This was originally written as a response to Nico's front page post, the one where he says "Good teams don't play a lot of close games". But it got kind of long, so I've made a separate FanPost out of it instead.
Nico, I'm surprised that this post hasn't been challenged more. Maybe it's true that all the stat guys have left AN. Or maybe they just don't want to post in a Nico post because it always turns into an ugly food fight when they do.
But it needs to be said.
As you know, I'm not an absolutist, so I'm not going to say "the numbers prove you wrong." Still, there is strong evidence that your "great and important point" is simply traditional conventional wisdom that does not accurately describe reality. On the other hand, Beane's observations, as unsatisfying as they feel to us, do make sense.
Good teams don't win by avoiding close games. Good teams win by outscoring their opponents. Slightly good teams win by slightly outscoring their opponents. Bad teams lose by scoring less than their opponents. Slightly bad teams — which is what the Oakland A's seem to be this year — lose by scoring slightly less than their opponents.
In all cases, there will be random fluctuations along the way that will look like meaningful patterns but really are not. Sometimes there will be close games, and sometimes there will be blowouts. Sometimes there will be slumps when the team achieves below its base talent level, and sometimes there will be streaks when the team overachieves. This is not just the nature of baseball; it is the nature of random fluctuation and is a basic mathematical truth.
We can model a simplified version of baseball in which each batter has a base skill level where any time he goes to the plate he has an X% chance of it resulting in a run. Different players have different values of X, where a good hitter might be 20% and a bad one might be 10% with everyone else scattered somewhere in between. If we suppose that each player simply rolls the dice and maybe creates a run or maybe doesn't, according to his basic skill level, what do we find? As you'd expect, we find that the team with better players will end up with the better record in the long run, but along the way there are close games and blowouts, and there are streaks and slumps both at the team level and at the player level.
Of course real baseball is more complicated. Scoring runs is a function not just of the hitter's skill, but of that of the pitcher and defense as well; and runs are usually scored by a combination of multiple events rather than just homers. But the fact remains that all the scoring fluctuations we see are adequately explained by assuming each player has a base talent level that doesn't change.
Does that mean I think players have a constant talent level that never changes? Of course not. I believe Daric Barton really is a better hitter now than he was two years ago, and Eric Chavez really is a worse hitter now than he was six years ago. Players can get better and players can get worse. Same goes for teams. I believe that this year's team is not as good as the 2002 team. But I don't believe that this team in June is horrible and this team in May was great. The team really hasn't changed that much.
So how do you know what's random noise and what's a real trend? What is the most meaningful scale at which to evaluate the team's performance? One person who asked that question and looked carefully at the game for answers is Bill James, and what he found is expressed in his famous Pythagorean win formula. This formula is not a law that stat-heads impose upon baseball to make it submit to their will; it is a pattern that is already in baseball which observers of the game have discovered there. What Bill James' pythagorean formula reveals is that a good way to evaluate a team is the total runs it scores and the total runs it allows. These numbers give a good representation of what the team's basic skill level is, better even that the win-loss record does.
If we look at the Oakland A's in this light, we see that in May the team scored 99 runs and gave up 113; during that month the team's record was 16-12, which is a significantly better result than normally associated with those total runs. So far in June, the team has scored 99 runs and given up 109; this month the team's record is 7-16 so far, which is a significantly worse result than normally associated with those totals runs. On the entire season the A's have scored 304 runs and given up 321. The team's record on the season is 35-40, which is just about right for those total runs.
What I see here is that the A's are a slightly below average team with a slightly below average record. They have been the same slightly below average team all season, but in May several games tilted in their favor and they looked better than they really are, and then in June several games tilted against them and they've looked worse than they really are. Judging from Billy Beane's recent comments ("We've lost a lot of close games"), I'd say he sees the same thing.
Lately I've heard a lot of explanations of what's wrong with the A's — they don't care enough, they've lost confidence in their manager, they're in too many close games, Lew Wolff wants to lose so he can move to San Jose, etc. Everyone seems to have a theory, and most of them rely on the idea that they understand the team or baseball better than Billy Beane does.
I'm not a debater. You all have a right to believe whatever you like, and I'm not going to say I've proven any of you wrong. I would just observe that the A's entire season so far is easily and adequately explained by the simple mathematical nature of the game. All these other theories I'm hearing are not necessary to explain anything, and I've seen scarcely any evidence for any of them. We are simply a slightly below average team playing exactly like a slightly below average team typically plays.