PDA

View Full Version : BP's 2004 PECOTA Projections


jeremyb1
01-23-2004, 01:57 PM
I realize that Baseball Prospectus' PECOTA player projections are too number oriented for some posters on this board. However, for those into statistical analysis or simply those with an open mind, the numbers can be very interesting.

For those unfamiliar with PECOTA, the system adjusts players career statistics for level, park effects, and other factors and then projects the player's upcoming season based on other players performance throughout history. Last season was PECOTA's first season and it proved to the the most effective projection system available. (http://premium.baseballprospectus.com/article.php?articleid=2515)

The 2004 projections for hitters were just released and some interesting observations about our club include:

- Jeremy Reed is projected to hit .279/.347/.436 against major league pitching next season compared to .262/.323/.429 for Rowand and .251/.318/.457 for Borchard.

- Willie Harris is projected to hit .253/.313/.380 next season.

- Crede and Olivo are projected to improve to a respectable .271/.328/.470 and .248/.307/.406 respectively although both also have a significant chance of collapsing.

- Konerko is projected to rebound to hit .271/.337/.465.

Frater Perdurabo
01-23-2004, 02:10 PM
Originally posted by jeremyb1
I realize that Baseball Prospectus' PECOTA player projections are too number oriented for some posters on this board. However, for those into statistical analysis or simply those with an open mind, the numbers can be very interesting.

For those unfamiliar with PECOTA, the system adjusts players career statistics for level, park effects, and other factors and then projects the player's upcoming season based on other players performance throughout history. Last season was PECOTA's first season and it proved to the the most effective projection system available. (http://premium.baseballprospectus.com/article.php?articleid=2515)

The 2004 projections for hitters were just released and some interesting observations about our club include:

- Jeremy Reed is projected to hit .279/.347/.436 against major league pitching next season compared to .262/.323/.429 for Rowand and .251/.318/.457 for Borchard.

- Willie Harris is projected to hit .253/.313/.380 next season.

- Crede and Olivo are projected to improve to a respectable .271/.328/.470 and .248/.307/.406 respectively although both also have a significant chance of collapsing.

- Konerko is projected to rebound to hit .271/.337/.465.

Ugh. Reading the linked article makes my head hurt (too much math, that's why I'm in the liberal arts.... :D: )

But I understand the point that the formula projects a wider allowance for potential variance.

If Reed produced those stats as a leadoff hitter for the Sox this year (his projections are much better than Harris), I think Lee, Frank and Maggs all would clean up with RBI, and the Sox could muscle their way to a division title.

I hope PECOTA is right....

doublem23
01-23-2004, 02:11 PM
On the subject of Baseball Prospectus, when does the 2004 book come out?

MarkEdward
01-23-2004, 02:13 PM
Originally posted by doublem23
On the subject of Baseball Prospectus, when does the 2004 book come out?

It'll ship out mid-February, but you can pre-order it right now and get like 30% off.

Frater Perdurabo
01-23-2004, 02:13 PM
In fact, all those projected numbers look pretty nice. What does the analysis say about other Sox players? Also, did this analysis in any way predict the possibility of Konerko's miserable 2003?

poorme
01-23-2004, 02:24 PM
I seriously doubt Borchard could hit .250.

jeremyb1
01-23-2004, 02:48 PM
As far as pitching goes, PECOTA projects:

- Buehrle is in trouble with an expected ERA of 4.42

- Loazia will regress but remain effective with a 3.93 ERA.

- Garland is projected to remain relatively constant with a projected ERA of 4.76 but features a decent chance of a breakout season.

- Schoeneweis is likely to be as much of a problem as most seem to fear considering the fact that his ERA is expected to drop to 4.55 assuming he continues to pitch most of his innings in relief.

- Felix Diaz is the leader for the 5th starters job with a projected ERA of 4.84, followed by Rauch (5.15 ERA), Pacheco (5.29 ERA), and Cotts (6.54 ERA). PECOTA suggests that the best candidate, Kris Honel (4.71 ERA) is not in the running. Since all these pitchers are young and have limited track records against major league pitching there is obviously variation here but part of me wonders if we wouldn't be just as good going with two from the group of young pitchers instead of Schoeneweis who is clearly more valuable in the pen.

Clearly PECOTA alone should not be relied upon and disasterous seasons that come out of nowhere such as Konerko's '03 season or Chris Singleton's breakout season in '00 cannot be projected but its interesting to see that a pitcher like Honel has a chance of making a solid contribution during the season.

quade36
01-23-2004, 02:58 PM
I don't believe in predictions in a season that hasn't happened yet, where one person's destiny is their own doing.

