OOBL Efficiency

-or-

What's Luck Got to Do (Got to Do) With it?

ESPN recently published an article on Team Efficiency- the basic premise being that you should be able to tell a team's record from its run differential, which you should be able to predict from TBW (total bases plus walks). The statistical analysis is, obviously, very simplistic and ignores a lot of things that go into winning baseball, like stolen bases. However, it presents a basic metric by which to measure how teams should have done vs. how they actually did. The conclusion of the article is that, on average, there is no such thing as a 'winning team attitude' or 'performance in the clutch'- these things average out in the long haul to how much you're able to produce. Of course, this isn't entirely true in an OOTP league- we have a quantitative clutch rating. But the analysis is interesting nonetheless.

Let's start with looking at how teams actually did compared to what their run differential predicts. Bill James said a long time ago that a team's expected winning percentage can be taken as the number of runs they score squared divided by the sum of the squares of the runs scored and runs allowed (win%=RS²/(RS²+RA²)). Before delving into how this turned out in OOBL, keep in mind that using the expected winning percentage given by this formula is exceedingly accurate- the average team misses its projected winning percentage by a whole 3.2 games (average for all teams since the advent of the 162 game schedule).

The first thing to point out is that the pythagorean method has produced a farily accurate predictor on average- the average difference between a team's predicted and actual number of wins is 3.2. This means that a team that won 3 more games than it should have based on run differential got significantly lucky, and could expect to win fewer games next year if the same lineup produces the same numbers. More interesting is that even the playoffs have produced only one real upset to this point- the Yankees over the Mariners.

What can we learn from the difference between the actual wins and the predicted wins? The most prominent one relates to the wild card teams- New York and Los Angeles. New York dramatically underachieved (well, got unlucky) in the regular season, but sneaked into the playoffs as the second best team in the AL on paper. They got lucky at the right time- against Seattle's pitching- and find themselves in the World Series. In the NL, Chicago got dramatically unlucky and missed the playoffs. LA snuck in, but lost against a significantly better team- St. Louis. This isn't to imply that Chicago is sitting pretty for next year. The pythagorean method incorporated players under/overachieving, which will change from year to year. However, you have to figure they are, at least preliminarily, the favorites for the wild card (5 game lead is significant statistical seperation between them and SF, or LA). Remember that, on average, teams perform within a few wins of their predicted wins, meaning Chicago should probably win more games next year, even if they get the same production from the same players.

But how about going one step further in 'luck measurement'- the luck of actually scoring and allowing runs? As a measure of what leads to runs being scored, I'm going to use total bases plus walks (TBW)- simply enough, getting on base and driving guys home. Obviously, the more TBW, the more runs you score, and the fewer TBW your pitchers allow, the fewer runs your opponents score. Unfortunately, I don't have access to the team total TBW allowed, so I'll be making a bad assumption that TB=1.58*HA (this is the average for all of baseball- incredibly, including pitchers, the NL actually slugs the ball harder than the AL, but the difference is negligible).

Clearly, there is a linear relation between TBW and runs; define this ratio as an efficiency (RS/TBW or RA/TBWa). This efficiency with which you turn TBW into runs is historically constant- just because you're more efficient at doing it one year has no relation to how efficient you'll be next year. From this, we can attempt to figure out how lucky a team is in scoring runs or allowing runs by taking your production of TBW and multiplying by the league average efficiency (.272) to arrive at expected runs and expected runs allowed. Very surprisingly, the assumption that on average, we can get TB allowed simply by multiplying hits allowed by 1.58 did not systematically skew the relationship with runs allowed- i.e., teams that allowed fewer hits didn't also systematically allow fewer runs.

Let's deal first with stats for the leagues offenses. A positive diff number means you scored more runs than you should have based on your production. The average deviation from the way we predicted runs scored is 30- a diff of larger than 30 is significant. This means, most prominently, that Florida, Cincinnati, and San Diego can expect to score more runs next year given the same production, and that New York, Seattle, Texas and St. Louis can expect less. The deviation from expected runs allowed is larger- 45- because of how we estimated total bases allowed. By taking a league average, we lost details like certain pitching staffs allowing fewer home runs because their pitchers are talented in that area, or ballpark effects (Colorado). However, there was no systematic error- all good pitching staffs don't necessarily give up fewer homers per hit than bad ones. Looking at the stats, a positive diff number means you allowed fewer runners than what you should have. This means that Seattle, Texas and Florida can expect to allow more runs given the same production next year, and Toronto, Kansas City and Milwaukee can expect to allow fewer runs (I'm ommitting Colorado because Coors means that you will give up more runs per TBW because more hits are homers; this doesn't affect the offensive numbers because we didn't have to take a league average to translate hits into total bases). Strangely, there was a correlation between very efficient offenses and pitching staffs and the OOTP clutch rating, but it was obscure. You had to have a LOT of players with GREAT clutch in order to be particularly efficient- e.g. Arizona's pitchers. And this didn't always ensure increased efficiency (White Sox)- good clutch players had to also be playing well in order to really make a difference.

Finally, let's look at the projected wins diff column- a positive number means you won more games than you should have. Boston, Oakland, Seattle, St. Louis, Arizona, and Los Angeles won more than they should have based on just raw production, and Kansas City, Milwaukee, and the Cubbies lost more than they should have based on just raw production. In many cases (like Arizona, Seattle, and St. Louis), this meant that great teams ran away with a division, but in once case- Los Angeles and Chicago (N), LUCK made the difference between going to the playoffs and watching them on TV.