Though it gained notice this year, MLB.com's "nasty factor" actually debuted in their gameday fields last season. While the goal of quantifying the effectiveness of a pitch presents obvious difficulties for objective measurement, the description of it sounds promising and so after I had re-worked my database to import the data I went digging into the long term trends. Is it worth paying any attention to? The results so far are disappointing.
Since I do not have access to the math that goes into spitting out the nasty factors, I have to rely on spot-checking the results against other tools we use to measure pitchers. After all, shouldn't something like Nasty Factor track well with overall pitcher quality? However, when I see that Tyler Chatwood's average Nasty Factor (45.5) is higher than that of Felix Hernandez's (44.6), Michael Pineda's (44.3) and Roy Halladay's (44.9) then I am left thinking that something is off.
Those values are all pretty close together and that's another problem. If I take all pitchers that have at least 250 recorded pitches this season and average their Nasty Factors, the results go from a highest Nasty Factor of 48 (Mike MacDougal) to a low of 40 (Daniel Schlereth) and the standard deviation is a mere 1.4 points. Such a small spread makes me skeptical there's anything useful there to tease out.
I did three additional checks as well, grouping by the results of each pitch, grouping by pitches resulting in a batted ball and grouping by the hit outcome of those batted balls. Looking at the pitch results shows the only clear marker that I could find; called strikes have an average Nasty Factor of 50 while all other pitch results hovered within a point of 43. In fact, the average swinging strike has a lower Nasty Factor (43) than the average pitch taken for a ball (45), which strikes me as a design flaw.
Looking at batted balls revealed only more clustering. If the average line drive comes on a pitch with a 41 nasty factor and the average pop up is hit off a pitch with a 42 Nasty Factor, is nasty factor able to tell us anything amidst the noise? The average batted out comes on a pitch with a Nasty Factor of 42. The average single is also at 42. The average double and triple are at 41 and home runs are at 39. Sure, it's a downward trend like you would expect but if the spread is that small than the individual rankings aren't particularly illuminating.
What all my investigations boil down to is that while the factors can swing from zero to 100, over large samples, they are simply too closely grouped. It behaves too similarly to a random number generator. This doesn't mean that Nasty Factor is entirely worthless. More data could turn up some correlations. MLBAM could tweak the formula that better represents tough pitches. I may be simply not finding interesting stories that are already embedded in the data. All of those are possible and I think the concept has potential. However, at the moment, I personally have not found the sort of correlations on a league-wide level that should be present for me to have confidence in the factors assigned to individual pitches.