The NET ranking is definitely an improvement in a computer ranking system – I am glad that the Selection Committee will be using more analytics than the very unsophisticated RPI ranking.  If you want proof of this and the fatal flaw of the RPI, you can look at the top.  As of last night’s games, the #1 team in the NET ranking was Virginia (who has only lost two games this season – both to Duke).  The #1 team according to the RPI is Kansas at 23-8.  To be fair, the Jayhawks are not a bad team, but #1 in the country is a stretch.  We could all argue if Virginia, Gonzaga, Duke, UNC (or a couple others) are the best team in the country – I think even Jayhawks’ fans would be willing to admit that all of those teams currently are better than Kansas based on their records.

So, why are the Jayhawks so high in the RPI – because of its fatal flaw.  The RPI is heavily weighted towards strength of schedule.  While I could argue that some of those teams are better than Kansas – it would be hard to find a tougher schedule  Kansas played at Kentucky, at Arizona State, vs Tennessee, Michigan State and Marquette on a neutral court, home against Wofford and Villanova – and that is just their non-conference games.  It doesn’t include their 10 Quadrant 1 games from the Big 12 regular season.  They only played 2 games against Quadrant 4 teams and 4 against Quadrant 3 teams.  Just to continue on that point.  Even Duke (who seems to always play a tough non-conference schedule), still had 5 victories against the Quadrant 4 teams.  The Jayhawks schedule this year is just brutal.

But playing a brutal schedule does not mean you should be the #1 team – maybe you should be praised for your courage (or criticized for being insanely ambitious – there is probably a fine line here although Kansas is talented enough each year to take on this challenge and survive it), but you still have to beat almost everyone.

North Carolina State is the example on the opposite side – they currently have a RPI ranking of 104 – they have 10 games against Quadrant 4 teams, and 5 more against Quadrant 3 (they even lost 2 of those games to Georgia Tech and Wake Forest).  But at the end of the day, the Wolfpack went 9-9 in the ACC (10-9 if you include today’s crazy 18 point comeback against Clemson – they were down 18 and managed to hit two free throws at the end of the game to win by 1) – any team that can win more often than lose in the ACC is better than 104th.

But that also brings out the flaw of the NET ranking.  They probably aren’t the 32nd best team either.  While they went 10-9 in the ACC, they went 0-7 against Virginia, Duke, UNC, Virginia Tech, Louisville and Florida State.  Basically, the Wolfpack won the games they were supposed to – and that led them to a pretty decent record.  We could debate things, but I can imagine we can find some teams that have a worse NET ranking that might deserve to be ahead of the Wolfpack.

The problem here is that the NET ranking looks at more predictive metrics such as scoring margin.  And no matter how you try to cap it, it is hard.  When you beat multiple bad teams by 50 points, it still looks good.   And when you play a team like Virginia or Duke, you can lose by 10 points – and compared to the rest of the country statistically – that looks like a solid performance.

Another example here is Penn State.  Despite a 14-17 record, they are 49th in the NET ranking.  If you look closer at their schedule, they played 15 games against Quadrant 1 teams – but they won 3 of them and lost 7 of those by single digits – so they were competitive in 10 of those games.   If you play at team like Michigan that on average wins by 12 points, and you beat Michigan by 6 points – your adjusted possession statistics go through the roof.

And that becomes the complication with all these ratings.and why the Selection Committee is so important.  There are 353 teams in Division I basketball – and obviously there is a big difference between playing Virginia and Chicago State.  But teams eventually play the schedule that they have (and the games against their conference are forced) – so how do you compare a 30 point victory to Chicago State to a 6 point loss to Virginia – what outcome is better.  If you can solve this problem, there are probably lots of media outlets (and maybe some Vegas books) that would like to pay you some money.

That being said, I am still glad that the committee moved to the NET ranking.  It might still be tricked by teams that have great adjusted scoring margins – but I think it still gives a better ranking system.  Just realize it still isn’t perfect.  It is simply a great start.