Saturday, March 21, 2009

Local Maximum Or Global Maximum?

When I discussed the contract that the Orioles gave to Nick Markakis, I projected his value to the team by assuming that he'll continue producing at the level that I have him projected at for 2009 (about 5.5 WAR). Now that may not be very fair to assume, so perhaps comparing Nick to other similar players and seeing how they preformed after the age of 24 may be more appropriate. This is essentially what BaseballProspectus' PECOTA model does, but this is a far (far) less detailed look at it using Chone's historical WAR. I pulled in the data for players comparable to Markakis, including those listed at Baseball Reference, ZiPS, PECOTA, as well as various scouting reports and just players who showed similar offensive profiles at the age of 24, concentrating on left-handed outfielders. Examples include Gary Matthews, Ben Grieve, Steve Kemp, Rusty Staub, and Ellis Burks.

The results, in handy-dandy graphical form (sorry, it's a little blurry; you can click for a clearer version, or just look at the general trends with Nick being in orange and the comparables average being red.):

With the bat, Nick started out just slightly above average but has had drastic improvement each year. His comps likewise started out low but got better relatively quickly, though not to the same degree. Unfortunately, that age-24 season was their peak performance, with a steady decline starting right after that.

Because he was very good on defense his first year, Nick saw only a small bump in WAR in season two. It was in 2008 that he really broke out, and his comps also saw a modest improvement going from age-22 to age-23, and much larger one going to age-24. Again, there is a fairly steady decline starting at age-25.

Nick is actually starting at a higher baseline, so his decline starting at age-25 (if it indeed occurs) will leave him as a more valuable player for the length of his contract. Nick was 40% better with the bat than his comps in his age-24 season, so I at least kept that advantage for him in projecting out his batting runs.

Even though several of his comparable players where done at a young age (Grieve and Kemp most immediately spring to mind), overall they were still able to provide above average offense well into their late 30's. The big age-related decline happens started at age-33, but they are still valuable hitters for a few years after that. With things like these, there is often a selection bias where only the players that were good enough to keep playing are measured. I tried to get around that by including zeroes for players after they were done through age 40 for everyone that isn't still active.

The WARs likewise show the comps as still being somewhat valuable (0.5-1.0 WAR) into their later 30's. There's a steady decline from 25 to around 31 (with a strange down-year at age-30) before a sharper one starts. This has Nick (assuming again that he's better than the composite) declining from 4 WAR to around 3.4 WAR in the first five years of his contract, with a dip to 2.5 WAR at age-30 before having a rebound to 3 WAR at age-31 (when he may be on another team). I would honestly be somewhat disappointed if Nick was just a solid occasional all-star level player for the next six years instead of the star perennial all-star level player he showed he could be last year. Not mad or upset - just disappointed. I am anxiously awaiting another data point to see whether we can adjust the graph to go more into Carl Yastrzemski territory instead of Jeff Burroughs territory.

Even given the downward sloping lines, production at the projected levels would be worth about $95 M to the Orioles (taking into account salary inflation and arbitration year discounts). I'm still (mostly) expecting Nick to be closer to a 5.5 WAR player than a 3.5 WAR player, but even with a reasonable decline he's still easily worth his $66 M contract. Oh yeah... it was a steal.

And sorry for the title.

No comments: