Baseball Musings: Wizard of Oz

April 25, 2005

Wizard of Oz

When do you start believing a trend is more than a trend? When do you start believing a player or a team has some intrinsic ability to accomplish a feat?

The White Sox are off to a 9-1 start in one-run games (they're 6-3 otherwise, still very, very good). But last year they also did very well in one-run games, going 28-18, 2nd in the AL behind the Oakland Athletics (33-19). That makes them 37-19 in the closest of games under the guidence of Ozzie Guillen, a .661 winning percentage.

We're very close to this not being just luck. It all depends on what you think the White Sox intrinsic winning percentage is. One can argue that in one run games the winning percentage of any team should be close to .500. In games that close, the teams are by definition evenly matched, and the contest can turn on a bad hop or a single mistake by a pitcher (see Joe Blanton last night). If teams should play .500 in one run games, then we would expect the White Sox to win between 21 and 35 games given 56 opportunities (the 95% confidence interval).

If instead, we base it on the White Sox 2004-2005 combined winning percentage, .541, we get a 95% confidence interval of 23 to 38 wins. The White Sox are in this interval, but very close to the high end.

A better way to get a team's intrinsic winning percentage over this period is to figure what their winning percentage should be based on their runs scored and runs allowed. The White Sox under Guillen have scored 951 runs and allowed 895. That works out to be a winning percentage of .530 under Bill James, Pythagorean Formula. At .530, the 95% confidence interval is 22 to 37 wins in 56 games. Guillen's right at the top.

At this point the White Sox record in one-run games sits on the fine line between luck and skill. I had my doubts about Ozzie Guillen's skill as a manager when he took the job; I have to give him lots of credit for the success so far.

Posted by David Pinto at 08:48 AM | Management | TrackBack (0)

Comments

It's nice to hear kind words for the White Sox, and especially Ozzie Guillen, from the sabermatic community. As a longtime White Sox fan I can only say that I don't really trust it yet. Years and years of let down have taken their toll.
Still...the argument that being able to play for that extra run (i.e. smallball) should help a team in a close game is an attractive one.

Posted by: Peder at April 25, 2005 10:16 AM

The Sox do always seem to get the clutch hit, no offense to Mr. James. They seem so much more professional than last year's outfit. And when Big Frank gets back, well, they have the chance to get better.

One thing I'm worried about is the pitching. Yes, it seems everyone has taken a step forward, but sooner or later we'll have to win a slugfest. I don't know if we can go anywhere if our pitching doesn't take us there.

Posted by: MikeQ at April 25, 2005 03:14 PM

Keep in mind that with 30 teams, one would expect 1.5 to be outside the 95% confidence interval.

Posted by: Yaks Hairbrush at April 26, 2005 12:23 AM

Bill James did a study in his '85 Baseball Abstract on fast starts and what they meant. The context there was the '84 Tigers and at what point in the season we should have been able to tell that they were going to have a great season. I'll take a look at it this week and see how that relates to Chicago's great start.

Posted by: zlionsfan at April 26, 2005 12:14 PM

Two items: one is that only one team started 16-4 between 1965 and 1984 (excluding the strike-shortened '72 and '81 seasons), and it finished 97-63. A good sign, but hardly a predictor.

The other is that James has a formula to estimate winning percentage based on current record, but it doesn't seem to work right.

Three steps: first, calculate X.
X = 100 - ([80 * (W / 3.5)]/[W + 2L + 4])

Next, calculate winning pct. for the remainder of the season.
Pct = (W + X)/(W + L + 2X)

Finally, calculate total wins by multiplying remaining winning pct. by remaining games. The problem is that in his example, he estimates that the '84 Tigers would finish 112-50 after starting 35-5. Using the formula, I have them finishing 105-57 - nearly spot-on, but not what the book says.

Posted by: zlionsfan at April 27, 2005 01:28 AM