“I just take my three swings and go sit on the bench. I’m afraid if I ever think about hitting it, I’ll mess up my swing for life.”—Dick Allen, on the knuckleball

“I just hope that Wakefield doesn't mess up the swings of our hitters for a couple of days.”—Cito Gaston, May 2009

“He messes you up.”—Dustin Pedroia, on Tim Wakefield

Knuckleballers are an odd bunch. Of course, much of what makes them weird also makes them wonderful. Each of us fell for the flutterball in a different way, but whether a Niekro first knocked your socks off, or you got zonked by a Zink (or a Haeger, for that matter) at Rob Neyer’s knee, the practitioners of the pitch have long exerted their hold on the baseball fan’s imagination. Perhaps their appeal stems from the novelty of seeing something rare; maybe it owes more to the inspiring sight of major leaguers swinging through pitches whose velocity readings resemble those that we’ve seen accompanying our own offerings, which dares us to dream.  If you’d like to get really heavy, feel free to meditate on the possibility that we feel some sense of existential identification with a pitch whose destination in life is so often unknown. More likely, though, the darn thing just looks kinda cool, which might be the best reason of all.

Whatever its origins, all of that interest has spawned a thriving myth factory surrounding the Tao of the knuckleball. One of the strangest such tales ascribes powers to knuckleballers that extend even beyond their appearances on the mound: the notion that the butterfly ball “messes up swings,” impairing hitters’ abilities for days after the pitch itself has stopped dancing its way to the plate. If true, we’ve been underestimating the degree to which a knuckleballer can make his impact felt; a direct effect on performance lingering across games could be a hidden factor confounding our attempts at analysis, and it seems like something worth looking into.

Where better to begin our inquiries than with the “Knuckle Knuckle King?” As the longest-tenured knuckleballer on the scene, Tim Wakefield must have earned a reputation for messing up a few swings in his day, right? Sure seems that way. In fact, after a cursory Google search, I discovered that the most recent referenceto his alleged abilities in that area appears to date from just over a month ago. Could it be, to paraphrase the Zombies, that a knuckleballer’s effects are felt even after he’s gone?

I can’t say that I expected to find any tangible evidence of such a legacy; the most I hoped for was a smug sense of satisfaction the next time I caught a member of the media perpetrating a time-honored falsehood upon the baseball-watching public. This particular old players’ tale reeks of confirmation bias and selective memory, and it certainly wouldn’t be the first such chestnut to crack under the strain of a little investigative work.  

On the other hand, if such an effect did exist, and Wakefield did exhibit it, it wouldn’t be the weirdest thing about Tim Wakefield, would it? Even ignoring the obvious idiosyncrasy of throwing a virtually spin-less pitch, we’re talking about a guy who’s still starting in the majors (in his 16thseason with the same club, no less) a month shy of his 44thbirthday, despite not throwing a pitch over the speed limit of a number of states all season (in Massachusetts, his pitches might get pulled over, but could probably get off with a warning). In addition to his failure to top 74 mph on the gun this season, Wakefield sports a career BABIP of .282 in over 3,000 innings pitched, and no overall platoon split to speak of. It’s possible that the most surprising result I could have obtained would have been one that suggested that he might actually be normal in some respect. Well, no such luck.

With the invaluable aid of some clever querying from Eric Seidman, I came up with the following table, which displays statistics for the batters who’ve faced Wakefield over the course of his career. It should be noted that this data does not include Wakefield’s 432ndstart, but it’s probably safe to say that if we find nothing lurking in the first 431, there’s nothing to be found. The top three lines show the batters’ rate statistics in the games that they played one day after facing Wakefield, two days after facing Wakefield, and three days after facing Wakefield, respectively. The bottom line represents those same batters’ overall lines in the seasons that they faced Wakefield, weighted by the number of times that they faced him:


Time Period









1 Day After









2 Days After









3 Days After


















If a moderate knuckleball “hangover” effect exists, this is essentially what we’d expect it to look like. After facing Wakefield, strikeouts go up, homers go down, and overall performance suffers, with rates rebounding on the second day of release, and returning to near-normal levels by the third day of deliverance. But before we start handing out hangover points and adjusting win values for knuckleballers, let’s load this up with some asterisks. The results do appear to be statistically significant—for instance, that day-after OBP value is 2.4 standard deviations from the mean, with a two-tailed P-value of .0164. That said, a test of statistical significance can miss any number of significant confounding variables that we might be able to suss out on our own (after all, how do you think we got Enron?).

