The NET ratings just came out and I must say, now that they tweaked the formula a few years ago to go from an opaque, 5-component ranking to use just 2 components, it really looks like they've replicated a lot of our Combined Rating. That rating, which as the name suggests combines our Strength and Success ratings into a composite, currently uses a healthy dose of pre-season information on the Strength side but when I run it without pre-season bias, the teams match up pretty well with the NET.
What I'm saying is, clearly the NET is using a similar philosophy now: use a true power rating (like Strength) that is based on point margins, or per-possession stats rated by strength of opposition, and combine that with a points-free Win vs. Loss system that heavily uses home/away court. In their own words:
The remaining factors include the Team Value Index (TVI), which is a result-based feature that rewards teams for beating quality opponents, particularly away from home, as well as an adjusted net efficiency rating. The adjusted efficiency is a team’s net efficiency, adjusted for strength of opponent and location (home/away/neutral) across all games played.
Here's how it looks for the season so far:
NCAA NET top 25, 12/5/2022
1. Houston 2. UConn 3. Purdue 4. Tennessee 5. Mississippi St. 6. Maryland 7. Sam Houston 8. Alabama 9. Virginia 10. Indiana 11. West Virginia 12. Texas 13. Illinois 14. UCLA 15. Utah St. 16. Saint Mary's 17. Duke 18. Kansas 19. Utah 20. Iowa 21. Auburn 22. Iowa St. 23. Fla. Atlantic 24. Kent St. 25. Arkansas
You can see the influence of a Strength, or Sagarin Predictor, or probably closest, a Kenpom type of system in use, in that the teams that rate high in those rate high in NET. The difference being that Sagarin and Kenpom have pre-season bias to smooth out the early weirdness; the NET doesn't.
You can also see a heavy influence of a wins-only system, and in particular one that really gives a heavy advantage to teams that win on the road or on neutral courts. Early in the season, road wins are at a premium. 95% of early games are overmatched opponents traveling to get beaten by a much better foe. That means that the home team wins almost all the time. Not only conference play do road wins start to add up. So early on, if you have road wins, you have everything.
This is why Sam Houston State is at #7, for example. The Bearkats have four D-I wins and a loss. Two of the wins are on the road and the other two are on neutral courts. It doesn't matter all that much who they beat at this point, given the lack of pre-season bias, just getting road wins is a big deal. But they happened to be Oklahoma and Utah that they beat on the road, which by this point is very impressive. Their one loss is on the road too, which mitigates it since losing on the road is what 95% of teams are doing right now.
Here is our Combined Power Rating for 12/5/2022, without any pre-season bias:
num Team record rating NET 1. Connecticut 9-0 44.10 2 2. Purdue 8-0 40.41 3 3. Houston 8-0 40.19 1 4. Maryland 8-0 35.89 6 5. Tennessee 7-1 31.38 4 6. Utah St. 7-0 31.12 15 7. West Virginia 6-2 30.25 11 8. Mississippi St. 8-0 30.17 5 9. Virginia 7-0 30.02 9 10. Sam Houston St. 7-1 30.01 7 11. Texas 6-0 29.91 12 12. Alabama 7-1 29.81 8 13. Indiana 7-1 28.72 10 14. Auburn 8-0 27.16 21 15. UCLA 7-2 26.97 14 16. Arizona 7-1 26.37 26 17. New Mexico 7-0 26.07 27 18. Arizona St. 8-1 25.83 35 19. Illinois 6-2 25.82 13 20. Kansas 8-1 25.46 18 21. Iowa 6-1 24.82 20 22. Gonzaga 5-3 24.22 29 23. Arkansas 7-1 24.13 25 24. Kent St. 6-2 23.72 24 25. Rutgers 6-2 23.71 30
Lots of similarity there, more than you'll see with other power ratings around the web. Both for the fact that almost everyone "smooths" with pre-season bias for the first few months, and that most are single-purpose, like either Strength or Success, or Kenpom, either to predict future games OR to "reward" accomplishments, but not to do both at the same time. That's what the NET tries to do, as well as our Combined rankings.
The other similarity is early-season stress on winning road games, which occurs naturally in ours since road wins are very valuable (due to rarity) early on. This might seem overvalued at the time, but the purpose of the NET isn't to measure who is best in December but to work in mid-March.
Another team that's a good example is Florida Atlantic, #23 in NET and #27 in "unbiased" Combined ratings. The Owls have three road wins, including one at Florida, and their only loss is a road loss.
Perhaps the best example of similarity? Look at Louisville. The Cardinals are 0-8 right now but few ratings would put them in the bottom three in the nation, right? Kenpom for example has them #206.
But from a wins and losses-only perspective, Louisville is among the very worst teams. Three of their losses are by a single point—but that doesn't matter in a wins vs. losses system. What does matter is how many home losses you have. And right now Louisville has FIVE home losses! And three more neutral-court losses. That makes the Cardinals the worst team in the nation in our Success rankings. Combine that with unbiased Strength (which has them #336) and you get this:
361. Hartford 4-7 -26.07 362. Louisville 0-8 -27.63 363. Houston Christian 2-7 -30.82
Compare that to the bottom three in NET:
361 Louisville 362 Houston Christian 363 Hartford
Wow. That's some pretty close agreement. But we were here first, like over 15 years ago. Should have asked us back then and we could have replaced that stupid RPI a lot sooner.
Comments