It’s been recently theorized that the strength of schedule used by most media outlets is inaccurate because it uses the previous season’s record and NFL teams tend to change over the course of an offseason. So what should we use to determine strength of schedule? Why Vegas of course.
It actually makes perfect sense. Vegas bookmakers make their living on accurately predicting the number of regular season wins for each team. Or at least setting the number and the vig so that they make the most money off of it. But in general, they’re pretty good at picking that number.
There’s two things we have to examine: 1) is using the number wins set by Vegas more accurate than using previous year wins; 2) if so, what actually are the rankings for strength of schedule for 2010?
Is it more accurate?
In a word: yes. I looked at the predicted wins from Vegas both from the 2008 and 2009 seasons and compared them against actual number of wins. After all, even though we’re talking strength of schedule, the important part of the equation here is how accurately the number Vegas pegs is the final number of wins. I also looked at the previous seasons number of wins as a predictor (PYP).
Take the 32 teams, compare the variances between predicted and actual and then see who does better. For example, in 2009, Vegas predicted 9 wins for the New Orleans Saints. Going by 2008 results, the Saints victories used in strength of schedule calculations was 8. The Saints won 13 games, giving Vegas a variance of 4 and PYP a variance of 5. I did this for all 32 teams.
No matter how you look at it, whether you take average variance or the standard deviation of the variance, Vegas does about 20-25% better than simply looking at the previous season numbers. That’s good enough for me.
Adjusted Strength of Schedule
Teddy Covers (I’m going to guess that’s not his birth name), the bookie in the ESPN article, uses expected wins (a complicated probability formula used by the Vegas guys that results in things like Dallas winning 10.1 games.) to determine his SOS. Unfortunately, in order to do that, you’ll have to figure out things such as implied win probabilities for both the over and under, cumulative binomial probabilities, sums of squared errors, etc. — basically everything that takes the fun out of football.
I’ve decided to update his numbers (they were done before the preseason) by using the regular season wins and the juice from Betus.com and set the number of expected wins based on the latest NFL odds. For example, the number of wins for the Carolina Panthers is 7.5 but the under is a huge favorite at -200. Clearly, Vegas expects the Panthers to win under 7.5 so the adjustment is necessary. In order not to over complicate things and possibly destroy my feeble brain, I’ve simply added or subtracted 1/2 a win if the juice is above -150 for the over or under. In the above example, the Panthers expected wins is adjusted to 7. On the other hand, a team such as the Bengals, with a -115 for both the over and under just stays at 8.5 wins. (If you’re a Mathlete and want to send me properly calculated zero-vig win totals, by all means, email me.)
Ok, now that we’ve gotten the boring math out of the way, here’s the latest and greatest Strength of Schedule for 2010. Feel free to use it to whine about how hard your team has it this year.
| Rank | Adj Win % | TEAM | Std Rank* | Std Win% |
| 1 | 0.541 | Philadelphia Eagles | (9) | 0.520 |
| 2 | 0.533 | Houston Texans | (1) | 0.547 |
| 3 | 0.527 | Tennessee Titans | (2) | 0.547 |
| 4 | 0.525 | Dallas Cowboys | (3) | 0.543 |
| 5 | 0.521 | Cincinnati Bengals | (4) | 0.539 |
| 6 | 0.521 | New York Giants | (7) | 0.527 |
| 7 | 0.521 | Buffalo Bills | (16) | 0.500 |
| 8 | 0.518 | Washington Redskins | (8) | 0.523 |
| 9 | 0.516 | New England Patriots | (6) | 0.531 |
| 10 | 0.516 | Cleveland Browns | (10) | 0.516 |
| 11 | 0.516 | Detroit Lions | (13) | 0.508 |
| 12 | 0.516 | Chicago Bears | (14) | 0.504 |
| 13 | 0.516 | Minnesota Vikings | (15) | 0.504 |
| 14 | 0.512 | Indianapolis Colts | (11) | 0.516 |
| 15 | 0.512 | Green Bay Packers | (22) | 0.488 |
| 16 | 0.510 | Jacksonville Jaguars | (5) | 0.535 |
| 17 | 0.502 | Denver Broncos | (24) | 0.484 |
| 18 | 0.498 | Miami Dolphins | (17) | 0.500 |
| 19 | 0.496 | Oakland Raiders | (19) | 0.500 |
| 20 | 0.496 | Pittsburgh Steelers | (21) | 0.492 |
| 21 | 0.496 | Carolina Panthers | (26) | 0.477 |
| 22 | 0.494 | New York Jets | (18) | 0.500 |
| 23 | 0.494 | Tampa Bay Buccaneers | (25) | 0.480 |
| 24 | 0.488 | Baltimore Ravens | (12) | 0.508 |
| 25 | 0.488 | Atlanta Falcons | (20) | 0.496 |
| 26 | 0.482 | New Orleans Saints | (27) | 0.469 |
| 27 | 0.482 | Arizona Cardinals | (32) | 0.445 |
| 28 | 0.479 | St. Louis Rams | (31) | 0.449 |
| 29 | 0.473 | Seattle Seahawks | (30) | 0.453 |
| 30 | 0.471 | Kansas City Chiefs | (23) | 0.488 |
| 31 | 0.469 | San Diego Chargers | (29) | 0.453 |
| 32 | 0.465 | San Francisco 49ers | (28) | 0.457 |
*Standard Rank and Standard win % based on traditional method of using previous season’s results to calculate SOS.
**Cumulative win % skews slightly higher than 16 games. 16.1 to be exact. I considered normalizing to 16 but decided .1 wasn’t going to kill you or me.
Bonus: Toughest Division in Football based on expected wins by each team
1. NFC East (.555)
2. AFC South (.531)
t3. AFC East (.508)
t3. AFC North (.508)
t3. NFC North (.508)
6. NFC South (.500)
7. AFC West (.477)
8. NFC West (.438)
