I wonder if Weiter's projections are skewed because of the existence of only one year of data. (Last year being his only year of proffesional ball) I noticed that Pecota has this tendency to regress players to the mean most of the time. For example, if a player hits say .350/.450/.650 between short season ball and A ball, then the next year does the same between High A ball and AA, he'll be expected to regress the next year, since the translations will show a marked improvement the second year over the first. Since Weiters has no experience before last year, there is nothing to regress to, hence the explosive projection.

That's not dismissing his obvious talent, but it concerns me when projecting from a single point. Its simply not enough data. If I were the GM of the Orioles, I wouldn't look at Pecota and say, "Ok Matt, we're going to bring you up to the majors, hand you the starting job, and bat you third based on this projection." Instead, I'd probably send him to AAA while keeping a close eye out to see if he continues to rake, and bring him up if he does. If he struggles, I might let him play the whole year in AAA and let him win a job next year on the club.

PECOTA gave me a stealth pick with its Weiter's prognosis. Finding two decent catchers is really tough in my fantasy league, and with a bias against "losing" teams like the Orioles, he'll be a great surprise to people --"who...the Orioles"? Even if he does 75-80% of what PECOTA projects, he'll be a worthy later-round pick to bolster up BA or RBI's. Thanks PECOTA!

How much have the minor league translations been tweaked between last year and this? Maybe in an effort to not be so conservative for rookies, changes have been adopted that may actually over-estimate rookie stats? Other than Pujols, what other rookie can even come close to a stat line Wieters projection? He might be the best catcher in the last 50 years, but even then I wouldn't bet on that kind of a production for at least a couple of years.

I really appreciate this analysis, mostly because I'm both an Orioles fan and someone who just traded my kitchen sink for the right to draft Wieters in my 100%-keeper fantasy league. But I also appreciate that you spelled out how PECOTA works for a guy like Wieters. This is something that Nate Silver used to do comprehensively for all top prospects on an annual basis, but unfortunately, hasn't done much of lately.

As hopeful as I am about Wieters, though, I am a bit skeptical that PECOTA's projections can come true, especially this year. I'd like to see how someone responds to irablum's question regarding Wieters' possession of only one year of pro-statistics. Although they're harder to translate, Wieters' college statistics might be interesting to see as well, to examine whether the guy ever had a dip in his performance.

But anyway, thanks for the analysis.

There is only one catcher in Wieters top 20 comparables and his attrition rate is a measly 3%. PECOTA is not factoring in the potential for injury-whether it be missed time or decreased production-nearly enough. Russ Martin is the #1 MLB catcher in terms of attrition at 11%. There is no way Wieters is more durable than that. His similarity index is 0.

Here is some supporting evidence that PECOTA is not regressing enough from "What the Stats Mean" on the minor league translation page;

"Regular Translation: This applies adjustments for league difficulty to the player statistics,
resulting in a changed line. Everybody, from every park, is translated to the same underlying
standard. Hitters go to a league that hits .270/.330/.420 with a .260 eqa, pitchers go to a league
that allows 9 hits, 1 home run, 3 walks, and 6 strikeouts per nine innings with a 4.50 ERA. For
majors, these should be a close - but not necessarily exact - match for the "adjusted for all-time"
numbers (the translation process can cause slight differences from the exact rating, typically no
more than two points of EqA, unless PA are really low.) Ideally, this is how the player would
perform if he were called up to the majors right now - allowing for the difference between the
real majors and the standard league. In practice, they fall short. Most of that is because the
player most likely to be promoted is the one who is playing farthest over his head, a form of
selection bias. Players in AA or lower are even more likely to fall short, because the translation
routine is calibrated for one-level jumps, and there does appear to be an additional drop in
performance for players who try to jump two (or more) levels at once."



While he's the best prospect in baseball, Wieters' PECOTA projection is nuts. PECOTA is basically saying Matt Wieters is Mike Piazza, not that when everything goes right he can become Mike Piazza.

Alex Gordon's PECOTA in 2007-.282/.364/.511 good for a .301 EQA.

Forget those counting numbers for him... they're based on 649 PA. No way a catcher who breaks spring training in the minors gets to that amount of playing time.

The thing is, at some point, someone will emerge as good as, if not better than, Albert Pujols, Alex Rodriguez, or, for that matter, Mike Piazza. There's no reason to believe that PECOTA won't predict this emergence. Maybe Wieters is that man. I can only hope...

