A few years ago when the NCAA college basketball NET ratings debuted, there was a firestorm of rage. The ratings were horrible, said people on Twitter, since they didn't at all match the AP top 25 at the time, nor power ratings like Sagarin or Ken Pomeroy. They have to throw them away and start all over! Said Twitter basketball.
The thing the Twitterverse didn't understand is that the NET ratings used no priors from the previous season, just the early results of that very season. And every power rating, no matter how good, looks goofy early on. That's just how it works. When you don't discriminate by your biases of how good a team should be, it's hard—or impossible—to tell how good the teams are with only a few weeks of results. It just is. Sagarin and Pomeroy for years have used last season's data and slowly "eased" it out of the formula so their early rankings aren't goofy. People who make power ratings know this. Twitter did not (surprising both Nate Silver and Jerry Palm were two of the more clueless critics).
I'm not saying NET is a great thing, but any power rating without priors looks ridiculous early on. Which brings us to our Strength power rating result after three weeks. Two issues here are: there have only been 3 weeks of games; and, there are a lot of FBS vs. FCS games that reveal very little about either team. The latter does reveal the pecking order, with FBS teams on top, but the game itself isn't great data at this stage and leaves only 2 FBS vs. FBS games for most teams and that doesn't make for consistent results.
Yet here we go: our All-Division ratings top 10 after three weeks of play.
Michigan is our surprise, or no-so-surprise #1. They've probably moved up the most in our Bayesian power rating (the one we use until the Strength power rating is ready, usually at about 6-7 weeks). They've definitely exceeded our expectations more than any team; we started them at #43 in our pre-season projections. So far they've crushed Western Michigan 47-14, Washington 31-10, and Northern Illinois 63-10. When it's too early to tell how good the competition is, nothing beats three blowouts.
Georgia is #2, almost 13 points behind the Wolverines. The Bulldogs beat Clemson which is great as long as you know that Clemson is a good team (the power rating sort of does, putting the Tigers #20, 4th best in the ACC). They also beat UAB 56-7 and South Carolina 40-13, so they have some blowouts of their own.
Mississippi is #3 after beating Louisville 43-24, Austin Peay 54-17, and Tulane 61-21. The latter is particularly helpful as Tulane almost beat Oklahoma, who ranks #18—in essence, that last result makes Ole Miss 35 points better than the Sooners who only won by 5 points.
Washington (1-2) is next at #4. Huh? Blame Michigan, whose rating is so hight that even Washington's 31-10 loss doesn't look bad. Or blame Washington's 52-3 blowout of Arkansas State for making the Huskies look good despite a 13-7 loss to Montana. Or blame Montana for only having 2 games so far, the other a 42-7 win. Either way, all this information converges to Washington at #4, and Montana at #7.
Strength power rating top ten: September 19, 2021
# Team rec Rating Offense[rnk] Defense[rnk]
1. Michigan 3-0 98.12 75.63 [ 2] -22.49 [ 4]
2. Georgia 3-0 85.86 73.50 [ 3] -12.35 [ 12]
3. Mississippi 3-0 81.37 68.44 [ 8] -12.92 [ 11]
4. Washington 1-2 79.59 62.41 [ 27] -17.18 [ 6]
5. North Carolina 2-1 79.43 78.22 [ 1] -1.21 [ 51]
6. Virginia Tech 2-1 79.04 51.58 [ 82] -27.45 [ 1]
7. Montana 2-0 78.22 53.30 [ 70] -24.92 [ 2]
8. Texas-San Antonio 3-0 77.16 64.86 [ 16] -12.29 [ 13]
9. Virginia 2-1 73.61 67.54 [ 9] -6.07 [ 26]
10. Iowa 3-0 73.33 52.95 [ 74] -20.38 [ 5]
Looking for Alabama? The close Florida win is probably what pins them at #14, one spot below James Madison. Also Miami's close call with Appalachian State and big loss to Michigan State took the shine of the Tide's 44-13 win.
The Crimson Tide are just a few spots ahead of the 1-0 Harvard Crimson, who beat Georgetown 44-9. If your team can do that, right now you should sit at #20 based on that one result. Are they ranked too high? Time will tell with more games, but we're going to assume that yes, Harvard is not a top 25 team. But if Notre Dame had played Georgetown they might have reached the same score, and no one would question Notre Dame being at #20. Fun note: Notre Dame is actually at #101 right now, due to their overtime scare vs. 0-3 FSU, their three point home scare vs. Toledo, and 27-13 win over Purdue.
What's the takeaway? There's not much so far. Michigan might be very, very good—or they might just be good. Washington may have an excuse for their terrible play so far, but not as much as this ranking gives them. UTSA isn't a top ten team but the teams on their schedule should probably take note.
Three or four weeks from now the Strength power rating will look pretty good, and where it looks confusing, it will be telling you something you should watch out for. Right now where it's confusing? That's just because it's confused. If you didn't know anything about any of these teams, you'd have a hard time making heads or tails of the results so far, too.
Comments