2009 PECOTA spreadsheet is available

by Llewdor on Feb 2, 2009 10:45 AM PST

More information is always good, right?

If you're a BPro member, Here you go.

If not, here's a fun (and possibly irrelevant) piece:

One of Phillipe Aumont's top comps: Chris Carpenter

One of Roy Halladay's top comps: Chris Carpenter

One of Brandon Webb's top comps: Roy Halladay

And here's a fun game. In which order does the 2009 PECOTA rank these players, by projected VORP:

Milton Bradley
Adrian Beltre
Dustin Pedroia
Cristian Guzman
Mike Cameron

Hint: If you listed Cristian Guzman anywhere other than first, you're wrong.

Tweet Comment 30 comments | 0 recs |

Do you like this story?

Comments

RE: "If you listed Cristian Guzman anywhere other than first, you're wrong."

Uhhh, what?

Would you mind giving us non-subscribers his projected VORP?

by marc w on Feb 2, 2009 11:17 AM PST reply actions

49

I like using semi-colons; they make me feel smart.

by Llewdor on Feb 2, 2009 11:56 AM PST up reply actions

11th best projected VORP in baseball.

I like using semi-colons; they make me feel smart.

by Llewdor on Feb 2, 2009 1:18 PM PST up reply actions

I 100% believe this.

by marc w on Feb 2, 2009 1:27 PM PST up reply actions

If you take all of his top 4 comparables, they dropped off from age 30 to 31, yet his improve rate is great and his collapse/attrition are almost non-existent. A similarity index of 31 is reasonably high, too.

by ubelmann on Feb 2, 2009 3:01 PM PST up reply actions

Good thing we don't use VORP anymore.

by Matthew on Feb 2, 2009 11:26 AM PST reply actions

Unless you're a mainstream sportswriter who wants to take the piss out of sabermetrics.

I will smash your face into a jelly.

by Phil Hatzenbuehler on Feb 2, 2009 11:42 AM PST up reply actions

So are you saying that defense matters?

by JI on Feb 2, 2009 11:47 AM PST up reply actions

Sort of my point.

I like using semi-colons; they make me feel smart.

by Llewdor on Feb 2, 2009 11:58 AM PST up reply actions

MLB Front Office Manager does!

I just picked that up a few days ago. Sure stats like that are outdated but seeing anything related to sabermetrics in a video game made me feel all warm and fuzzy inside.

by CKel on Feb 2, 2009 4:35 PM PST up reply actions

Jose Lopez +11 at second base

Clay’s new defensive metric isn’t any better than his old one.

by davidcameron on Feb 2, 2009 11:50 AM PST reply actions

What is the new one?

by JI on Feb 2, 2009 11:54 AM PST up reply actions

He hasn't told us yet.

Apparently it will get explained in this year’s book.

I like using semi-colons; they make me feel smart.

by Llewdor on Feb 2, 2009 11:56 AM PST up reply actions

At least they finally realized that replacement-level did not mean:

Hits like Vidro and fields like Raul

by Matthew on Feb 2, 2009 11:56 AM PST up reply actions

Eh.

Maybe it’s because I no longer pay money to BP, but with the improvement in systems like CHONE and ZiPS over the last few years catching up to PECOTA on overall accuracy and BP’s flat out mess in terms of the stats that they use, I find PECOTA close to irrelevant these days.

by Matthew on Feb 2, 2009 12:50 PM PST up reply actions

That could be a problem with PECOTA instead of the defensive metric

The value listed on the spreadsheet is the projection and not the historical performance. Since PECOTA is heavily driven by finding the most comparable players, this could just mean that Nate Silver isn’t weighting defensive performance very heavily when he makes his list of comparable players, and then a lot of the players on Lopez’s list were good defensive players.

I know that in the past, the defensive performances in PECOTA went inversely with the percentiles while the offensive performances were better for higher percentiles. This always struck me as wrong. It’s certainly true that a lot of players who are good at defense aren’t that great at offense and a lot of players who are good at hitting aren’t that great at defense, but if you take someone who is average at both to begin the season, I don’t think you would usually find an inverse relationship between their offensive and defensive performances.

Which is a long way of saying that I don’t really trust Nate’s methodology for forecasting defensive performance regardless of what metric he chooses.

by ubelmann on Feb 2, 2009 1:01 PM PST up reply actions

Moot point

It’s hard to be completely sure (since PECOTA’s the blackest of the boxes), but I don’t think Nate’s using defensive ability at all for similar players. Which makes sense because defensive ability doesn’t really factor into offensive ability (at least I can’t think of anyone who’s found that, myself included), so it doesn’t help us with projecting offense.

Defensive aging, on the other hand, seems quite a bit simpler. Players generally peak early and then go on a downward slope. The error bars in defensive ability tends to obscure more rigorous defensive projections – it’s hard to compare what’s coming to what happened before when we only have a general idea of what happened before!

--
Dan Szymborski
dan@baseballprimer.com

by D.Szymborski on Feb 2, 2009 6:50 PM PST up reply actions

I agree that offensive and defensive abilities seem pretty uncorrelated

Eyeballing the PECOTA cards in the past has given me the impression that the offensive and defensive projections were (negatively) correlated in PECOTA percentiles, but glancing at a few cards now, maybe that has changed in the last couple of years or I happened upon a few weird player cards.

My main objection to Nate’s defensive projections is that I have no idea how he’s making them. The only times he ever even briefly explains the projections, the main point that he makes is that hitters are compared on EQA and pitchers are compared on EqERA. I’ve never seen any evidence that he makes separate comparisons (or ignores player comparisons altogether) for the defensive projections.

It could be that Nate is just using a generic aging curve, but that’s not really PECOTA’s style. As you said, PECOTA is the blackest of the boxes, so we won’t really know unless Nate decides to drop by, but to Dave’s original point, I think it’s awfully silly to pick one of Nate Silver’s PECOTA-projected defensive values and decide that Clay Davenport’s new defensive statistic is significantly flawed, especially since we know that there is nowhere near 100% agreement even amongst good advanced defensive metrics.

by ubelmann on Feb 2, 2009 7:35 PM PST up reply actions