We relaunched our NFL Elo power ratings on Thursday with a nifty interactive that lets you track where your team stands and how likely it is to win each game. But Elo isn’t the only statistical power rating out there; in fact, our NFL previews used predictions and rankings generated by ESPN’s new Football Power Index (FPI), a system of far greater complexity. Since we plan to use both metrics in our NFL coverage throughout the season, we thought it would be interesting to look at how closely the two ratings mirror each other, as well as which teams fare better in each. (Note: To make an apples-to-apples comparison, I converted FPI to the same scale as Elo. I also updated the Elo-equivalent FPI ratings of New England and Pittsburgh to reflect Thursday night’s season-opening game.)
| ELO | FPI | |||||
|---|---|---|---|---|---|---|
| RANK | TEAM | RATING | RANK | TEAM | RATING | |
| 1 | Seattle | 1678 | 1 | Green Bay | 1653 | |
| 2 | New England | 1675 | 2 | New England | 1653 | |
| 3 | Green Bay | 1601 | 3 | Seattle | 1642 | |
| 4 | Denver | 1595 | 4 | Indianapolis | 1635 | |
| 5 | Dallas | 1590 | 5 | Denver | 1622 | |
| 6 | Indianapolis | 1581 | 6 | Dallas | 1592 | |
| 7 | Baltimore | 1577 | 7 | Philadelphia | 1564 | |
| 8 | Pittsburgh | 1552 | 8 | Baltimore | 1553 | |
| 9 | Philadelphia | 1545 | 9 | New Orleans | 1540 | |
| 10 | Arizona | 1541 | 10 | Pittsburgh | 1536 | |
| 11 | Cincinnati | 1540 | 11 | Kansas City | 1532 | |
| 12 | Kansas City | 1540 | 12 | Miami | 1530 | |
| 13 | Detroit | 1538 | 13 | Arizona | 1526 | |
| 14 | Carolina | 1536 | 14 | Cincinnati | 1525 | |
| 15 | Buffalo | 1534 | 15 | Carolina | 1515 | |
| 16 | San Francisco | 1529 | 16 | Buffalo | 1510 | |
| 17 | San Diego | 1523 | 17 | Houston | 1508 | |
| 18 | Houston | 1505 | 18 | Detroit | 1501 | |
| 19 | New Orleans | 1486 | 19 | San Diego | 1497 | |
| 20 | Miami | 1484 | 20 | N.Y. Giants | 1496 | |
| 21 | N.Y. Giants | 1480 | 21 | Atlanta | 1495 | |
| 22 | Minnesota | 1479 | 22 | St. Louis | 1484 | |
| 23 | St. Louis | 1477 | 23 | Minnesota | 1477 | |
| 24 | Atlanta | 1462 | 24 | San Francisco | 1463 | |
| 25 | N.Y. Jets | 1440 | 25 | N.Y. Jets | 1458 | |
| 26 | Cleveland | 1422 | 26 | Chicago | 1444 | |
| 27 | Chicago | 1419 | 27 | Cleveland | 1409 | |
| 28 | Oakland | 1389 | 28 | Washington | 1394 | |
| 29 | Tampa Bay | 1377 | 29 | Tampa Bay | 1358 | |
| 30 | Washington | 1371 | 30 | Jacksonville | 1351 | |
| 31 | Jacksonville | 1357 | 31 | Tennessee | 1349 | |
| 32 | Tennessee | 1339 | 32 | Oakland | 1346 | |
For the most part, Elo and FPI agree about the landscape of the league; the correlation coefficient between the two metrics is 0.94. But the fit isn’t perfect.
Elo thinks more highly of 17 teams — including the San Francisco 49ers, Oakland Raiders, Detroit Lions, Seattle Seahawks and San Diego Chargers — than FPI does. Among the teams that FPI favors, on the other hand, are the Indianapolis Colts, New Orleans Saints, Green Bay Packers, Miami Dolphins and Atlanta Falcons.
It’s tough to pinpoint exactly why the discrepancies exist in many cases (so many factors go into the FPI’s construction, in particular), but they’re worth keeping an eye on as the season progresses.
