance of wild card 1.20902
To make playoffs 1.81378


http://www.baseballprospectus.com/statistics/ps_oddspec.php

White Sox, .500, 1.81% chance of making playoffs

Yankees, 7 games under .500, 19.53% chance of making playoffs

:rolleyes:

IndianWhiteSox
05-31-2007, 01:38 PM
As some of you know BP predicts how the season will turn out using PECOTA (Player Empirical Comparison and Optimization Test Algorithm) (http://en.wikipedia.org/wiki/PECOTA)

Their prediction for the White Sox is laughable

Chance of winning division .60477
Chance of wild card 1.20902
To make playoffs 1.81378


http://www.baseballprospectus.com/statistics/ps_oddspec.php

:?:

Why are you surprised? This is typical BP bull****.

CLR01
05-31-2007, 01:39 PM
White Sox, .500, 1.81% chance of making playoffs

Yankees, 7 games under .500, 19.53% chance of making playoffs

:rolleyes:


Even the WSI softball team is at 3%

Chisox353014
05-31-2007, 01:39 PM
According to BP, we still only have a 98% chance of winning the 2005 AL Central.

Fenway
05-31-2007, 01:42 PM
White Sox, .500, 1.81% chance of making playoffs

Yankees, 7 games under .500, 19.53% chance of making playoffs

:rolleyes:

and the Devil Rays who have the same EXACT record as NYY .21686 chance

what a farce

of course I will agree with the Boston number 97.55666 :tongue:

DumpJerry
05-31-2007, 01:47 PM
Even the WSI softball team is at 3%
We traded Chips for that Gregory Pratt guy.

Luke
05-31-2007, 01:55 PM
Same old garbage from them.

Sox It To Em
05-31-2007, 02:16 PM
Six percent to win the division seems somewhat reasonable to me, but only a 1.8% chance to make the playoffs? How can our chance at winning the wild card be lower than our chance at making the playoffs?

PECOTA sucks. Stats are something that should be used to evaluate what happened, not to predict what will happen. There's just too many unpredictable variables.

kittle42
05-31-2007, 02:30 PM
You're all right, the way the Sox are playing, they honestly have to be at least a 4% favorite. Anything less is an insult.

Fenway
05-31-2007, 02:38 PM
for the Cubs

31.98042% to WIN the division
2.84577& to get the Wild Card

34.82619% to get in

The Pirates who have a better record than the Cubs .93097%

Luke
05-31-2007, 03:17 PM
for the Cubs

31.98042% to WIN the division
2.84577& to get the Wild Card

34.82619% to get in

The Pirates who have a better record than the Cubs .93097%

We can rest assured that MLB is going to make everyone play 162 games despite BP's analysis that it's all over on May 30th.

southside rocks
05-31-2007, 03:29 PM
Out of curiosity -- what is BP's track record with their predictions? Are they often right? Or are they just another flavor of opinionation that amounts to nothing every October, like the "analysts" on sports radio?

Just wondering, if anyone knows.

Chisox353014
05-31-2007, 03:57 PM
for the Cubs

31.98042% to WIN the division
2.84577& to get the Wild Card

34.82619% to get in

The Pirates who have a better record than the Cubs .93097%

Those numbers seem a bit high to me, especially the 31.98042% to win the division. I would think it's more like 27.8464535273473638%.

WizardsofOzzie
05-31-2007, 04:46 PM
Seriously, where do you sign up for a job there? I'd love to get paid for making **** up as I go!

Johnny Mostil
05-31-2007, 05:13 PM
Out of curiosity -- what is BP's track record with their predictions? Are they often right? Or are they just another flavor of opinionation that amounts to nothing every October, like the "analysts" on sports radio?

Just wondering, if anyone knows.

Good question. This (http://web.archive.org/web/20060531215636/http://www.baseballprospectus.com/statistics/ps_oddspec.php) should show what they said a year ago today. (If it doesn't, I'll see if I can cut and paste.)

EDIT: Link appears to work.

Here are some interesting tidbits from the 5/31/06 analysis, comparing ultimate playoff teams with their BP chances that date

NYY, ALE champ: 57.7% chance of winning division, 76.5% of making playoffs
MIN, ALC champ: 0.2% of winning division, 0.5% chance of making playoffs
DET, AL wild: 14.3% of winning wild card, 70.3% chance of making playoffs
OAK, ALW champ: 15.2% of winning division, 15.5% chance of making playoffs

NYM, NLE champ: 77.1% chance of winning division, 85.1% chance of making playoffs
STL, NLC champ: 67.7% chance of winning division, 74.8% chance of making playoffs
SDP, NLW champ: 7.1% chance of winning division, 12.3% chance of making playoffs
LAD, NL wild: 11.6% chance of winning wild card, 63.9% chance of making playoffs

One year ago today, BP had the Sox with a 36.0% chance of winning the division, a 17.5% chance of winning the wild card, and a 53.5% chance of making the playoffs.

It also had five AL teams with a greater than 50% chance of making the playoffs, but only three NL teams with at least a 50% chance of doing so.

ws05champs
05-31-2007, 05:36 PM
Out of curiosity -- what is BP's track record with their predictions? Are they often right? Or are they just another flavor of opinionation that amounts to nothing every October, like the "analysts" on sports radio?

