FanPost

Much Ado About Projections

We've ad a little to-do lately about projection systems and their accuracy.  Well, actually the to-do involved quite a bit more than that, but that part bored me.  I did realize that there doesn't seem to be any repository of projection data; at least not within a couple top google hits.  So, I put together a collection of the 2007 projections I have at hand:  PECOTA (from Baseball Prospectus 2007 & 2008), Bill James and The Hardball Times Annual Previews 2007 & 2008.  

Some of these were reasonably good, and some were laughably wrong (see Braun, Ryan in THT 2007 Preview).  There's a lot of good work that goes into these, but there's clearly a lot of room for improvement.  The Brewers, in particular, caused some trouble in 2007 due to the youth on the team, and thus the absence of performance data to draw on.

Without further ado, the numbers.

Rickie Weeks
James 2007     .271/.354/.437
THT 2007       .249/.338/.398
PECOTA 2007    .275/.358/.456
Actual 2007    .235/.374/.433

James 2008     .265/.374/.453
THT 2008       .261/.367/.447
PECOTA 2008    .265/.368/.454

I'm probably not going to comment on every one of these, but Rickie's 2007 was simply not projectable (is that a word?).  In The Bill James Gold Mine, he points out that Rickie managed to increase his walks from 30 in 2006 to 78 in 2007, and barely increased his OBP.  Bill James and PECOTA had Rickie pegged as a decent-average work-the-count kind of player, and except for the atrocious hitting early in the season, that's pretty much what he was.

JJ Hardy
James 2007     .252/.325/.403
THT 2007       .260/.328/.417
PECOTA 2007    .263/.329/.419
Actual 2007    .277/.323/.463

James 2008     .274/.330/.462
THT 2008       .273/.324/.446
PECOTA 2008    .273/.329/.441

Ryan Braun
James 2007     N/A
THT 2007       .235/.294/.414
PECOTA 2007    .283/.341/.504
Actual 2007    .324/.370/.634

James 2008     .326/.383/.660
THT 2008       .289/.338/.549
PECOTA 2008    .300/.367/.575

If anyone had Braun projected to have a season like that, I didn't see their projection.  Still, THT looks pretty silly on this one.

Prince Fielder
James 2007     .280/.360/.517
THT 2007       .269/.350/.486
PECOTA 2007    .292/.375/.529
Actual 2007    .288/.395/.618

James 2008     .289/.388/.591
THT 2008       .301/.397/.587
PECOTA 2008    .287/.384/.560

Corey Hart
James 2007     .283/.338/.483
THT 2007       .252/.312/.434
PECOTA 2007    .288/.352/.518
Actual 2007    .295/.353/.539

James 2008     .297/.357/.534
THT 2008       .285/.343/.511
PECOTA 2008    .288/.358/.528

Bill Hall
James 2007     .268/.330/.485
THT 2007       .265/.326/.483
PECOTA 2007    .277/.345/.520
Actual 2007    .254/.315/.425

James 2008     .263/.328/.471
THT 2008       .265/.330/.467
PECOTA 2008    .267/.335/.482

I thought the projections were unreasonably harsh on Hall last year, and it turns out they were too optimistic.

Geoff Jenkins
James 2007     .273/.351/.476
THT 2007       .264/.348/.442
PECOTA 2007    .268/.343/.449
Actual 2007    .255/.319/.471

Johnny Estrada
James 2007     .291/.334/.427
THT 2007       .280/.327/.416
PECOTA 2007    .280/.326/.423
Actual 2007    .278/.296/.403

I have only ever had 2 reasons for looking at preseason projections:

  1. In the offseason, we have to talk about something.  In that case, I like to gather as many projections as I can and discuss where they're off and where they're probably correct.  It's harmless fun and a launching point for discussion.
  2. Fantasy baseball.  This is the one where I would actually like for the system to be more correct than not.  For me, I like PECOTA's accuracy.  It doesn't get the huge up or down swings (and it doesn't seem likely that any system can do that), but it does a pretty good job of pegging the talent level of the various players.  Across 18-25 players on a fantasy team, it does a pretty good job.
I don't have any larger points.  Like I said, i didn't find a collection of these projections in some quick googling, so I threw them together.