Has the NCAA’s New NET Rankings System Impacted Sports Bettors?
When the NCAA retired the old RPI system in favor of the NCAA Evaluation Tool, or NET, it was welcomed as a step in the right direction. The RPI didn’t take into account margin of victory and treated a one-point win the same as a 40-point blowout victory. It also didn’t take into account game location and simple team stats such as offensive and defensive efficiency. NET was supposed to fix all of those problems and be the key metric that the NCAA Selection Committee relied on come tournament time.
However, when the first version of the NET rankings was released in late November, there were plenty of skeptics including Nate Silver from 538. In these initial rankings, Ohio State was tabbed as the #1 team in the nation and there were other headscratchers such as Loyola Marymount at #10 and Radford at #22. But in the weeks following, sample size took over and the NET rankings started to take shape. They began to make much more sense and intuitively felt like a good representation of what had taken place on the floor. It turns out that the NET rankings are highly correlated to the ratings at KenPom.com – the gold standard in college basketball circles among media, fans and, most importantly, bettors. Kudos to the NCAA for finally embracing analytics in an effort to get the best field possible on the court come March Madness.
As a sports bettor, I was excited that the NCAA was going in a new direction this season. Change often creates opportunities for savvy bettors who do their homework. With margin of victory entering the equation in regards to tournament berths and seeding, there was a chance that the mentality could change for head coaches of certain teams this season. Instead of emptying the bench in a 20+ point blowout, would they think twice about it now? Would the impact be even greater for teams that were on the bubble? Would the margins of victory increase significantly in matchups that were easy to exploit? These were some of the questions that entered my mind before the season began.
Now that we’re more than two months in the season, we have a big enough sample of games to evaluate how NET has impacted teams and, more importantly, bettors. With margin of victory being at the crux of this theory, let’s first see how average margin of victory stacks up this season versus year’s past for all lined games.
Average Margin of Victory
2015-16 Season: 11.65 points
2016-17 Season: 11.77 points
2017-18 Season: 11.80 points
2018-19 Season: 12.71 points
Margin of victory is up nearly a full point this season over what it has been in year’s past. That may not sound like a lot, but a full point is enormous in the world of data testing. However, some things are not always as they appear. Most of this difference is due to more mismatches scheduled in the 2018-19 season than in prior seasons. The average line in games this season is 8.9 points versus an average of 7.8 over the last three seasons. As a result, most of this difference can be attributed to scheduling more than anything else. It’s possible that head coaches are softening their schedules in an attempt to game the margin of victory metric used in NET. I think it’s too soon to tell and remember that many of these games were negotiated more than a year in advance.
Next, let’s take a look at things from a betting perspective. Betting big favorites seemed like it could be a viable strategy under the new NET system. However, the betting results definitely don’t provide any evidence of that. Favorites of 20 or more are 113-131 ATS this season for -31.1 units against closing numbers. If we lower the cut off to favorites of 15 points or more, the results don’t get any better. Those big favorites are a subpar 242-257 for -40.7 units. The results are a bit better ATS when you look at opening lines rather than using closers, but they are still below 50% ATS. It appears that linesmakers have done a pretty good job making the number on games with bigger spreads.
But these results don’t necessarily mean that coaches and teams are disregarding NET entirely. Logically, the only time of the game where coaches might consider margin of victory is at some point during the second half when the game is well in hand. Consequently, it doesn’t make a ton of sense to only review the entire game from a point spread perspective. As bettors, we should focus more on the period of time where the NET could play a role in the coach’s decision-making – the second half. Let’s see how those results look in games where this could be a factor.
I eliminated all games where the margin at halftime was less than 10 points. Those games would be far too close for any coach to start thinking about margins of victory in terms of padding their NET ratings. They are much more concerned about simply winning the game at that point. So let’s take a look at the second half results for games in which the margin was 10+ points at halftime. Teams that were leading by 10 or more points at halftime have recorded a 361-319 ATS record (53.1%) this season for +10.1 units of profit. Wow, now we are onto something – or are we?
I decided to increase the threshold to halftime leads of 15+ points, which reduced the overall sample substantially. My theory was that many of the games with halftime leads of 10-14 points would end up being closer games in the second half, and that coaches wouldn’t be worrying much about winning by margin. They’d be focused on just winning the game period. It doesn’t take much for a 10-point halftime lead to quickly evaporate. The results of this smaller sample were somewhat frustrating. When winning by 15+ at the half, the team leading is just 154-153 ATS in the second half this season. When those teams were favored in the second half, the results are even worse for the team in the lead at 82-94 ATS. That means that teams that were underdogs in the second half went 72-59 ATS (55%). In many of those games, the team leading big at the half was an underdog for the full game. This was a significant blow to my theories.
My limited testing reveals that teams with big leads in the second half are probably not playing any more aggressively to build their lead and take advantage of the new NET system. We could obviously take things a step further and look into specific play-by-play data late in the second half to see how teams approach the final 5-10 minutes of the game. That would lead to a more accurate test so we can fully take into account score and situation at that time. But that’s a rigorous process and I’m fairly confident that it wouldn’t yield anything meaningful based on the overall second half results. But remember that this is only year one of the NET system. I will continue to monitor these results because I do think that some coaches will look to exploit it once they understand it better. We also may be better served drilling down into individual head coaches rather than looking at this globally. The rub there is smaller sample sizes and less overall confidence in the findings. I’ll be back with an update at some point in the future.
Eric Waz is a professional sports bettor living in Las Vegas. You can find him on Twitter at twitter.com/WazBettorIQ.