One thing that makes the projection difficult to swallow is the idea that a catcher, any catcher, would see 649 at bats in a season. Russell Martin had 553 AB last year to lead all catchers, and he started 8 games at 3B.

Nothing against Wieters, but even if he makes that triple slash line, he won't see those kind of counting numbers, at least not this year.

To complete the "PECOTA was awesome with rookies last year" argument, you should look up its projections for Jay Bruce, Joey Votto, Lastings Milledge, Michael Bourn, Adam Jones, Justin Upton, Jacoby Ellsbury, etc. 2 ROYs is a small sample size.

For what its worth, ZiPs, CHone and Bill James had better projections for Soto and most of the ones I listed above. In my very limited experience, PECOTA is overly rosy compared to those projection systems on young players like Josh Phelps, Wilson Betemit, Chris Young, Milledge etc.

Lets start a comparative study with Weiters

PECOTA .311/.395/.544 (572 ABs)
Bill James 311/407/526 470 (470)
CHONE .274/352/439 (419)
Oliver 294/373/487 (453)
Baseball Forecaster 341/434/576 (200)

(Had to use Abs instead of PAs).

Not only is 2 ROYs a small sample size, but it's a clear case of selection bias.

The best rookie in each league (determined after the fact) is almost certainly going to be a player who exceeded his forecast.

If we similarly judged PECOTA only by looking at MVPs we'd probably see a list of players who performed in their 90th percentile projection or even better.

The thing that makes Weiters a "fantasy darling" is that he is a catcher. He could underwhelm and go .270/15 HR/70 RBI and still rank as a top 10 catcher. He could have just one hot month and still outproduce most catchers even if the league ends up catching up to him. If Weiters played any other position (besides maybe, second base), he'd be a much riskier pick.

I'd draft him wanting a minimum of .270/15/70 and hoping for for .280/20/80 while keeping J.R. Towles in mind. Problem is, the hype might be built up to the point that he is overdrafted when he should instead be the 5th or so catcher to come off the board. I'd probably grab him in the 6th round or so if he lasted that long, and if not, take a Victor Martinez-type as a consolation.

Just an interesting tidbit here. The Bill James Handbook has also projected Weiters to have quite a season (and I've been fairly impressed with the Handbook's projections as I've purchased the last 4 years versions).

the Handbook projects
.311 AVG
.407 OBP
.526 SLG

They project fewer plate appearances but still with 25 doubles, 25 home runs, and 100 runs created.

Pretty similar stuff. Should I draft this kid ahead of the likes of Mauer? Soto? McCann? Martin? I'm torn...

Below is a link to a review by Nate Silver of PECOTA's first projection for Tim Lincecum, who also had an historically epic first pro season. Given the spectacular forecasts of future performance by PECOTA of Lincecum in 2007 and now Wieters in 2009 and how good that Lincecum forecast proved to be, I thought you might find it worth the re-read.

http://www.baseballprospectus.com/article.php?articleid=6125

I think the key is the fact that Wieters hasn't played above AA (and only about half a season at AA). I think it's insane to think Weiters will hit his PECOTA weighted mean. Look at Jay Bruce and Alex Gordon, how did that go?

The competition is 15% tougher as you move up each league in the minors, and there is a big 25% in the competition difference from AAA to MLB. Assuming a young player improves 10% each year (by aging and facing tougher competition or learning the competition), we could possibly (or impossibly) expect Weiters' .365/.460/.625 to increase to .401/.506/.687 this year. But then factoring in his leap to MLB and the basically 40% tougher competition, he would clock in at .256/.322/.438. Maybe that's low, but I'd buy that more than .311/.395/.544.

If he kept improving at a 10% clip in MLB for two years, he would reach .310/.390/.530. That kind of looks like his PECOTA, now, doesn't it?

So, it looks like PECOTA is giving Wieters full credit for playing at his full potential at MLB this year, when it should be factoring in the tougher competition.


[Aside--It also looks like PECOTA only looks at his AA numbers. If you take his averages for last year, his projection should come out even lower. Let me put it this way...would you project a High-A player who hits .345/.448/.576 to hit .311/.395/.544 in MLB a year later?]

Wieters has an unusually high BABIP of .383 compared to his LD% of 16-18% in 2007:

http://firstinning.com/players/Matt-Wieters-a/

Is that not a reason to temper PECOTA's excitement or does it consider such factors already?

Based on this projection, I'd be willing to offer my entire Strat-O-Matic draft just to move up and take Wieters.

Now I just have to apply PECOTA to figure out who is going to finish last in my league and get the #1 draft pick!