Saturday, April 28, 2007 | 7:25 a.m.
.
With apologies to Calvin Johnson, JaMarcus Russell and especially Mel Kiper Jr., this weekend's NFL draft probably won't have much of an effect on how the 2007 pro football season plays out.
The best indicator of this, for my money, is that Las Vegas oddsmakers have not only already released betting lines on the first week of NFL games, but also individual team regular-season victory totals for 2007.
The Las Vegas Hilton sports book posted its team-by-team win totals Sunday - nearly a week before the highly hyped draft - with the New England Patriots (12-4 last year) and their projected total of 11 victories topping the list. Oddsmakers attached the lowest number on the board, five wins, to the Oakland Raiders (2-14 a year ago).
As a starting point in analyzing NFL season win totals, it's hard to beat a method embraced by offshore oddsmaker Simon Noble that uses the number of points scored and allowed by each team in the previous season to determine whether the team overachieved or underachieved in terms of its win-loss record.
Another way to put it is how "lucky" or "unlucky" the team got.
The formula is a football version of the old baseball Pythagorean method popularized years ago by author and statistician Bill James. According to Noble, the football equation is a team's points scored squared divided by the sum of points scored squared and points allowed squared. That result is multiplied by 16, the number of regular-season games, to yield a baseline number.
Noble suggests modifying the baseline number by moving it toward the league mean of eight victories. If the baseline number is between 5 1/2 and 10 1/2, adjust it a half-game toward eight; otherwise adjust it a full game toward eight, according to Noble.
Roster changes involving veteran players can then be taken into account, with the draft - given its emphasis on young, inexperienced players - playing a minor role at best in the analysis.
Take the Philadelphia Eagles, for example. Last season they scored 398 points and allowed 328 (finishing 10-6). Plugging those numbers into the formula yields a baseline of 9 1/2 wins, and adjusting toward the mean gives Philly a projected total of nine wins. It's no coincidence the Hilton listed the Eagles' over/under at - whaddya know! - nine wins.
Casual fans or occasional gamblers might be surprised that the Hilton posted a total of "only" 10 1/2 wins for the San Diego Chargers, who led the league with 14 victories a year ago. But the formula suggests 10 1/2 is right on. The Chargers scored 492 points and allowed 303 last season for a baseline of 11.6, indicating San Diego was lucky to have amassed 14 wins. Adjusting toward the mean gives you 10.6, which jibes with the Hilton's number.
In a May 8, 2006, column I pointed out five plays for the 2006 NFL season generated by Noble's formula. Those plays went 3-2 and won 1.2 units (assuming a risk of one unit when the play was an underdog and risking enough to win one unit when the play was a favorite ). That's obviously a tiny sample, but I'll continue to lean on this formula because its methodology makes sense.
The Buffalo Bills present a good example of including roster changes in the analysis. The formula for the Bills yields a projected line of eight wins compared with the Hilton's listed total of six. Buffalo, however, lost playmakers Nate Clements, Willis McGahee and Takeo Spikes to trades and free agency, so I agree with ratcheting down its projected total.
I'll have no wager either way on the Bills, but following are seven plays for 2007 suggested by Noble's formula:
archive
