Hearts full of youth,
Hearts full of truth,
Six parts gin to one part vermouth.
– Tom Lehrer, “Bright College Days”
In The Only Thing that Matters is October, I derived parameters for Major League Baseball that track Lehrer’s recipe for martinis, with luck taking the role of gin, and skill relegated to the role of vermouth.
In earlier posts, I’ve fretted a good deal about what I thought was a strikingly large role of chance in backgammon, e.g., All That Luck. Can it possibly be true that the ratio of such to skill in baseball is twice as large as it is in backgammon? What does that mean for the way baseball is played?
My first attempt to derive the luck parameter for baseball relied on estimates from fivethirtyeight.com of the prospects for each of the teams in the current postseason. I have been able to confirm that finding, with another approach.
My current tourney simulator is not particularly good at modeling long, multiple round robins like the Major League Baseball season. But, with a little fiddling, I was able to get it to simulate one of the past season formats – the one that was in use between 1962 and 1968, when ten teams in each league played each other 18 times. Since I was using the whole league, I didn’t need to set an elite threshold. And, sure enough, setting luck to six yielded average results that are pretty similar the actual win totals in baseball during that period:
|simulation wins||actual wins|
The chief difference is that the teams at the bottom of the table do a bit worse, on average, than expected – perhaps that reflects the fairly common practice in which teams that find themselves with little chance to win the league sell off some of their better players to the active contenders late in the season.
So what does this mean for baseball?
Perhaps it’s no accident that baseball’s regular season, at 162 games, is far longer than the seasons for other major sports. If the effect of skill is so subtle, a long season will be necessary for the truly superior teams to emerge.
That suggests that various kinds of baseball that don’t have the luxury of playing long seasons are likely to produce more random results. NCAA baseball seasons consist, these days, of 60 to 70 games for the top teams. For little league, seasons are required to be at least 12 games, and 16 seems to be about the average number.
The fewer games that are played, the less likely it is that the team at the top of the standings is, in fact, the best team. But there’s a strong tendency, in the minds of most sports fans, to want to accept actual results as definitive markers of merit. That’s a less valid assumption for baseball than it is for most other sports.