With the possibility of random variation more or less dismissed, the first explanation that comes to mind is a contextual one: if Wakefield has usually pulled up the rear of the rotations to which he’s belonged, the batters opposing him would have faced top-of-the-line talents in subsequent games, which could account for the observed offensive dip. I didn’t rule out this possibility, but it’s worth noting that in the 13 times that he’s started a season in a rotation, Wakefield has first appeared, on average, in the third game on his team’s schedule, ranging from his only opening-day start, with Pittsburgh in 1993, to his stint as a fifth starter for Boston in 2008 (and encompassing every point in between).

In light of injuries and in-season fluctuations in performance, start-of-season position might not be the best proxy for rotation slot. However, if Wakefield has managed to hold down the equivalent of a No. 3 spot during his time in the rotation, we might have expected his opponents to overperform their seasonal lines in the two games following his efforts (though it’s also possible that Wakefield’s rotation mates have generally outpaced their counterparts with other teams).

There remains more work to be done here, in addition to a more rigorous adjustment for the strength of the starters in games following Wakefield’s outings. If the knuckleball hangover exists, we would expect its most pronounced effect to be felt in the immediate aftermath of the knuckleballer’s departure. Batters have hit .259/.313/.403 against the pitchers who followed Wakefield in games that he began, but without adjusting for the fact that most of the pitchers against whom those numbers were accumulated were late-inning relievers, whom we would expect to suppress production, those figures add little fuel to the fire.

Another measure we could take to assess the magnitude and veracity of the effect might yield considerably more entertainment value: we could simply look at more knuckleballers. As the busy social schedule of a sabermetric writer allows, I’ll try to do just that in the weeks to come. Until then, I welcome any comments or suggestions that might uncover an alternative explanation, beyond Wakefield’s hangover-inducing deliveries; what’s more, if the effect’s existence holds up, his upcoming opponents would be delighted to hear your recipes for a remedy.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
One thing you could try is restricting the sample to instances where the batter's next game came against a different team.

Or rather than using raw stats, use the delta between the next starter's season stats and the stats generated by batters against him after facing Wakefield the previous day.

Of course there are also park effects. Then again, Fenway is noted as a hitter's park, so park effects may actually amplify the statistical significance here.
Ha. Very cool. I had often wondered about this myself. I look forward to a more rigorous analysis.
Fascinating. I guess we don't knuckleballers throwing BP, then?
Nice article. The fact that most batters are facing the same team (Red Sox), is probably the largest potential bias of the study -- as Nate Sheetz points out.

Look forward to a follow-up.
Once you've fleshed this out, I'd be interested in seeing your thoughts on how the Red Sox could use (or could have used) this to gain a tactical advantage.
I really enjoyed this. I will definitely read any further research.
I'd like to see a comparison of these numbers between Wakefield and his rotation mates. We might get a better idea of how abnormal his numbers are.
I think one of the attractions to a knuckleball pitcher from the fans standpoint is the idea that we could pitch in the majors if we could develop a knuckler. We know we couldn't throw 95 mph, or throw a 12 to 6 curveball. But if properly taught and with a little bit of luck ; we believe we could throw a knuckleball.
"Batters have hit .259/.313/.403 against the pitchers who followed Wakefield in games that he began, but without adjusting for the fact that most of the pitchers against whom those numbers were accumulated were late-inning relievers, whom we would expect to suppress production, those figures add little fuel to the fire."

This seems to be the avenue to investigate. Can you see how pitchers who relieve Wakefield perform in that game and compare it to those same relief pitchers who relieve other starting pitchers? The Red Sox bullpen has been relatively stable and the relievers tend to throw a lot of innings so there should be a good chunk of data to consider.