I don't remember how long ago it was, but I used to really like WHIP. It seemed that every baseball article out there describing a pitcher was using this stat along with ERA to describe the pitcher's skill level and effectiveness. Even today, it is one of the standard categories in fantasy baseball (but that's a story for another day). Times have changed. You don't even have to be on one of the more saber-slanted blogs to notice the almost extinction of the stat.
Let's take a look why. WHIP tells us how many base runners a pitcher allows for each inning he is on the hill. Makes sense right? Certainly a good tool for the casual observer. Unfortunately, not all base runners are created equal. Would you rather have a pitcher with an impossible 0.50 WHIP, but whose only base runners are home runs, or a guy with a more pedestrian WHIP of 1.20, but only allows singles and walks. Ya, I like player number two as well. To really hammer home this point, let's compare WHIP to WAR, the best tool we currently have for evaluating a player's overall value (all information courtesy of FanGraphs). I ran a simple regression using every pitcher who pitched 100 innings in either 2008 or 2009, giving us 272 data points. Plotting WHIP against WAR, we have an equation of "Predicted War = 13.05 -7.71(WHIP)" (maybe should have scaled WHIP to make the coefficient a bit more intuitive. Oh well) and an R Squared of .481.
Now that number means that WHIP isn't completely useless, but it isn't terribly valuable either. By applying the formula, I compared the difference between each players actual WAR and predicted WAR. On average, the model was off by 1.04 WAR, a very significant amount considering the average pitcher was only worth 2.52 WAR.
To put that in a broader perspective, xFIP had a .670 correlation with WAR, and even ERA had a .528 correlation. Just goes to show that there is a lot more to being a pitcher than simply limiting the number of base runners allowed. I am certainly not reinventing the wheel here, but I think it is interesting how fast an "advanced" stat can fall out of favour.