clock menu more-arrow no yes mobile

Filed under:

Spin Rates: Four-seam Fastball Edition

Finding some correlations using whiff rates (with better graphs this time).

MLB: Chicago Cubs at St. Louis Cardinals Jeff Curry-USA TODAY Sports

Last week, I began looking for relationships that give insights into why spin rates of pitches have been an obsession of some over the last few years. I’ll give a short summary of the findings of that piece, but if you’d like to review it in full, you can do so here (special thanks to the comment section for the feedback in the last piece, which helped refine this process immensely). Essentially, the initial analysis showed that the spin rate of various pitches had no visible effect on opponents’ wOBA or exit velocities. There were some issues with the analysis, however. One problem is that we didn’t control for pitch velocity. As you can see below, average velocity and spin rates of fastballs tend to increase together, even if only slightly, but it could still be enough to mess with our results.

For this reason, we have to select a narrower range of velocities as a limit for our analysis. Plotting the average velocities of MLB fastballs in 2021, we see that the majority of pitchers who qualify (minimum 50 ABs where they threw a four-seamer) had an average fastball velocity of between 93 and 95 MPH. Our player pool for this analysis will consist of players whose fastballs fall within that range, which should still give us a lot of data to work with.

Our other problem was that wOBA was a questionable stat to use to look at the effect of spin rates. There are a couple of reasons for this. For one, wOBA needs a substantial sample size before it becomes stable enough to use for predictions. In a lot of the cases we’re looking at, the sample size is probably small enough that opponent wOBA is more or less random, which helps explain why we don’t see any correlation with spin rates. Even after selecting our fastball velocities, spin rate doesn’t seem to affect wOBA at all:

(Quick note: I included labels on points that represent notable former and current Cardinals. The point that goes with the label is at the bottom left of the name. For example, the point for Jack Flaherty is located at the bottom of the F in Flaherty.)

This also leads us to our second potential problem with wOBA: We don’t know for sure that spin rates noticeably affect balls in play. One way to try to confirm this is by plotting spin rates with another offensive statistic. Baseball Savant provides slugging percentage in its spreadsheets as well, so I plotted that to see if it showed any relationship with spin rates:

Slugging percentage seems to be random as well, and similar results are seen when trying to make the same graph with average exit velocity. This all suggests that spin rates don’t actually affect the hitter’s quality of contact if he hits the pitch.

Every analysis we’ve done that includes balls in play has failed to show why spin rates are such a topic of interest. So what about balls that haven’t been put in play? Plotting whiff rates (whiffs/total fastballs for each pitcher) gives us the first inkling of a correlation, with higher spin rates matching with higher whiff rates:

Increased whiff rates should lead to higher strikeout rates, which should lead to decreased offensive production from opposing hitters. The relationships there make sense, but to prove it I plotted whiff rates against opponent wOBA (despite wOBA’s flaws stated above), and we actually do see an inverse relationship between the two:

So it seems that our problem with wOBA might be less about sample size and more about how we draw up the comparisons. I do find it strange that there’s no relationship directly between wOBA and spin rates, but we can draw an indirect correlation using this method. It’s a question I may try to tackle in the future.

The 2021 data provides some interesting insight into our question on spin rates, but do these trends hold when using a larger sample size? To find out, I repeated the analysis but used data from the 2016 to 2021 seasons. The distribution of fastball velocities was similar enough that I decided to keep the 93 to 95 MPH limits for the data:

Direct comparisons that include balls in play also return plots that are random, much like the 2021 data, supporting our assumption that spin rates don’t affect the quality of contact:

However, the relationships involving whiff rates hold very nicely over this stretch of time, indicating that we may be on the right track using these comparisons for our analysis:

The takeaway from all of this is that while spin rates of fastballs seem to have no direct bearing on the quality of a hitter’s contact, they do actually affect whether the hitter makes contact with the pitch at all. Batters seem to hit high-spin fastballs just as hard as low-spin fastballs and the real differences are seen when looking at how often hitters make contact with the pitch to begin with. If they aren’t making contact, they aren’t doing damage or even moving runners over, which is something every pitcher is going to be happy with.

I still find it strange, however, that there is no direct correlation (at least as far as we’ve seen) with the quality of contact. If a pitch is hard to make contact with at all, it seems as though the pitch should be harder to hit squarely even when contact is made. That, though, requires a whole other method of analysis, and I’m trying to keep this article from becoming a full-blown book chapter, so I’ll leave it for another day. This became a little more of a stream of consciousness post than I intended, but as always, feedback and thoughts about methodology are welcome and encouraged.