That’s one of the first things you hear when discussing how good a pitcher is. How good is his fastball? Does it have late movement? Does he command it well? Pitchers are generally gauged largely in relation to the strength of their fastball, but is this a true indicator of talent and overall effectiveness? How important is this, really? Can a pitcher with an ineffective fastball still be effective overall? Does an elite fastball mean a sure-fire elite pitcher?
I decided to try and find out. My math skills are most certainly lacking, as are my Microsoft Excel skills, but I compiled some data, made some charts, and constructed some graphs to try and illustrate how much (if any) of a correlation there is between fastball velocity/effectiveness and things like WAR, xFIP and even the dreaded mother-of-all-non-categories, wins.
Before I get into it a brief description of some of the advanced statistics I’ll use is needed. FBv is the average fastball velocity of a given pitcher and wFB is the effectiveness of said fastball measured in runs above average.
Among qualifying starters, Ubaldo Jimenez led the league in FBv at 96.1 mph and knuckleballer R.A. Dickey was at the bottom of the league with an 83.9mph FBv.
Tim Hudson led the league in wFB at 32.1 while James Shields scoured the bottom with a wFB rating at -25.1.
I’m going to use xFIP in this analysis instead of FIP because xFIP attempts to adjust for the park in which the pitcher is playing, which would indicate a more accurate reading of pitcher talent.
I limited my subjects to the top 50 qualified starters in xFIP, which should roughly consist of the best 50 starters in baseball (roughly the top third of all starters). Roy Halladay led the league in xFIP at 2.92.
What I found when I plugged in the numbers was that fastball velocity and fastball effectiveness were only a marginal indicator of how good a pitcher really is; and that it’s no more of an indicator than wins.
The idea, in case you’re not a math-whiz, is that the steeper the line, the more of a correlation there is between the two numbers represented. Each dot on the graph represents a pitcher. This first graph illustrates that xFIP and WAR are strongly related; generally speaking, the better the xFIP, the higher the WAR.
In the second graph, xFIP is compared to fastball velocity. Again, there is a correlation between higher FBv and lower xFIP, but it isn’t very drastic; certainly not as drastic as the relationship between xFIP and WAR. But how about fastball effectiveness?
This graph illustrates the relationship between fastball velocity and effectiveness. It turns out, the relationship is pretty weak. There’s a slight correlation, but perhaps not as much of one as you’d expect. A great arm an effective fastball does not necessarily make.
As expected then, the relationship between fastball effectiveness and xFIP is also not too consistent. In fact, the most effective pitchers tend to have marginally above average fastballs. The dot at the top belongs to Roy Halladay whose fastball isn’t by any means elite, but that is mostly because he doesn’t use his fastball as often as his cutter and sinker, which are far more integral to his success.
To give the argument some context, I wanted to see how much of a correlation there was between something like wins, which most of us now understand tells us very little about how well a pitcher has performed, and WAR and xFIP. It turns out that wins are actually more of a predictor of success than fastball velocity and effectiveness.
In conclusion, it seems we overvalue fastball effectiveness. Yes, it’s important to have an effective fastball, but it isn’t the best indicator of how a pitcher will do. It is possible to have an ineffective or below-average fastball and still be a fairly effective pitcher. Shaun Marcum, for instance, has a changeup that is so good (the best in baseball actually) that he can afford to chuck up a second or third-rate fastball with a -9.6 wFB rating.
Click here for the chart from which these graphs were made
As always, statistical information from FanGraphs
And please, feel free to tell me why this is flawed in the comments. I’m sure it is somehow; I was never good at math.