Just wondering, if anyone knows.
They probably don't keep track of that statistic (or at least make it public knowledge).

StillMissOzzie
05-31-2007, 06:03 PM
Six percent to win the division seems somewhat reasonable to me, but only a 1.8% chance to make the playoffs? How can our chance at winning the wild card be lower than our chance at making the playoffs?

PECOTA sucks. Stats are something that should be used to evaluate what happened, not to predict what will happen. There's just too many unpredictable variables.

Ummm, that's a 0.6% chance to win the division, NOT 6%, according to these geniuses.
Today's lesson in probability: Pr(making the playoffs) = Pr(win division) + Pr(win WC)

SMO
:gulp:

soxtalker
05-31-2007, 06:09 PM
I don't understand the outcry against BP's predictions. They are based on some assumptions and a series of algorithms. I think that a more interesting question is to ask why they were right when they were right and, even more interesting, why they were wrong when they were wrong. The wikipedia link gives a nice description, though it is complex. It looks like they base a lot on past (3 years plus minor league) performance of individual players. So, this could break down if individual players do much better than the model. (I bet that is less likely, as they probably improve the model over time.) Alternately, it could also change if the teams make changes -- substitute different players. Isn't that what Minnesota did last year?

downstairs
05-31-2007, 07:05 PM
I don't understand the outcry against BP's predictions. They are based on some assumptions and a series of algorithms. I think that a more interesting question is to ask why they were right when they were right and, even more interesting, why they were wrong when they were wrong. The wikipedia link gives a nice description, though it is complex. It looks like they base a lot on past (3 years plus minor league) performance of individual players. So, this could break down if individual players do much better than the model. (I bet that is less likely, as they probably improve the model over time.) Alternately, it could also change if the teams make changes -- substitute different players. Isn't that what Minnesota did last year?

Some people just don't like thinking about math and algorythyms, and some do. I get a bit disturbed when people completely dismiss all theoretical thinking/statistical predictions. Hey, if its not your cup of tea, fine- there's plenty more to enjoy about baseball.

I happen to love it. And I think the current models, while flawed, are facinating in what they can predict. And they often do.

By the way, the PECOTA model gives the sox a 1.8% chance of making the playoffs. In all honesty, the "traditionalist" model of looking at how they're playing and some basic stats would lead anyone to think about 0.0%.

:cool:

I'd love for once someone to fight the fire with fire... show me why the pythagorean formula from James is inaccurate in numbers. I can understand its flaws, but I think its a heck of a lot better than pure wins and losses.

southside rocks
05-31-2007, 09:01 PM
Good question. This (http://web.archive.org/web/20060531215636/http://www.baseballprospectus.com/statistics/ps_oddspec.php) should show what they said a year ago today. (If it doesn't, I'll see if I can cut and paste.)

EDIT: Link appears to work.

Here are some interesting tidbits from the 5/31/06 analysis, comparing ultimate playoff teams with their BP chances that date

NYY, ALE champ: 57.7% chance of winning division, 76.5% of making playoffs
MIN, ALC champ: 0.2% of winning division, 0.5% chance of making playoffs
DET, AL wild: 14.3% of winning wild card, 70.3% chance of making playoffs
OAK, ALW champ: 15.2% of winning division, 15.5% chance of making playoffs

NYM, NLE champ: 77.1% chance of winning division, 85.1% chance of making playoffs
STL, NLC champ: 67.7% chance of winning division, 74.8% chance of making playoffs
SDP, NLW champ: 7.1% chance of winning division, 12.3% chance of making playoffs
LAD, NL wild: 11.6% chance of winning wild card, 63.9% chance of making playoffs

One year ago today, BP had the Sox with a 36.0% chance of winning the division, a 17.5% chance of winning the wild card, and a 53.5% chance of making the playoffs.

It also had five AL teams with a greater than 50% chance of making the playoffs, but only three NL teams with at least a 50% chance of doing so.

That is interesting -- thanks. They seem to have a bit better track record than the jabbering radio goofs, but then, that's not saying much...

chisoxfanatic
05-31-2007, 09:22 PM
Seriously, where do you sign up for a job there? I'd love to get paid for making **** up as I go!
They're less accurate than a Chicago weatherman (other than Skilling)!

BP has a higher odds of being wrong than right.

WLL1855
05-31-2007, 10:49 PM
I don't understand the outcry against BP's predictions.

In a nutshell, I've got just as good a chance of predicting the playoff teams as BP and all I've got to use is the daily paper and my gut.

The outcry is that telling me the chances something might happen (using whatever model someone has cooked up) isn't the same thing as telling me what is going to happen. The clowns at BP act surprised/shocked/defiant when the outcome of the actual game being played doesn't happen according to their model (see 2006 World Series).

Players get hurt, change teams, go on hot streaks, and change or lose their roles all the time. Sabremetrics simply can't account for all of the chaos theory present in the game of baseball to tell you what is going to happen. Its fine as a "what happened" analysis tool but that's about it.

The Dude
05-31-2007, 10:55 PM
for the Cubs

31.98042% to WIN the division
2.84577& to get the Wild Card

34.82619% to get in

The Pirates who have a better record than the Cubs .93097%

That's because their division sucks and ours is the best in baseball. I can't see our Sox making the playoffs but you'd think we'd get a bigger % than that.