/cdn.vox-cdn.com/uploads/chorus_image/image/61054287/usa_today_10712360.0.jpg)
The NCAA’s announcement that it would ditch the hated RPI metric and replace it with a new “NET” rating system for postseason selections was met with cheers of glee among college basketball statheads.
But how would previous seasons have been evaluated under NET? And how would that have affected the Virginia Cavaliers’ tournament positions?
Using media reports of what the new NET components will be, Bart Torvik compiled a data set of how NET would compare to RPI over the past 10 years.
For your perusal, some raw historical data for 4 of the 5 NET components (acc. to reports): https://t.co/aeu4brXTmK
— Bart T rvik (@totally_t_bomb) August 28, 2018
While not an exact reproduction of NET, this does give us a pretty close approximation to work off of for historical comparison purposes.
Here’s how the last 10 years of UVA basketball would have looked:
UVA Men’s Basketball: NET vs RPI, 2008-18
Year | Team | Uncapped Scoring Margin | Scoring Margin w/ 10-pt Cap | Raw Efficiency Margin | NET | RPI |
---|---|---|---|---|---|---|
Year | Team | Uncapped Scoring Margin | Scoring Margin w/ 10-pt Cap | Raw Efficiency Margin | NET | RPI |
2008 | Virginia | 2.3333 | 1.2667 | 0.029 | 0.5 | 0.44697 |
2009 | Virginia | -2.4643 | -2.8214 | -0.038 | 0.3571 | 0.27869 |
2010 | Virginia | 1.871 | 0.6129 | 0.031 | 0.4839 | 0.42857 |
2011 | Virginia | -0.5484 | 0.5161 | -0.003 | 0.5161 | 0.46809 |
2012 | Virginia | 9.3871 | 4.0645 | 0.152 | 0.7097 | 0.69504 |
2013 | Virginia | 8.9063 | 3.8438 | 0.143 | 0.6563 | 0.64655 |
2014 | Virginia | 10.6471 | 6.2941 | 0.176 | 0.8235 | 0.82667 |
2015 | Virginia | 14.5938 | 7.375 | 0.251 | 0.9063 | 0.90132 |
2016 | Virginia | 10.697 | 5.8182 | 0.175 | 0.7879 | 0.82707 |
2017 | Virginia | 10.9688 | 4.625 | 0.18 | 0.6875 | 0.67568 |
2018 | Virginia | 14.1515 | 7.697 | 0.231 | 0.9394 | 0.93464 |
From 2014 onward, there’s not much to complain about with where Virginia itself was rated and seeded. UVA has consistently been comfortably in the tournament and—all but once—a top-two seed.
But what about the 2012 team that snuck in as a ten-seed before getting smoked by Florida? Or the 2013 squad that missed the Big Dance entirely?
The 2012 team was 53rd in RPI on Selection Sunday; Florida was 30th. Understandably, UF ended up seeded better. However, if you re-rank the teams by this NET approximation, Virginia is ranked 55th—but Florida is 62nd. It’s much more likely the two teams would have been seeded about the same and not been slated to meet at all.
Could that nine-handed team have limped into the second round against someone else? Probably not. But would that “someone else” have had Bradley Beal, who burned UVA to the tune of 14 points and 11 boards? Definitely not.
Now to 2013. That squad—the last one before London Perrantes’ auspicious arrival—missed the tournament with a selection-day RPI ranked 76th in the country. Under NET, their ranking actually moves down two spots, to 78th. Those bad losses to ODU, Clemson, and Wake Forest were going to drag them down no matter what.
No set of seeding choices stands out as much to UVA fans, however, as the relative seeding of Virginia and Michigan State in 2014 and 2015. Would the new metric have rated each team differently enough that the only reason Virginia fans would have to hate MSU is for being the detestable, morally corrupt institution that it is?
Virginia and Michigan State: NET vs RPI, 2014-15
Year | Team | Uncapped Scoring Margin | Scoring Margin w/ 10-pt Cap | Raw Efficiency Margin | NET | RPI |
---|---|---|---|---|---|---|
Year | Team | Uncapped Scoring Margin | Scoring Margin w/ 10-pt Cap | Raw Efficiency Margin | NET | RPI |
2014 | Virginia | 10.6471 | 6.2941 | 0.176 | 0.8235 | 0.82667 |
2014 | Michigan St. | 10.6176 | 4.8529 | 0.154 | 0.7647 | 0.74074 |
2015 | Virginia | 14.5938 | 7.375 | 0.251 | 0.9063 | 0.90132 |
2015 | Michigan St. | 8.4706 | 3.8235 | 0.129 | 0.6765 | 0.65625 |
The 2014 UVA team had the eighth best adjusted RPI in the country on Selection Sunday; Michigan State had the 18th. Virginia’s run to the ACC Tournament title put momentum on its side, enough to vault the Hoos up to a #1 seed. But Michigan State—despite winning the Big Ten Tournament—slotted in as a #4 in UVA’s quadrant.
The NET method actually reflects poorly on both teams for 2014, dropping Virginia’s NET ranking to 11th. But the Spartans drop all the way to 29th. UVA might have dropped off the #1-line to a #2 or #3 seed, and Michigan State probably would have been somewhere in the 6-7-8 range. Anything under a 4-seed makes you an underdog in the Round of 32; chalk, if you’re to believe it, would have MSU getting knocked off by a higher-seeded team before hitting Virginia in the Sweet Sixteen.
2015 is where UVA would have benefited the most from the NET metric. According to RPI, UVA was #7 on Selection Sunday; MSU was 23rd. That justified a 7-seed for the Spartans and a second round matchup with Virginia.
But the NET rankings tell a dramatically different story. The Hoos stay at #7—but Michigan State drops to 65th. Had the NCAA ditched RPI and gone with the NET ratings that year, there is a decent chance the 2015 MSU team doesn’t even make the tournament. And if the Spartans did make it in, they likely would have been one of the “last four in,” a 10- or 11-seed having to play in through Dayton.
One final note. In 2018, UVA was the #1 team in the country in both RPI and NET on Selection Sunday. They were going to be a #1 seed no matter what, and probably the #1 overall no matter what. Changing from RPI to NET moves UMBC from 111th to 62nd. The Retrievers should have been something like a 14- or 15-seed, laying in wait to ruin someone else’s Las Vegas trip.
What does this all tell us? For one, seriously, Michigan State can go to hell. Two, changing the NCAA’s selection metrics to focus on things like net efficiency probably would have helped three of the early Bennett years—and maybe one more recent—end differently for UVA.
And three: basketball season can’t get here soon enough.