Almost all television broadcasts do it. Most stadiums do it. FOX has been doing it all throughout the World Series. In the onscreen graphic that comes up when a batter walks to the plate, you'll always see three numbers, and often, those three numbers only: batting average, home runs, and runs batted in. The limitations of HRs and RBIs compared to modern stats like wOBA are widely known. But exactly how poorly do the traditional Triple Crown stats assess player performance, compared to the sophisticated tools we have now? And how bad are they compared to the also-popular "triple slash" set of stats, AVG/OBP/SLG, which is just as easy as AVG/HR/RBI to use?
A quick aside: The great thing about wOBA is that it's essentially as close to perfect as we can get. It describes total offensive performance in a single number that's both context-neutral and comprehensive. Unlike other advanced stats, it doesn't pick and choose certain attributes to measure, and all of its coefficients are calculated from actual baseball events instead of being conjured out of thin air. It makes for a great baseline we can work from.
So, armed with this knowledge, the Triple Crown stats were easy enough to test. I took every qualified batter season from 2009-2011 and ran correlation tests with batting average, home runs, and runs batted in, compared to wOBA.
Another quick aside: The measuring stick here is R2. In a nutshell, R2 gives an indication of how much of one statistic's variability can be explained by the other stat, or to give an example, an R2 of 0.62 between wOBA and FictionalMadeUpStat says that FictionalMadeUpStat is responsible for up to 62% of what wOBA says. Since we're treating wOBA as our perfect barometer of value, this would say that FictionalMadeUpStat accounts for 62% of player value.
So, the results?
AVG: 0.45
HR: 0.47
RBI: 0.46
It turns out all three of the Triple Crown stats are at around the same level of suck. Batting average only represents 45% of offensive value, and a tally of home runs or runs batted in barely fares any better. To put that in perspective, here's the R2 for the triple slash stats, AVG/OBP/SLG:
AVG: 0.45
OBP: 0.78
SLG: 0.84
And if you add OBP and SLG together to form OPS?
OPS: 0.97
The world doesn't exactly need yet another "use these stats, not these antiquated old ones!" argument, but this really isn't the most difficult thing to fix. To borrow a common refrain from Tom Tango's blog, if you want to use non-statistical methods, be my guest. But if you're going to use numbers? Don't use those, use these.
The World Series returns to St. Louis with Game Six tomorrow at 5:05 PM Pacific.
