If you’re anything like me, this time of year is the true sports dead zone. Football is gone, spring training games won’t start until the end of the month, and the NBA and NHL bore me to tears. To get my fix, I went so far as to watch Puerto Rico play Venezuela on ESPN3 – In Spanish. I don’t speak Spanish. Then, I suppose maybe you aren’t anything like me and watching the Pistons gives you all the sports fix you need, but then what are you doing reading Tigers blogs? What we do get this time of year is a steady trickle of projections for the 2011 baseball season, to feed the fan that has nothing better to occupy their mind than this simple question: “are we going to be good enough this year?”.
I’ll start you off with some numbers, readily available projections on Tigers batters: Bill James projections by Baseball Info Systems via FanGraphs, Marcel the Monkey projections from tangotiger.net via FanGraphs, RotoChamp projections from rotochamp.com via FanGraphs and CAIRO projections from RLYW the Replacement Level Yankees Weblog. I’ll be using one stat that rolls everything into one, and wOBA is a good a stat as any. ZiPS doesn’t publish wOBA (at least not that I’ve been able to find) so that got left out. For reference, the MLB average wOBA last year was .321
So why do these projection systems give such different results? Excellent question. Marcel the Monkey is the (deliberately) simplest, but it still uses a weighted average of the past 3 years numbers, player age and expected regression to the mean. The rest do that and more, though Marcel might predict more mean regression than the rest. If I had to guess (and why not?) I’d say it looks like RotoChamp weights last year more heavily than the others. Bill James projections have the reputation of being overly optimistic, though they aren’t as crude as something like Marcel, especially for rookies and second-year players. From the little I know, CAIRO projections seem like ones whose methodology I would approve of – using minor-league stats, batted ball breakdowns and age-related decline for individual traits.