from Amazon
Friday, December 18, 2009
Corey Koskie v Justin Morneau
Corey Koskie, from 2000 to the end of his career, had 3495 PA, with 22 WAR, according to BaseballProjection.com. Justin Morneau, from 2003-09, had 3557 PA and only 16 WAR. How the heck did Corey Koskie get 6 more wins than Justin Morneau?
With roughly 700 PA per 162 games, that gives each of them almost exactly 5 full seasons. That puts Morneau at 3 WAR per season (an unsurprising figure), and Koskie at 4.5 WAR, which is hugely surprising. The way to think of the win scale:
0 - minor leaguer
1 - bench player
2 - average player
3 - very good player
4 - great player
5 - superstar
6+ ridiculously good
Morneau, like most good-hitting firstbasemen are considered “4” when they are actually “3”. So, no surprise to see him with 16 WAR in 5 seasons. But Koskie? Let’s break him down. BProj has him at +48 runs above average on offense over 5 full seasons, which makes him a solid hitter. Fangraphs has him even higher, at +79 runs (see wRAA), with a career wRC+ of 115 (contributes 15% more runs than the average hitter). This number might be more appropriate as it is park adjusted, which translates closer to +68 runs. Baseball-Reference.com has him at +72 runs. It should be noted that BProj has him that low because it includes some peripheral stats that the other sites don’t have. But, roughly speaking, he’s somewhere between +50 and +70 runs as a hitter.
How about as a fielder? BProj has him as being +42 runs above the average fielding 3B, which again is a solid figure. His UZR at Fangraphs is +49 runs starting in 2002. He was another +22 runs in 2000-01 the last time MGL published those stats, for a total of +71 runs. That’s an excellent fielder. Once again, his fielding is somewhere between +50 and +70 runs. Furthermore, the average fielding 3B is a bit better fielder than what you’d find at a “neutral” position (i.e., one-eighth of each of the 8 positions on the field). Not much though, but roughly 2.5 runs per season, or an extra +12.5 runs we give for 5 full seasons.
If we add it up, a very good hitter, a very good fielder, at a slight premium position, and you have someone who is +112 to +152 runs better than average over 5 seasons, which is +11 to +15 wins above average. The average player contributes +2.25 wins above the “replacement” level player each season (and these days, more like +2 wins for the average NLer, and +2.5 wins for the average ALer). So, over 5 seasons, the average ALer is going to come out about +11 to +12.5 wins ahead of the replacement level player. Add that to his +11 to +15 performance, and you have +22 to +27 wins above replacement.
And that, ladies and gentlemen, is how BProj has Corey Koskie at 22 WAR, which makes him roughly one of the 100 most productive nonpitchers of the 2000-09 decade, even though he only had the equivalent of 5 full seasons.
Paraphrasing Bill James, a stat that always surprises is usually wrong, a stat that never surprises is probably useless, and a stat that confirms 80% of what you know, and surprises you the other 20% of the time is the one that we should pay attention to. Morneau is part of the 80% and Koskie is part of the 20%.