Mays Copeland has made a player rater. He has explained how he derives the values of the rater, and how he prices those values. He’s made some software that makes it easy to customize those ratings and prices for your league. He has also, most impressively, created an interface that allows the user to use a variety of projection systems (Marcel, CHONE, ZIPS, others I’m forgetting, and even a composite of them all) to create a list for their league’s format.Â
The problem? The smell test. I ran the numbers for my Tout Wars NL league and the answers stunk. JJ Putz was named the fourth most valuable closer in the NL, by virtue of his 19 saves (!). F-Rod ranked first, with a projected 35 saves. The cloud, it seems, doesn’t always compute. Will Albert go for $49, as the PG suggests? In a word, no. But that isn’t the mistake (it is only a symptom).
Bid prices are different than projected values. Ignoring this truism means imagining that Johan Santana might be worth $48, as the LPP site suggests, but we all know that is wrong, even if Johan is healthy, which he may not be.
The problem is that there is no automatic pricing system based on projections that is going to properly price players for auction. Why? Because projections are, inevitably, 25 Â percent (or more) wrong. Even my projections aren’t perfect. We value projections because they fix for us what a player is expected to do, but what matters are the prices everyone else is willing to pay. And how our expectations stack up against them. That market evaluation is where the heavy lifting of draft prep comes. It combines the information in the projection with our assessment of risk, and filters it through our knowledge about the league we play in.
A system that projects Santana for $48, when he will not go for more than $38 in your league, is failing to properly allocate $10, compounding the effect of the error.
I dig Mays Copeland’s efforts, and maybe he isn’t selling the Pricing Guide as a list of bid prices, but rather as a way to compare different sets of stats (projected and real), though that wasn’t what I took away. My problem is that when he disses other sites, like the rototimes player rater, he offends me. Actually, he makes me an enemy. It isn’t that the rototimes rater is perfect, it isn’t, but it’s a lot better than anything Mays has come up with. His strict adherence to category scarcity blithely ignores the way people actually play the game.Â
From the spirit of his site, I would think Mays would find ways to improve the rototimes rater. but instead he chooses to diss it and promote himself. That would be okay if he got it right. Mays hasn’t yet.
Update: I went back and checked his 2008 prices, figuring that would be a better test of the Price Guide. It is better. The Putz problem is a result of the projections, not anything the pricer is doing wrong, but he’s still giving Steals and Saves full value, so he prices Mariano Rivera at $49 last year. That is probably correct in a math sense, but doesn’t reflect the way the game is played. Also, his 2008 stats have Matt Holliday in the AL, which makes them useless for evaluating what actually happened last year. So, an interesting efffort showing some promise, but there are kinks to work out.