Football Geekery

This entry will be boring to people who aren't interested in analysis of the 2002 NFL season. If you're such a person, you may want to wait 'til my next entry. Otherwise, prepare to geek out.

As I mentioned yesterday, I did good in my football pools this year. Tonight I finished rounding out my spreadsheet with all the scores and decided to do a little analysis of the NFL teams, and make some predictions for the playoffs.

First, some notes about what my spreadsheet does:

Essentially, I use what is called the "Pythagorean Method" to estimate winning percentages for each team based on their points scored and allowed. (As a detail, the exponent I used for the calculation was 2.85 during the season. It turns out scoring was up somewhat for 2002, so a 2.90 exponent might have been better. I'm not sure how much difference it would have made, though.) The simplest approach to making my picks, then would have been to compare the estimated winning percentages (EWP) and pick the team with the higher percentage. For my ranked pool, I would rank teams with a larger differential between them and their opponent with a higher ranking.

I made three adjustments to this basic approach:

1. First, I computed the average EWP of all the opponents for each team. (their OEWP) Teams which were played twice were counted twice by this method. I then adjusted the points scored and allowed for that team by the OEWP; their points scored were multiplied by (0.5 + OEWP) and their points allowed were divided by the same amount. This basically adjusted for "strength of schedule"; a team which played teams who were stronger than average would get extra credit for doing so.
2. Second, I weighted individual games so that more recent games counted more heavily. In particular, the four most recent games would be given more than one game's worth of weight, and the prior games would be given less. This was to attempt to account for teams on hot streaks or who had recently made a significant adjustment (brought in a better - or worse - quarterback, for instance).
3. Third, when projecting the winner for a match-up, I gave the home team a 6% advantage in their EWP. This "felt" like the right number, and at least one article I read suggested that it's fairly close to the real home field edge, for the typical team. (Obviously the Packers hosting the Dolphins might not measure up the same way.)

Now, on to some analysis:

Teams with the Strongest Schedules: Pittsburgh (.538), Atlanta (.537), New England (.537). The Patriots generally floundered against good teams with good running games. Both Pittsburgh and Atlanta had rocky stretches. This is probably a reason why.

Teams with the Weakest Schedules: Philadelphia (.444), New York Giants (.469), Green Bay (.471). All three of these teams are playoff-bound, and their schedule probably had something to do with it. Philly played a lot of really bad teams, and only three relative powerhouses (Tampa, Tennessee and the Giants twice). Even with Donovan McNabb, they're probably not as good a team as they appear at first glance.

Looking at these SOS numbers, I suspect that I should have changed my handling of weighting teams by strength of schedule, as there isn't really a lot of difference in my formulae between teams playing a .538 schedule and teams playing a .520 schedule. Something to consider for next year.

Teams Which Should Have Made the Playoffs but Didn't: Miami, Kansas City.

Teams Which Shouldn't Have Made the Playoffs but Did: Cleveland, San Francisco.

These are the marginal playoff teams, all of them basically flawed in some way. What's interesting is that my system thinks that the AFC had several teams better than the NFC West champion (Denver was another). Cleveland doesn't really have any business being in; Miami would have been a better choice.

Four Really Bad Teams: Arizona, Cincinnati, Houston, Detroit. No surprises here; my system picks Houston as the worst team in the NFL. Detroit is actually the best of these four, which isn't a huge surprise, either. Detroit had a few capable players during the season; Arizona and Cinci are moribund, and Houston is obviously still getting off the ground. I'd expect Houston to zip past the other three next year, assuming they can assemble an offensive line.

Underrated Team: Jacksonville. They actually outscored their opposition 328-315, against a nearly-neutral schedule. Unfortunately they went 6-10. It's easy to say that they pounded on the bad teams and got pounded on by the good teams, but they actually beat KC and Philly early in the season, and made a game of it against several playoff teams (they lost to Indy twice, the Giants, Pittsburgh and Cleveland each by 7 points or less). They were somewhat unlucky. They're not a team on the way up, though.

Overrated Team: San Diego. The Chargers seemed to be considered playoff contenders for much of the season, thanks to 6-1 start, but their 9-7 record seems a little lucky. They beat some good teams in close games, but also got beaten convincingly by Denver, the Jets, Miami, and Oakland. Their defense was good enough to make a game of it most of the time, which counts for something, but they probably should have finished 7-9, and would have if they hadn't eked out wins over KC and SF. However, they will probably get better over the next two years.

Now for my predictions for the playoffs:

Wild Card Games

Pittsburgh over Cleveland: Pittsburgh does have a wacky quarterback situation, but overall they're a much better team all around than the Browns, especially since Tim Couch is injured. Even with Couch, the Steelers should make short work of the Browns.

Atlanta over Green Bay: Even in Lambeau? Yes, even then, but it's a fairly close call. The Packers limped into the playoffs and may face the Falcons without Donald Driver. Unless Favre brings his A game, the Pack will be at home the following week.

Giants over San Francisco: The closest of the picks. Both teams have good-but-not-great offenses and defenses. The Giants finished stronger, but the 49ers rested their regulars the last two weeks. A really tough call.

Conference Semifinals

Tennessee over Pittsburgh: Could be a close game. Tennessee has been very strong in the second half of the season, though, and as long as McNair doesn't collapse in a pile of bones and painkillers, they ought to prevail.

Tampa Bay over Atlanta: Not really a contest. Tampa beat the Falcons convincingly twice during the season. The Falcons' only hope is that Tampa can't get an offense going at all due to their quarterback injuries.

Philadelphia over Giants: The Giants won the final game of the season 10-7, but lost earlier in the season. If Donovan McNabb returns and is healthy, then Philly has a much better chance. As it is, we could end up looking at another 10-7 score.

Conference Championships

Tampa Bay over Philadelphia: A battle of teams with great defenses and injured quarterbacks. Could be close, 6-3 close.

Super Bowl

Overall, this looks like the Raiders' year. Their defense can be exploited, but it's not bad. Their offense is tremendous, and they have fewer question marks of the "our quarterback is hurt" variety than the other teams. I think the game to watch is the Oakland/Tennessee game, assuming it happens. Unless Philly or Tampa get their regular quarterbacks back, I don't think they'll be able to make the Super Bowl a real dogfight.

But hey, I've been wrong before!