The book also just provides the median forecast. PECOTA projects a full distribution of the player's performance, in essense showing what the top 10% of players most closely comparableto this player did, what the top 20% did, what the top 30% did, etc.

The median forecast tends to be fine for rate stats like BA, but it tends to underestimate counting stats. To get the full forecast, you need the online source.

GotowarMissAgnes wrote:The book also just provides the median forecast. PECOTA projects a full distribution of the player's performance, in essense showing what the top 10% of players most closely comparableto this player did, what the top 20% did, what the top 30% did, etc.

The median forecast tends to be fine for rate stats like BA, but it tends to underestimate counting stats. To get the full forecast, you need the online source.

ah, that does sound cool!! Thanks for the explanation...*goes to BP website...*

I was going to say, it seems like this system would push projections towards the mean. I mean, there just aren't a lot of "really good" hitters out there, so when you're trying to compare an elite one to others and if "expands" its tolerance, you'd just be including more "worse" players in to the equation, which would tend to draw their projections towards the median. I mean, they certainly minimize net error through this proceedure, but with guys like Albert Pujols and Alex Rodriguez, you don't really expect any sort of regression there. It seems like it would be worse with younger players since they'd have fewer "top-teir" seasons with which to differentiate themselves from the more middling players.

But then again, they've got Howard down for 41HRs, so who knows.

"I don't buy everything I read,
I havn't even read everything I've bought"

"I find it more comforting to believe that this [life] isn't simply a test."

George_Foreman

General Manager

Posts: 4351

Joined: 16 Apr 2004

Home Cafe: Baseball

Location: at Morimoto's, eating $50 worth of sushi

George_Foreman wrote:I was going to say, it seems like this system would push projections towards the mean. I mean, there just aren't a lot of "really good" hitters out there, so when you're trying to compare an elite one to others and if "expands" its tolerance, you'd just be including more "worse" players in to the equation, which would tend to draw their projections towards the median. I mean, they certainly minimize net error through this proceedure, but with guys like Albert Pujols and Alex Rodriguez, you don't really expect any sort of regression there. It seems like it would be worse with younger players since they'd have fewer "top-teir" seasons with which to differentiate themselves from the more middling players.

But then again, they've got Howard down for 41HRs, so who knows.

An alternative way of understanding this is that PECOTA simply gives you more information. The other forecasts give you one number. In one way or another, this is some measure of the central tendency of this player's likely performance. PECOTA gives you that....but it also gives you a whole range of additional information. If you observe Howard or Wright or Hermida, or have other information, you can say, 'I think this guy is more like an 80% percentile guy" and use that forecast, rather than the central tendency.

1) Do the books provide any different info than the website, or is it just a matter of convenience?

2) Do subscribers actually have access to the formula they use, or is it a big family secret like the KFC recipe?

1) Not sure, haven't book the book. I think it gives you more in depth analysis of players individually and maybe even prospects.

2) It's not as simple as a formula where you input the players past statistics. It draws upon databases of thousands of players and finds comparable players, and what career paths they followed. I think it also adjusts for alot of other factors to: playing time, lineup, home ballpark etc.

I understand that it's not a "simple" formula, but it still is based on a formula of some sort. I'm just curious as to exactly what factors they use and how much weight they are given.

That's why they call me whiskers -- 'cuz I'm curious like a cat (sorry, obscure Will Farrell reference that I probably misquoted).

WittyC wrote:I understand that it's not a "simple" formula, but it still is based on a formula of some sort. I'm just curious as to exactly what factors they use and how much weight they are given.

That's why they call me whiskers -- 'cuz I'm curious like a cat (sorry, obscure Will Farrell reference that I probably misquoted).

Sorry Witty, wasn;t sure what level of understanding you were at. But no, they don't really go into any detail (from anything that I have seen) about exactly how PECOTA works.

WittyC wrote:I understand that it's not a "simple" formula, but it still is based on a formula of some sort. I'm just curious as to exactly what factors they use and how much weight they are given.

That's why they call me whiskers -- 'cuz I'm curious like a cat (sorry, obscure Will Farrell reference that I probably misquoted).

And, for the same reason Coke won't share its formula, BP isn't interested in sharing their formula.

WittyC wrote:I understand that it's not a "simple" formula, but it still is based on a formula of some sort. I'm just curious as to exactly what factors they use and how much weight they are given.

That's why they call me whiskers -- 'cuz I'm curious like a cat (sorry, obscure Will Farrell reference that I probably misquoted).

And, for the same reason Coke won't share its formula, BP isn't interested in sharing their formula.

Exactly what I figured (well, I alluded to KFC, but yeah).