If we actually believe this then we could assume the future in anything can be mathematically proven.

maurice
01-23-2004, 03:06 PM
I don't believe in predictions in a season that has happened. That's cheating. :D:

Jeremy has conceded that future performance can't be mathematically proven. It's impossible to predict or account for all variables.

I find it interesting that the attempted mathematical projections are very close to the numers I would have given off the top of my head. For example, the projected starters' numbers are consistent with my argument that we need to add a #1 starter, not a #5.

doublem23
01-23-2004, 03:13 PM
Originally posted by quade36
I don't believe in predictions in a season that hasn't happened yet, where one person's destiny is their own doing.

If we actually believe this then we could assume the future in anything can be mathematically proven.

Man, talk about overanalyzing. No one is saying here that since PECOTA says Loaiza will have a 3.93 ERA, it means he will have a 3.93 chiseled in stone. They are projections. Obviously some things break the laws of statistics. The very fact that we need a science like statistics says that things are uncertain. These projections are just intelligent guesses based on lots of data and history.

And, we can pretty much mathematically prove most of life. Someone who starts smoking a pack a day when they're 16 isn't likely to live to 65 and many do not. Just because your uncle's friend's sister's husband's nephew's T-ball coach has been smoking since he was a teen and is 90 years old doesn't mean that stat is invalid.

kittle42
01-23-2004, 03:18 PM
Originally posted by quade36
If we actually believe this then we could assume the future in anything can be mathematically proven.

I can prove that it will be 4:18 p.m. an hour from now.

kittle42
01-23-2004, 03:19 PM
Also, more accurate projections could be had if one just asked former utility infielder Bill Pecota.

jeremyb1
01-23-2004, 04:29 PM
Originally posted by maurice
I don't believe in predictions in a season that has happened. That's cheating. :D:

Jeremy has conceded that future performance can't be mathematically proven. It's impossible to predict or account for all variables.

I find it interesting that the attempted mathematical projections are very close to the numers I would have given off the top of my head. For example, the projected starters' numbers are consistent with my argument that we need to add a #1 starter, not a #5.

I think Loaiza still most likley fits the bill as a number one. I also have more faith in Buehrle than PECOTA does based on some intangibles like his second half improvement after altering his mechanics and watching the hard luck he faced in the first half. We'll have to see.

Dadawg_77
01-23-2004, 04:36 PM
Originally posted by quade36
If we actually believe this then we could assume the future in anything can be mathematically proven.

Read up on Chaos math, if we ever truly master that, watch out.
but that is a subject for the parking lot.

Pecota has a built variance and over all it preformed rather well for it first time in use.

mantis1212
01-23-2004, 04:54 PM
I wonder what PECOTA said about Loaiza last winter?? Right on the money I'm sure...

hold2dibber
01-23-2004, 10:45 PM
Originally posted by jeremyb1
IThe 2004 projections for hitters were just released and some interesting observations about our club include:

- Jeremy Reed is projected to hit .279/.347/.436 against major league pitching next season compared to .262/.323/.429 for Rowand and .251/.318/.457 for Borchard.

- Willie Harris is projected to hit .253/.313/.380 next season.

- Crede and Olivo are projected to improve to a respectable .271/.328/.470 and .248/.307/.406 respectively although both also have a significant chance of collapsing.

- Konerko is projected to rebound to hit .271/.337/.465.

What do you think about a Reed/Rowand platoon in CF? Does anyone know if Reed has shown any significant discrepancy in terms of performance against lefties vs. righties? Over the last three seasons, Rowand has an .819 OPS vs. lefties and a .697 OPS vs. righties. Maybe a good way to work Reed into the majors without putting too much pressure on him? Or maybe he'd be better off playing every day in Charlotte or Birmingham. Thoughts?

bc2k
01-24-2004, 12:21 AM
Originally posted by jeremyb1
but part of me wonders if we wouldn't be just as good going with two from the group of young pitchers instead of Schoeneweis who is clearly more valuable in the pen.

I agree. I also think Josh Stewart and Dan Wright should be added to that list. A Spring Training competition would be an obvious solution. Too bad the Sox have already resigned Schoeneweis.

jeremyb1
01-24-2004, 02:11 AM
Originally posted by mantis1212
I wonder what PECOTA said about Loaiza last winter?? Right on the money I'm sure...

I think you're missing the point here. PECOTA features probability for breakout seasons, improvement, collapse, and attrition to guague the likelihood of seemingly improbable events such as Konerko's disasterous season or Loaiza's breakout season. The fact that PECOTA doesn't project highly unlikely events such as Loaiza's '03 season does not prove the system is completely flawed. For instance, the projection system could fail to project Loaiza's season and still be highly accurate about 95% of its projections, still making the projections incredibly valuable and far better than any other predictions humanly possible.