I am not a fantasy expert. In fact, I’m pretty lousy at fantasy (as my various league-mates can attest), and the more I learn about “real” baseball the worse I get at fantasy. I’m not here to give fantasy advice (ahem), or tell people how it should be played (I participate and enjoy different kinds of leagues). I’m not even here to defend fantasy baseball (Carson already did). This isn’t even really a “fantasy” column. I simply want to suggest that traditional 5×5 (W, K, ERA, WHIP, S; BA, R, RBI, HR, SB) fantasy categories aren’t as retrograde as one might think, and in fact, may be in one sense more “realistic” than more recent fantasy scoring methods.
How can a writer for FanGraphs, of all places, be serious about traditional fantasy categories being more “realistic” than, say, ERA? This site doesn’t even use that to value pitchers! Steals on the same scale as home runs? Runs and RBI? Batting Average? Pitcher Wins? Even [terrible GM du jour] doesn’t care about those! Well, get ready to squeeze those mind grapes.
What I’m calling the “realism” of traditional categories isn’t come based their being “analytically correct.” In fact, I think it comes from the exact opposite, from their relative arbitrariness (for lack of a better word). While here at FanGraphs we talk about teams “buying wins” on the market, the teams, of course, aren’t literally buying wins, but players who can help them win [insert Royals/Astros/Mets joke here]. This is best done not by looking at how many “runs” a hitter might score or drive in, or projecting how many “wins” a pitcher will get, of course. Smart front offices will look at projected linear weights runs above/below average, or runs saved above/below average using some sort of defense-independent stat like FIP, tRA, or some kind of component ERA.
While these are superior methods ways to judge how many runs a team will likely score and allow and thus how many games they will probably win, we also know that they aren’t a “perfect fit” to actual baseball games. We know, that the team runs scored column is (almost) never identical to their wRC, or (non-calibrated) BaseRuns. Sometimes terrible hitters like Jose Guillen and Tony Batista rack up lots of RBI. Sometimes Scott Feldman wins more games than Zack Greinke. That doesn’t mean that teams should be going after Jose Guillen,* or trading Zack Greinke for Scott Feldman. This is the reality of the “looseness of fit” between our analysis of the game and the way actual games turn out.
* Wait, I’m a Royals fan. I meant to say, “Jose Guillen is due for a classic age-34 ‘breakout,’ and teams would be foolish not to look into taking on half his salary off of the Royals’ hands.”
Yes, part of this is adjusting projections for a fantasy context, e.g., how many runs will this high-OBP guy get score now that he’s in a better lineup, or how many games will Roy Halladay win now that he’s in the best team in an easier division. But I’m primarily (and obliquely) addressing something different — a conceptual gap. In real baseball, there’s a gap between how many runs a team “should” score and allow according to linear weights (or whatever), and how many they do. In more advanced fantasy leagues using say, linear weights, that gap isn’t present. I’m sure baseball GMs (the smart ones with good teams, anyway) would love baseball to be that way, so they wouldn’t be subject to those random variations. But reality has a way of evading our conceptual grasp. In this way, the experience of traditional (rather than “sabermetric”) fantasy more realistically reflects the experience of baseball reality.