BP Sox projection.

[doublem23]

Originally Posted by Nellie_Fox

If BP doesn't think that using the PECOTA projected RS/RA to plug into Pythagorean W/L, then they should say so, and certainly not do it themselves. Since they do it, that seems to say they think it's a legitimate thing to do.

It's because every-****ing-body knows that how they do it. It's why RS/RA are the 3rd and 4th column. How the **** else do you guys think you win and lose games? It's based on runs.

Originally Posted by asindc

What fact does it underline for those who use PECOTA's RS/RA projections to publish actual W/L projections?

That they're publishing the PECOTA projections?

Originally Posted by sullythered

Then BP should not project wins, at all. They should only publish projections on RS/RA, so as to avoid confusion.

First, if you actually go to their site, you'll see they do list RS/RA in the projections and second, what point would be served by not just doing the computation? Anyone with a 10-cent calculator can turn RS/RA into Pythagorean W-L. It's a very simple formula.

[asindc]

Originally Posted by doublem23

It's because every-****ing-body knows that how they do it. It's why RS/RA are the 3rd and 4th column. How the **** else do you guys think you win and lose games? It's based on runs.



That they're publishing the PECOTA projections?



First, if you actually go to their site, you'll see they do list RS/RA in the projections and second, what point would be served by not just doing the computation? Anyone with a 10-cent calculator can turn RS/RA into Pythagorean W-L. It's a very simple formula.

Obviously. They are also obviously publishing W/L projections. Noting that fact does not make someone anti-intellectual or anti-BP.

[kittle42]

Originally Posted by amsteel

If people knew how weather was predicted they would lose their ****.

Weather isn't played on paper, and there is no use to predicting it until the days are finished.

[sullythered]

Originally Posted by kittle42

Weather isn't played on paper, and there is no use to predicting it until the days are finished.

Is it just in my world, or doesn't everybody always make fun of how wrong the weatherman always is?

[amsteel]

Originally Posted by sullythered

Is it just in my world, or doesn't everybody always make fun of how wrong the weatherman always is?

Yet people still find weather predictions useful.

[sullythered]

Originally Posted by amsteel

Yet people still find weather predictions useful.

Yup, and so is PECOTA. But doing a full season projection on an entire team is like doing a weather forecast a year in advance.

[blandman]

Originally Posted by kittle42

Weather isn't played on paper, and there is no use to predicting it until the days are finished.

[blandman]

Originally Posted by sullythered

Yup, and so is PECOTA. But doing a full season projection on an entire team is like doing a weather forecast a year in advance.

http://www.youtube.com/watch?v=vBeGQAvabOE

[blandman]

Originally Posted by amsteel

Yet people still find weather predictions useful.

Probably because most of the time they're right, and very rarely are they really off. Sounds familiar...

[sullythered]

Originally Posted by blandman

Probably because most of the time they're right, and very rarely are they really off. Sounds familiar...

And large scale, long term predictions are nearly impossible. Sure does.

[amsteel]

Originally Posted by blandman

Probably because most of the time they're right, and very rarely are they really off. Sounds familiar...

Any good modeler know the strengths and weaknesses of their model, whether it's a flood model that under predicts rainfall for El Nino years, a hurricane model that has trouble tracking hurricanes in the North Atlantic, a financial investment model that consistently loses money in the medical sector, or a statistical baseball model that perennially under predicts wins for the White Sox. That's just how numerical models work, they are finite representations of an essentially infinite system.

In addition, any good forecaster knows not to trust any single model due to these inherent weaknesses, rather using an ensemble approach that helps identify potential outliers. I'm sure someone, somewhere has a model that nails the White Sox, but misses the Brewers every year. A single data point (or analysis of a single model's results) means very little.

Hell, the inconsistency in the BP model provides forecasting guidance: BP predicts the 2013 Sox to have 77 wins, but everyone knows they always under predict the White Sox by 7 wins a year so we can qualitatively conclude the Sox will likely have somewhere around 84 wins.

I think people overvalue BP since they are the most established name in the field. While at the same time undervaluing other models out there since they aren't BP. At the same time, I don't think BP gives a **** what their model says since we've spent 400 posts here arguing about it (as I assume other teams' fans have been doing) and I would assume they love the attention, regardless of whether its positive or negative.

[dickallen15]

This is what Nate Silver had to say about the time PECOTA was correct about the White Sox in his book The Signal and the Noise:


In 2007, for instance, PECOTA predicted that the Chicago White Sox -- just two years removed from winning a World Series title -- would finish instead with just seventy-two wins. The forecast was met with howls of protest from the Chicago media and from the White Sox front office. But it turned out to be exactly right: the White Sox went 72-90.

Now, Nate, I believe, invented the PECOTA, and says it is predicting win totals. You would think an intellectual like himself, the guy who came up with the system, would realize he didn't predict records, just RS/RA. And if you are going to boast about being right, you should be able to take the criticism when wrong or at the very least, be a little more humble when being correct, considering the tools haven't been all that accurate overall.

[doublem23]

Originally Posted by dickallen15

This is what Nate Silver had to say about the time PECOTA was correct about the White Sox in his book The Signal and the Noise:


In 2007, for instance, PECOTA predicted that the Chicago White Sox -- just two years removed from winning a World Series title -- would finish instead with just seventy-two wins. The forecast was met with howls of protest from the Chicago media and from the White Sox front office. But it turned out to be exactly right: the White Sox went 72-90.

Now, Nate, I believe, invented the PECOTA, and says it is predicting win totals. You would think an intellectual like himself, the guy who came up with the system, would realize he didn't predict records, just RS/RA. And if you are going to boast about being right, you should be able to take the criticism when wrong or at the very least, be a little more humble when being correct, considering the tools haven't been all that accurate overall.

[SCCWS]

Pecota also predicted that this commentary on their prediction would exceed 27 pages in comments.

[kittle42]

Originally Posted by SCCWS

Pecota also predicted that this commentary on their prediction would exceed 27 pages in comments.

Based on the fact that few understand it, or really any statistical model.