Wednesday, October 22, 2003

Since Gregg Easterbrook has been fired from his NFL-commentary job at ESPN, Iberian Notes will try to fill the gap Easterbrook leaves.

Now, I love Bill James's writing about baseball and statistics and conventional wisdom, but there are a lot more individual pitcher-batter pairups in baseball, enough to make individual statistics like on-base percentage or strikeout-walk ratio comparable among various players. You can use stats to say that player X is a better hitter than player Y.

That's harder to do with stats in football. Sure, Priest Holmes is obviously an above-average running back. But he doesn't necessarily have the highest yards-per-game average or the most touchdowns-per-minutes played or whatever because things all depend on the team and the context in which the play is called--Priest is a lot more likely to rack up big yards in a game against a battered defense than he is against Tampa Bay, for example. Also, the whole team has an effect on Priest's success, much more so than in baseball. If the O-line doesn't make its blocks and if the WRs don't decoy the cornerbacks and if the QB botches the handoff, Priest loses two yards instead of gaining eight. Whether Barry Bonds jacks a homer or not depends only on him and the pitcher.

So it seems to me that, while established statistics are a pretty good indicator of some things about individual players, full team stats are the best way to judge the quality of a football team as a whole.

What's the point of football? To win the game. How do you win games? By scoring points and not giving them up. Let's say that the ratio of a team's points scored to points given up is a measure of its quality. Teams that score a lot and don't get scored on much will tend to beat teams that don't score much and do get scored on a lot. I buy that. That seems clear to me. Our only problem is sample size--that is, you can't base a firm opinion on the results of only one or two games. You need five or six games to establish that a team has a pattern of scoring more points than it gives up.

Well, every team has played either six or seven games at this point. That's a between a third and half of the 16-game NFL season. I'm going to say that's enough information to figure out each team's ratio of points scored to points given up and use it as the basis for a Power Rankings.

Rank. Team, Games Won-Lost, Points For / Against, Ratio

1. Vikings, 6-0, 179/104, 1.72
2. Colts, 5-1, 178/105, 1.70
3. Chiefs, 7-0, 208/125, 1.66
4. Rams, 4-2, 170/108, 1.57
5. Broncos, 5-2, 178/115, 1.55
6. Dolphins, 4-2, 118/77, 1.53
7. Cowboys, 5-1, 150/100, 1.50
8. Seahawks, 5-1, 146/104, 1.40
9. Buccaneers, 3-3, 134/97, 1.38
10. Titans, 5-2, 194/150, 1.29
11. Forty-Niners, 3-4, 159/126, 1.26
12. Bills, 4-3, 138/110, 1.25
13. Packers, 3-4, 200/166, 1.20
14. Patriots, 5-2, 145/126, 1.15
15. Panthers, 5-1, 118/105, 1.12
16. Ravens, 3-3, 134/126, 1.06
17. Jets, 2-4, 94/94, 1.00
18. Browns, 3-4, 112/121, 0.92
19. Saints, 3-4, 152/168, 0.90
20. Giants, 2-4, 105/123, 0.85
21. Bengals, 2-4, 111/132, 0.84
22. Eagles, 3-3, 95/119, 0.80
23. R******s, 3-4, 135/171, 0.79
24. Raiders, 2-5, 125/161, 0.77
25. Steelers, 2-4, 111/146, 0.76
26. Jaguars, 1-5, 110/154, 0.71
27. Chargers, 1-5, 115/169, 0.68
28. Lions, 1-5, 101/160, 0.63
29. Texans, 2-4, 100/170, 0.59
30. Bears, 1-5, 97/176, 0.55
31. Falcons, 1-5, 114/220, 0.52
32. Cardinals, 1-5, 82/140, 0.46

There's obviously a correlation between the points scored/allowed ratio and the win-loss record. I actually used to be able to figure some of that stuff out. Can't anymore. Anyway, if we take that ratio as an indication of the real quality of a team, we need to name the teams that are better than their records (and are likely to see their won-lost records improve) and the teams that are worse than their records (who are likely to see their won-lost records decline).

Better than Won-Lost Record: Rams, Buccaneers, Forty-Niners, Jets
Worse than Won-Lost Record: Patriots, Panthers, Eagles, R******s

Of course, we also need to pick our playoff winners; we won't be able to tell if this system works until it's tested in the playoffs. If I get most of these right I claim the system works.

National Football Conference
East: Cowboys
West: Rams
North: Vikings
South: Buccaneers
Wild Cards: Seahawks; Forty-Niners or Packers

American Football Conference
East: Dolphins
West: Chiefs
North: Ravens
South: Colts
Wild Cards: Broncos; Titans or Bills

And, of course, we have to put our money where our mouth is and make some betting picks. The only two games where I think using this chart gives you a significant advantage are the St. Louis Rams at the Pittsburgh Steelers and the San Francisco 49ers at the Arizona Cardinals. Both the Rams and the Niners are far above their respective rivals in the rankings, about 0.80 percentage points of difference in both cases. The current betting lines are the Steelers favored over the Rams by 1 1/2 points and the Niners over the Cards by seven. I think the Rams will beat the Steelers outright and the better-than-their-record Niners will blow out the Cards, by far the worst team in the league.

I am not a gambler, but doesn't the following betting strategy make sense? Only Bet When You Know More Than the Average Punter, and by that I mean 'bettor', not 'kicker'.

Our ratings don't give us a clear advantage over the average Joe in any other game, since the betting lines correspond in general to our ranking chart. So we don't bet on those games; we bet on teams the charts show as big favorites over their opponents. We're figuring that not only will we not need the points they're spotting us on the Rams, but that the Rams will stomp the very bad Steelers even though the Steelers are at home, and that the Niners are going to blow out the Cardinals by more than a touchdown. There'll probably be more Niners fans than Cardinals fans at that game. The crowd-noise factor and home-field advantage will work against Arizona. Those both seem like pretty safe bets to me, especially the Rams-Steelers game.

NOTE: I am not risking any real money on this, of course, nor do I recommend that you risk yours, unless it's some small-time harmless office pool. If you do so and you lose, it's your own fault. If you do so and you win, send me a check.

UPDATE: Football Outsiders is trying to pick up the no-TMQ slack. Check them out, and check out their Bill James-style system for analyzing teams' and players' performance. Observe their rankings and see how they compare with ours; then go over to Sports Illustrated and see how the Outsiders' take (complicated, based on stats) differs from our take (simple, based on stats) and from CNN-SI's Peter King's (based on his own feelings).

No comments: