The most annoying thing about well-written baseball books is their tendency to be naïve – nay, infantile – in their lack of statistical rigor. It's far too easy to tell charming "Just So" stories about stolen bases, sacrifice flies, intentional walks, and slugging averages than it is to do a proper analysis of what really matters on the diamond.

Take the "Big Inning Theory" (please!). My goat was gotten again last month as I read Thomas Boswell's How Life Imitates the World Series, a collection of entertaining if uneven essays from the 1970's and early 80's. In "The Big Bang Theory and Other Secrets of the Game" Boswell claims:

... Baseball is a game of big innings. In a majority of games, the winning team scores more runs in one inning than the loser does in nine innnings. To grasp the game, we must start there.

He goes on to say:

Look at the World Series since 1960. In those 133 games, the winning team followed the Big Bang Theory by scoring as many – or more – runs in one inning as the other team did in the whole game 106 times: 80 percent. In 75 games the winner scored more runs in one inning than its foe did in all nine. In only 27 games did the big-inning notion not apply. That breakdown – 75-30-27 – is worth considerable cogitation.

Hold the phone and look closer. Is a game in which the winner's Big Inning consists of only one run any sort of evidence for the value of a Big Inning? Hardly! Yet such games are included in Boswell's count. How about a shutout, in which the losing team scores zero and any Big Inning is thus totally irrelevant to the game result? Hardly! Yet those games are likewise part of the Big Inning myth-conception.

Inconveniently for Big Inning theorists it's straightforward to look at the actual scoring in the precise 133 World Series games cited by Tom Boswell. From 1960 through 1981, http://www.baseball-reference.com/ reveals that there were 74 games – not Boswell's 75 unless I've repeatedly miscounted – that fit what I'll call the Strong Big Inning Theory, SBIT, in which the winner scores more in one inning than the loser's total. In 8 of those 74 SBIT games the winning team scored only a single run in its biggest Big Inning. Slicing the data another way, fully 23 of the 74 SBIT games were shutouts in which the losing team plated nil and any Big Inning was meaningless overkill.

So instead of a truly Big Inning that dominates the majority of baseball games, at best such an inning appears to possibly have significance less than 40% of the time. How about the Weak Big Inning Theory, WBIT, when the winner scores at least as many runs in one inning as the loser does in toto? Boswell's 106 such cases – actually 105 by my count – include fully 16 in which the winner's so-called Big Inning consisted of a single run. (The same 23 shutouts that occurred among SBIT games likewise happened during WBIT games, as must be the case by definition.)

Question: should a baseball manager make decisions designed to maximize the number of runs scored in a single inning of a game (the Big Inning Theory) at the expense of reduced total run production? Answer: Not if s/he wants to increase the team's chance of winning!

(cf. WorldSeriesLines (22 June 2002) and HeartOfTheOrder (3 July 2002), SparkyAndSandy (24 Jul 2002), LeonardKoppett (23 Jul 2003), ...)

TopicRecreation - TopicScience - TopicLiterature - 2006-01-31

(correlates: LeufNet, SparkyAndSandy, WorldSeriesLines, ...)