Predicting 2022's National Champions Using Historical Kenpom Data

Predicting 2022's National Champions Using Historical Kenpom Data

February 14, 2022 (Updated 3/7 at the end of the article)

Chase Mikkelson

So how do we use Kenpom data to predict national champions BEFORE the tournament starts? Well this is why I made the graph above. Contrary to what those two articles I linked said, the data shows that the restrictions on Kenpom DER and OER are a lot wider from looking at the past 20 champions. I'm just going to go ahead and say this now, 2014 UCONN is going to be considered an outlier (I'm sorry Shabazz Napier) since their OER prior to the tournament starting was 57th, way outside of any other champions' OER. So with that out of the way, we have 19 past champions with 19 different combined OERs and DERs.

The range I found to be true, shown by the graph, is in order to be in competition for a national championship, you have to have a OER (prior to the tournament) of 21st or better, and a DER (prior to the tournament) of 43rd or better. With the margin I found, the defensive rank is a lot more forgiving than the previously mentioned top 20 in the articles I linked. Even though it lengthens the list of teams that could contend for a national title, it still keeps it at a relatively low 10-12 teams that you should pencil in as your champion when filling out your bracket.

I've seen articles, such as here and here that say different Kenpom OERs and DERs predict national champions, however both of those articles share the same opinion: Teams that win the national title HAVE to have a top 20 DER in order to win the national championship. When the holy "top 20" DER is tossed around it's because the team had a top 20 DER AFTER the tournament ended, when the final season rankings come out.


That doesn't really help when it comes to filling out brackets, however, since of course teams that win the tournament are going to have top 20 DER, along with a highly ranked OER, they just won 7 games in a row by beating the best competition.

So, what are the 10-12 teams that SHOULD, according to my research, cut down the nets on April 4? I'm writing this on February 14, and like I said previously we're still about a month out from the tournament starting, so as of February 13th (when I pulled the rankings), here are the 11 teams that I think could win the title:

11 Teams, including the top 9 Kenpom ranked teams (as of 2/13) should win the title. Illinois snuck in as of 2/13, but when I'm writing this on 2/14 they are actually out, so Illini fans consider yourselves lucky this time. My fellow Jayhawks are also barely in there, with a 39th ranked DER that slimly fits into the top 43 or better that I established previously. New 3/7 chart here

There are some surprise teams missing, such as Texas Tech, who I think is a great team that could make a deep run come March, along with Purdue, who have been pretty hot and cold this season (same with Tech).

As a KU student who has watched practically every game this season, I personally don't think we have what it takes to win the title this year, but the numbers don't lie. We'll see if the Jayhawks' DER dips below the top 43 mark and kicks them out of this table.

11 teams can seem like a lot, and it is, but I'm only using 2 metrics to measure title hopes, so I think 11 is a good number considering that fact. If you want to use this article come March, go ahead, but it's not the end-all-be-all for tournament predictions, because we know how crazy March Madness can be.

I'm going to update this article probably every week, after the Saturday games (hopefully) until tournament time to change the title contenders as the Kenpom ranks ebb and flow.

(Last Updated 3/7) (Tweet about this graph and chart)

data taken from kenpom.com, coded with hoopR

By Chase Mikkelson | twitter | linkedin | youtube