RANDOMIZING PITCH SEQUENCES. One experiment that Lindbergh and Miller had envisioned trying was to use a random number device to determine the sequence of pitches that their pitcher would throw. They would do so by determining beforehand what percentage of pitches to a given batter would be fast balls, what percentage would be sliders, and so on. Location of pitches could also be randomized.
In the last five years, there has been a lot of attention paid to the sequence of pitches. Major league hitters are said to learn how a pitcher is trying to get them out by learning the pattern of pitches and looking for a certain pitch.
I have been curious about randomizing because one theory in elementary game theory is that players have subconscious tendencies and can be “read” by an intelligent opponent. The way to avoid being read is to randomize.
This article on Randomized Strategies by Steven LaValle says: “One important example in which a randomized strategy is of critical importance is when making decisions in competition with an intelligent adversary. If the problem is repeated many times, an opponent could easily learn any deterministic strategy. Randomization can be used to weaken the prediction capabilities of an opponent.”
Miller writes: “You are not reading about the success of our random-number-generated pitch-calling, though, because of course we hadn’t implemented it—-hadn’t even got close.”
Maybe some day….
My first thought from this (and maybe it’s sort of a tangent) is, “Wouldn’t pitchers and catchers be reluctant to randomize the pitches instead of designing the sequence themselves? Especially if they try it and the randomization yields better results. It would be a blow to their pride.” Then again maybe they’d prefer whatever method wins them more games.