Should Jays Fans Worry about R.A. Dickey?

September is fretting season for Toronto Blue Jays’ fans. If the team is doing badly, the hand-wringing is all about whether the front office will make the off-season moves needed to field a competitor next year. If the team is doing well, it’s all about whether they will choke down the stretch, and who is holding them back.

This September the Jays-fans’-angst gaze has fallen heavily on R.A. Dickey, who was a big part of last year’s pennant run, and who is clearly having one of his rockier seasons, winning only 9 of his 28 starts with an ERA of 4.60. When Dickey is on, he is unhittable. When he is off, he might as well be tossing BP (*batting practice). So far, he has mostly been off.

What makes Dickey unhittable when he is good is his knuckleball, a pitch few use because few can throw it. Under the right circumstances, a knuckleball has a highly unpredictable flight path, which means that when the pitcher releases the ball hitters (as well as catchers!) cannot anticipate where it will cross the plate. A knuckleball literally wobbles en route, bobbing and weaving in a seemingly random manner. To paraphrase Muhammad Ali, Dickey’s knuckleball can float like a butterfly and sting like a bee. You can see this very clearly in this slow-motion .gif.

But what makes knuckleballers unhittable in one context makes them mortal in another. Compared to hard-throwing pitchers with the standard fastball/sinker/changeup repertoire, knuckleballers don’t have anywhere near as much control over where their pitches go. Inconsistency is inherent to the knuckleball-throwing enterprise. This is because the knuckleball is the perfect chaotic system.

In physics, a chaotic system is one that is extremely sensitive to initial conditions and easily perturbed by stochastic (i.e., random) influences. The defining feature of a knuckleball is that it leaves the pitcher’s hand without any rotation whatsoever. As the ball moves toward home plate, ambient turbulence will knock it off course, and perhaps also induce a slow spin in an unpredictable direction. Every other pitch—fastball, forkball, curveball, sinker—leaves the pitcher’s hand with a deliberately-imparted spin that will, as a result of Bernoulli’s principle, decrease air pressure on a specific side of the ball and induce a motion in that direction. Good pitchers can place these pitches very accurately. Good batters can identify them shortly after they leave the pitcher’s hand and anticipate where they will go. With a well-thrown knuckleball under ideal conditions, they have no idea. Throwing a knuckleball is a bit like curling without giving the rock an initial rotation: as it travels down the sheet, it will inevitably pick up a random spin of its own and go somewhere totally unpredictable. In curling, that’s insane. In baseball, it’s brilliant.

The problem is that very minor changes in initial conditions can turn a knuckleball into a big fat grapefruit that screams “Hit me please!” These minor changes can include the roughness of the ball’s surface, the length of the pitcher’s fingernails, the timing of his release, whether the roof of the SkyDome (er, Rogers Centre) is open or closed, atmospheric pressure, wind speed, humidity, and so on.

The fact that a knuckleball is a chaotic system means that, by their very nature, knuckleballs have a relatively high circular error probable, or CEP. Even with perfect initial release (i.e., the pitcher knows exactly where he wants the pitch to cross the plate and releases the ball appropriately), the odds that it will cross the plate at any given distance from the target are significantly greater than with any other pitch, whose tight rotation minimizes mid-flight perturbations. Since knuckleballs are relatively slow pitches, unless circumstances are just right, they will have flight paths that are relatively easy to anticipate and—if their flight path takes them over the plate—relatively easy to hit.

It is no surprise, therefore, that knuckleballers as a group have less impressive records than hard-throwing pitchers. They will be unhittable a smaller proportion of the time. According to the Bleacher Report, the top 10 knuckleballers of all time have net negative win-loss records (49 percent), an average ERA of 3.81, and an average career win total of 180 games. Only one knuckleball pitcher—Phil Niekro—has won more than 300 games. In contrast, the ten best starting pitchers of all time—none of whom threw a knuckleball—won 61 percent of their games, had an average ERA of 2.73, an average career win total of 330, and a dramatically higher total award haul.

Top 10 knuckleballers W% ERA W
1 Phil Niekro 0.535 3.49 318
2 Tim Wakefield 0.527 4.40 200
3 Charlie Hough 0.506 3.77 216
4 Hoyt Wilhelm 0.479 2.67 148
5 Dutch Leonard 0.560 2.99 139
6 Wilbur Wood 0.455 3.60 164
7 Joe Niekro 0.511 3.96 221
8 Tom Candiotti 0.480 3.97 151
9 R.A. Dickey 0.417 5.58 109
10 Bob Purkey 0.451 3.71 129
Top 10 starting pitchers W% ERA W
1 Walter Johnson 0.580 2.35 417
2 Christy Mathewson 0.610 2.37 373
3 Ed Walsh 0.547 2.12 195
4 Mordecai Brown 0.612 2.26 239
5 Grover Alexander 0.605 2.96 373
6 Pedro Martinez 0.649 3.00 219
7 Roger Clemens 0.649 3.22 354
8 Randy Johnson 0.664 3.41 303
9 Cy Young 0.601 2.66 511
10 Tom Seaver 0.595 3.00 311

Why would any manager field a knuckleballer then, you might ask? Because when the knuckleball works, it really, truly, genuinely is unhittable. Even the best power pitcher is hittable when he is throwing his best stuff. The odds are low, but they aren’t negligible. Under the right conditions, the odds of hitting a knuckleball are negligible. And in the average rotation, a good knuckleball pitcher—who will also, of course, have other pitches up his sleeve—is very likely to be one of the top five starters.

All of this is by way of saying that unless Dickey is injured, distracted, or otherwise impaired, there is no reason whatsoever to worry about his having had a bad season thus far. It’s in the nature of the pitch he throws. He could have a stellar October without doing anything different at all.

So, at least, we may reasonably hope. Chaos theory says so.