Projected to hit .400/.600/.800 by PECOTA

*Warning: If you’re not a follower of fangraphs, BPro, THT, etc, then this post is not for you.

Nothing annoys me more than people who can’t stop felating their proprietary projection system. This annoys me especially when that proprietary system has a terrible recent track record. To be fully clear, I’m not complaining about authors who use a projection as part of their piece. If I’m writing an article about Carlos Ruiz, it might be useful to say that despite public sentiment, a wide range of projection systems expect Ruiz to be slightly worse in ’10 than he was in ;09.  And then I’d go on to cite why I agree/disagree, harp on the importance of catcher defense, and speculate about Chooch’s favorite flavor of ice cream.  That’s fine. No, what I’m complaining about is when that projection IS the piece.

By now, a couple of you (or maybe just Ruckus) know that the specific system I’m referencing is BPro’s PECOTA. I understand why BPro’s staff fall all over themselves to use PECOTA, it’s a pay site and to keep the money flowing they need to convince the payees that they’re getting a super sweet service. Nevertheless, as someone who ostensibly pays to read the site (my former boss at 3N2Sports actually foots the bill), it’s incredibly annoying to open an article titled “Jason Heyward: How Optimistic Can We Be?” and get nothing but PECOTA projections and database queries.

The author of that particular post, Kevin Goldstein, is a very able prospect maven. He has access to all kinds of resources that none of us can touch.  He knows things about Jason Heyward, he has expert knowledge on the player.  So why the hell is he posting vanilla PECOTA projections? We already know that PECOTA isn’t very good at doing projections. Despite all the gizmos and gadgets thrown in, a simple 3 year weighted average with age regression (Marcel) outperformed PECOTA at least the last two years.

I’ve already posted on how to interpret a projection.  A projected stat line is simply the average expectation of the entire range of likely outcomes. Picture a Bell Curve. And as I noted above, there’s a time and place for that. A fantasy owner may want to look only at a projection line for simplicity’s sake. An author may want to cite a projection in the course of an article. The projection itself is not meaningful, it is the interpretation that gives it meaning. An article about projections is never worth reading if there isn’t an interpretation. Too bad the BPro staff has shown no sign of getting the message.

I think too many people assume that stats-literate baseball fans lose sight of the game. However, when people start writing articles about players based solely on the output of a projection system, it becomes a fair criticism.

_____

PS It’s finally warm in Minnesota and I’m sequestered in a windowless office until 4:00…the result is I’m a little more ornery (and verbose) than my normal self.

Update: I’ve started a discussion with Tom Tango on this subject, which may or may not result in a good conversation.