PECOTA Loves Wieters

by R.J. Anderson on Feb 3, 2009 7:00 AM EST

At least relative to the rest of the league. On the WARP leaderboard, after Albert Pujols and Hanley Ramirez comes Matt Wieters, projected at 7.9 WARP. If I were an oddsmaker, I would put that one at "highly unlikely". Here's why:

I'm a fan of Wieters. He's awesome. Perhaps the best prospect in baseball in years and he's almost certainly bound for stardom and glory, but being worth ~8 wins above replacement is extremely difficult. This year only Albert Pujols, Chase Utley, Chipper Jones, and Hanley Ramirez came within a half win of 8.

Even if you assume Wieters will get 20 replacement value runs, 22 positional, and 10 defensive, you still need him to be worth 27 offensive runs to equal 79 runs above replacement. Wieters hit worth ~28 offensive runs in Double-A and ~26 in High-A, so he's done it before, but assuming he spends minimal time in Triple-A and then leaps right into fulltime major league duty in the American League East, should we really expect him to be the best offensive catcher next year?

This is not a PECOTA-only problem, and frankly I still wonder if WARP's replacement level baseline still isn't a bit too low, but this is one of the more the most glaring cases of a projection system overestimating a player. And to think, this is simply the weighted mean.

Tweet Comment 19 comments | 0 recs |

Do you like this story?

Comments

weighted mean

R.J.

I agree with this article and I’m glad somebody wrote it.

I had a question for you, what did you mean by “And to think, this is simply the weighted mean.”

I have a vague idea of what you’re trying to say, but I think this point deserves some serious elaboration.

Thanks.

by cavegravedave on Feb 3, 2009 8:21 AM EST reply actions

Sure.

What I meant is that this isn’t Wieters 75th or 90th percentile projection. For example, last year Evan Longoria’s weighted mean had him at 6.8 WARP while his 75th said 7.8 and 90th said 9.6.

by R.J. Anderson on Feb 3, 2009 9:52 AM EST up reply actions

Has BP ever stated what their replacement level is set at? I always hear that it’s way too low but I’ve never heard an actual number for the level.

Bay City Ball | Beyond the Box Score

by xanthan on Feb 3, 2009 10:56 AM EST reply actions

They have.

From the glossary:

It should be noted that a team which is at replacement level in all three of batting, pitching, and fielding will be an extraordinarily bad team, on the order of 20-25 wins in a 162-game season.

That’s about a .125-.150 win percentage. Most sabermetricians use something between .290 and .350 for rep-level.

I has a blog. | Ask a sabermagician!

by cwyers on Feb 3, 2009 11:01 AM EST up reply actions

WARP's rep leve has been bumped up for this year, I believe. Haven't seen details, though.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Feb 3, 2009 11:12 AM EST up reply actions

Yeah, that’s really low. Thanks Clin.

Bay City Ball | Beyond the Box Score

by xanthan on Feb 3, 2009 11:19 AM EST up reply actions

It's actually worse than I'm letting on here.

Sum of total WARP:

2006 – 2411.4
2007 – 2423.6
2008 – 2567.9

So, for those years, the implied (rather than assumed) rep-level given by WARP is something like .003, .001 and -.028. (Yes, that’s a negative win percentage. No, I’m not just being clever – that’s the actual rep-level needed to get total WARP and team wins to reconcile for 2008.)

This is what happens when you don’t pay very close attention to your assumptions – the assumption of linearity in run scoring and runs to wins only holds up in a certain band of performance, which works fine for most MLB teams but holds up very poorly at the low end of the WARP metric. (This is why Win Shares does its zeroing out, if I’m not mistaken.)

Meanwhile, looking at Fangraphs:

2006 – 561.8+458.3=1020.1
2007 – 569.4+460.1=1029.5
2008 – 576.2+459.9=1036.1

That gives rep-levels of .290, .288 and .287.

I has a blog. | Ask a sabermagician!

by cwyers on Feb 3, 2009 1:24 PM EST up reply actions

Do we know that PECOTA is using the updated WARP baseline?

I broke out my ’07 PECOTA spreadsheet (last year I subscribed) and the leaders in projected WARP were:

Last First WARP
Pujols Albert 9.2
Mauer Joe 8.7
Jeter Derek 8.2
Cabrera Miguel 7.9
Wright David 7.8
Tejada Miguel 7.7

If Weiters has a projected WARP of 7.9, that doesn’t seem to me like they’ve updated the rep-level yet.

I has a blog. | Ask a sabermagician!

by cwyers on Feb 3, 2009 1:46 PM EST reply actions

I put a question about this in Sheehan's chat on BP right now.

Judging by my track record in putting questions about BP’s metrics into their chats, this one will get ignored too. (I’m not convinced that BP has blacklisted me a la JC Bradbury, but they do seem to ignore these sorts of things.)

I has a blog. | Ask a sabermagician!

by cwyers on Feb 3, 2009 1:55 PM EST up reply actions

i doubt much of anything has been updated yet.

as evidenced by the assumption being wieters getting 650 PAs or something. that would be true if computers played and managed. iirc, they don’t really roll out the PECOTA stuff until mid-February.

by larry on Feb 3, 2009 2:17 PM EST up reply actions

I don't believe those PA or IP totals are anything other than something Marcel-like.

So they should be ignored for now. Clay’s going to update roster guesses as the spring moves forward, evidently.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Feb 3, 2009 2:59 PM EST up reply actions

that's correct.

they’ve go the raw data (go AC360!) but it needs the human touch. you know, the heartbeat.

by larry on Feb 3, 2009 3:21 PM EST up reply actions

PECOTA is absolutely right in what it's predicting.

Or at least, it’s correct in the aggregate. Weiters is being predicted to get 650 PAs total between the majors and the minors. And I think that’s probably correct – barring injury, he’s likely to play full-time between AA, AAA and the majors.

Remember – PECOTA predicts a combination of a player’s major league performance and the major league equivelency of his minor league performance. It’s not PECOTA’s function to sort out what part of his production will be in the minors and in the majors – that’s the depth charts, which will be out… whenever, I guess.

I has a blog. | Ask a sabermagician!

by cwyers on Feb 3, 2009 7:39 PM EST up reply actions

650 PAs from a catcher? Who got 520 PAs last year between A+ and AA ball?

The minor league season is shorter, and there’s not a lot of reason to bet on any catcher to get 650 PAs with a full season in the majors. Not that it’s a big deal,but that seems high.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Feb 4, 2009 5:33 PM EST up reply actions