Flatlanders Attack!

Cite overwhelming anecdotal evidence

If you're not lucky enough hear WFAN broadcast out of New York, you're missing a treat. However, after this you are just lazy as the best of their shows are available as audiocasts over the web. Their top rated show are hosted by Mike Francesca and Chris Russo, who know their stuff and provide great entertainment in discussing baseball. They give call in guests ample opportunity to express their views, but you better come with your best stuff because they are smart and they do their homework. They are, however, flatlanders. The current Red Sox series has incited them to attack the Red Sox style of baseball, which is code for Bill James, Moneyball, Theo, Billy Beane (everyone must pick on Billy, all the time) and all who dare question the wisdom of the stolen base, sacrifice bunt, and the game winning RBI. They blame the appearance of Red Sox lethargy on emphasis on base percentage and good pitching, at the expense of little ball ways of moving runner over. One of their correspondents indicated that sabermetric methods were just plain wrong, as he had overwhelming anecdotal evidence as to how stolen bases and sac bunts directly lead to winning ball games. This naturally deevolved into a rant that stopped just short of the Red Sox being un-American and the virtues of high payrolls.

But the truth is, the numbers don't lie. As overwhelming or anecdotal the flatlanders evidence may be, there is little statistical evidence that our treasured strategies are effective. So this raises two questions: are the statistics lying (as they often do) or what is it about anecdotes? For sure, the numbers don't lie, but they may not be telling the whole truth. Over a long season, considering every equal inning state as equal, considering every inning as equal, the truth is clear: you have to steal 75% of your stolen base attempts just to break even, and every sac bunt success is just a breakeven proposition.

There are many situations where the alternative to stolen base/sac bunt breakeven is worse than breakeven. Situations like the pitcher batting, or an overwhelming pitcher on the mound, or an extremely fast runner, or the next batter is just a high average singles hitter. These are situations different from the average inning state assumptions: average batter, average pitcher, average hitter in the on deck circle. When these planets align just right, of course the expected number of runs out of a particular situation will differ from the average inning state. The opposite side of the coin exists, of course: one would never bunt with the #8 batter up, 1 out, and the pitcher up next. So in any given at bat, the actual number of expected runs may differ from the average for the inning state because of who is up and who is on deck. But this happens for all the time for all inning states, not just candidates for bunting or stealing. And, it is these differences in average inning state and actual current state that a manager is paid to exploit. That's why we put the better batters at the top of the lineup, bat righties vs. lefties, and bring relievers into the game in the late innings. On average, though, bunting and stealing are losing propositions.

Another possibility for the argument is confusing inning state with game state. The worst part of the stolen base is that if you are thrown out and the next batter homers, you have thrown away a run that may be very valuable. In the early portions of the game the extra run matters quite a bit. In the late portions of a close game, all you may need is one single run, and it doesn't matter whether the runner stealing 2nd or the home run hitter. These differences should show up however in positive game state changes. Running the numbers through the game state matrix however, show very little variability in bunting or stealing profitability based upon the inning number or the home team lead.

The part about overwhelming anecdotal evidence is interesting. I think it's a matter of perception. When it works, the extra base taken followed by the hit that scores the run, it's exhillarating, it seems that a run was produced out of nothing. When it doesn't work, it's like "Well, we gave it a good try". Let's face it, small ball is exciting baseball. But each run, exciting or dull, counts for one and only one run. It’s difficult, however, to remove emotions from our perception of each play. With small ball, we get a selective memory, where we remember the successful plays and forget the ones that didn't work. Sort of like a golf round. Is it possible that the way some runs are scored, being inspirational, are more important than others (factoring out game state considerations?) If so, they are they are going to be extremely difficult to measure. Overwhelming anecdotal evidence gets you thrown out of math class, doesn't fly in a courtroom (remember OJ), lacks sufficient rigor among sabermetrarians, but apparently gives you instant credibility on a radio talk shows and on tabloid sports columns.