Monday, August 7, 2006
A different way of looking at it
SG’s last post ran the rest of August projections using log5 to predict that the Yankees would go 12-12 over the next 24 games and go into September 2 games behind the Red Sox. I’ve got some issues with his conclusions.
First of all, there his projection of the Yankees/Sox 5-game set. He made a mistake somewhere, because Boston is projected to go 3-2 that series, not 4-1. They’re not going to win 80% of their games against the Yankees at Fenway. That changes things to the Yankees and Sox being tied for first on 9/1, with the Yankees percentage points ahead.
The second is his HFA adjustment of .042. That’s from Diamond-Mind’s 2002 sim on postseason records, and they used the 2002 HFA of .542 to adjust the advantage that season. Well, it’s not 2002, and HFA this season is only .535, and .540 historically. That makes a marginal difference.
The third is that he’s simply projecting the outcome of each series by rounding the numbers. If the Yankees have a .499 expected winning percentage against the White Sox in US Cellular Field, then he projects they’ll go 1-2, which is a .333 winning percentage, when they’d be only slightly less likely to go 2-1. (They actually should win .478 of the time there). If you don’t round the numbers the Yankees should have a .556 winning percentage over the next 24 games rather than a .542. No, they won’t be winning and losing partial games, but they’re also not going to be missing every single break. Well, probably not.
This is more of an issue for the Red Sox, as they have a real expected winning percentage of .565 but a rounded winning percentage of .625. Sure, if they get the breaks they could do even better than that. But then, they also should have gone 5-2 this last week rather than 3-4.
Then, of course, there is using straight winning percentage for the projection. Not a huge issue, but there’s also Pythagorean Record and BPro’s W3% to try.
Using Pythagorean Record, and using a .535 HFA, and not rounding off the wins and losses until after you add them all up, the Standings on September 1st would look like:
| Team | W | L | Pct. | GB |
|---|---|---|---|---|
| Yankees | 79 | 53 | .598 | - |
| Red Sox | 78 | 56 | .582 | 2 |
If we used BPro’s W3% instead of Pythagorean record (W3% factors in the strength of schedule), the standings would be:
| Team | W | L | Pct. | GB |
|---|---|---|---|---|
| Yankees | 80 | 52 | .606 | - |
| Red Sox | 78 | 56 | .582 | 3 |
And if we go with just plain old W-L record for or projections, but adjust the HFA to .535 and don’t round until afterwards, the standings would be:
| Team | W | L | Pct. | GB |
|---|---|---|---|---|
| Yankees | 79 | 53 | .598 | - |
| Red Sox | 79 | 55 | .590 | 1 |
A lot of things can go wrong between now and September 1st, though. A lot of things can go right, too. This is a very tough stretch the Yankees are starting tomorrow, but the Red Sox are hardly going through an easy stretch themselves.
Page 1 of 1 pages: