clock menu more-arrow no yes mobile

Filed under:

AP, KP, the NET, and trying to figure out who is actually good

Between human voters and various computer numbers, it’s getting confusing to figure out who is good and who is just a metric mirage

College of Charleston v North Carolina Photo by Grant Halverson/Getty Images

How do you even keep this straight any more? Xavier beat a good Creighton team to pick up a quad one win at home on Wednesday night. How good is Creighton? Well, that depends on who you ask. They aren’t ranked in the AP top 25, KenPom has them at 19th, Torvik at 21st, and the NET at 27th. That’s not too wide a spread. The Jays are apparently an almost top 25 team.

So how about that Villanova win? On the road at the Finneran Pavilion is a good win. Well... Nova isn’t ranked in the AP, KenPom thinks they are 68th, Torvik thinks they are 90th, and the NET has them at 92nd. Want to be even more confused? ESPN has them at 82nd. So how good is that win? Apparently no one knows.

The idea behind the NET and computer metrics is that they measure things out more fairly and less subjectively. Human eyes lead to human error lead to Dylan Sinn ranking Charleston in the top 10. The AP poll is nonsense. Unfortunately, the NET is throwing up some weird results as well. Take a look:

NET KenPom Torvik AP ESPN
Team A (15-1) 15 43 32 28 54
Team B (14-3) 19 20 19 12 21
Team C (17-1) 45 79 53 22 79
Team D (8-4) 39 76 60 ---- 40
Team E (11-6) 32 23 43 38 29

So which of those teams is the best? You win no prizes for figuring out that Team B is Xavier.

The only other ranked team there is Team C. They are 22nd in the AP poll and voters clearly think they are good. That team is 79th in both KenPom and ESPN’s BPI. So are they good? Do they merit a tournament bid? The humans think so, NET thinks they are on the back end of the bubble, and Torvik thinks they are just worse than that. How do you parse such wildly disparate numbers? Pat Kelsey would tell you his Charleston Cougars should be in the tournament. The metrics don’t really agree, but the people do. For what it is worth, their only loss is their only Q1 game and their best win is Kent State.

Team A, then, seems like a lock for the tournament. 15th in the NET, receiving votes that would put them 28th in the nation, and 32nd in Torvik. That’s all pretty good. KenPom, though, sees a team barely in the top 50 and ESPN doesn’t even think they are that good. Would you put a team that at least two metrics think is mediocre in the field? The NET would, and it would have them as a top four seed. That team has two quad one wins and 11 of their 15 in the bottom two quads. They are, Xavier in the RPI style, getting a couple good wins and then hammering bad teams. Does that make Florida Atlantic better than X? The NCAA thinks so.

Say there is one spot left. We know that 38th is a soft cut line on the NET. One of Team D and Team E must be picked. You don’t know who they are, so who do you take? NET wants you to take E, but it’s really close. Team D is 2-3 in Q1, Team E is only 1-6. Does that matter? Should it? Who knows? The metrics all prefer E by some degree, but D seems to have scheduled better, they have fewer games in the bottom two quads, and won more big games. The metrics, though, are painful. Why? Because Team D lost to Grand Canyon. The teams you are trying to decide between are Sam Houston State as D and North Carolina as E.

In that case, the NET has the two teams very close, down to the point of splitting the quads to see who won what and where. The rest of the metrics think it isn’t even close and UNC should easily go ahead of Sam Houston. Which is the better team? That depends on who you trust. Not every team is as even across the table as Xavier. NET loves FAU, ESPN hates them. ESPN has UNC in easily, Torvik would leave them out.

These are somewhat cherry picked to prove a point (mainly that having Charleston ranked is ludicrous), but it shows the difficulty in using just one system to decide who is good and who isn’t. We tend to blend Torvik and KenPom when previewing teams and games. Someone who leans more into the NET will come to the conclusion that Florida Atlantic is far better than Creighton.

So how do you keep this straight? College basketball remains a bit more art than science now. At least the art has changed into deciding which metrics measure teams better as opposed to whether some goof from Fort Wayne thinks Charleston would win the Big East. It’s improvement, but the debate will rage on.