I think some of our previous opponents losing is also dragging us down. Everytime CU loses it is a double hit to our resume. Eventually it will be a triple hit so I would appreciate it if they would stop shitting the bed.
2024-2025 MBB computer projections thread
- Thread starter Sigmapolis
- Start date
Really sucks that CU somehow beat UConn, because we'd have killed UConn, the way they were playing, then, and it would be a great boost to our NET, right now.
It's just important to remember the difference between predictive ratings (Kenpom, Barttorvik, and basically what NET appears to be) vs. descriptive (strength of record, wins above bubble).
Would Iowa State be a significantly better team if Broome had missed the putback in Maui and Iowa State had made one more free throw to win by one? Or would Iowa State be a materially worse team if Jefferson had missed his shot in Lubbock to send it to OT, or Williams had made his last shot? I think the answer to both of those is no. The W/L matters and should matter in terms of ranking, seeding, etc. But in a sport with a lot of randomness, it doesn't necessarily tell the full story about how good a team actually is and what it will do in the future.
Houston is kind of a more extreme version of Iowa State last year--no Q1 wins going into conference play, some difficult losses, but a lot of big, efficient wins that led to better predictive metrics than most would think based on the W/L results. Those obviously proved out to be pretty much correct for Iowa State last season. We'll see if the same happens for Houston this year.
Wait for it to be quadruple when we bump into them in KC.
#1 should have #8 and then #9 and #16 in its "pod," right?
Colorado could very well be #16 and pull two upsets.
Just can't get rid of them this year.
Looking back, I think it's not surprising UConn lost all their games at Maui. I mean who in their right mind would want to take a trip to Hawaii with Dan Hurley?
meh, give me the wins we have and being 15-1 with a NET of 5.
Yes, the winner of the 9-16 game faces 8 for the chance to play 1.Wait for it to be quadruple when we bump into them in KC.
#1 should have #8 and then #9 and #16 in its "pod," right?
Colorado could very well be #16 and pull two upsets.
Just can't get rid of them this year.
I'm somewhat guessing but I think the apparent contradiction is due to tempo. Tempo as a metric is offense-oriented -- it measures offensive possessions, not defensive possessions (ref: https://kenpom.com/blog/help-with-team-page/).
Therefore, if a low-tempo team (Houston) meets a high-tempo team (e.g. Alabama, a team they lost to), the defensive efficiency (which Houston excels at) can still be maintained even if they lose because they are defending a higher-tempo team. If your defense forces an offense to shoot 20% but they still get their normal number of possessions, you're still defensively efficient.
The key thing is that Alabama's reduced offensive efficiency w/high tempo may still out-score Houston's less reduced offensive efficiency w/low tempo.
Put pseudo-mathematically:
Code:
ORtg(Houston)*AdjT(Houston) - DRtg(Alabama) < ORtg(Alabama)*AdjT(Alabama) - DRtg(Houston)
That is a good insight. Might explain why when you look at past champions, offense ratings appear to be more important than defense ratings.
I don't think this is right. Tempo is an overall metric, not something attributed to just offense or defense, though it's easier for the offense to impact it. Possessions are just calculated using offensive stats because that's what's available in a box score. The number of defensive possessions is just the opponent's number of offensive possessions. As noted in the link you shared, tempo takes an average of the team's possessions and the opponent's possessions.
I think the thing that confuses people about NET is you're hearing about Quad 1 wins all the time, and it's NET that defines those quadrants, but I don't think "Number of Quad 1 wins" factors directly into the NET score. Kind of makes sense, you need to know the score before you can know if a win/loss is in a certain quadrant or not.
You guys are overthinking this. Houston is still very high in the metrics, because they lost 3 very close games and otherwise have been destroying teams. From a predictive metrics standpoint and not from a W-L standpoint, it's important to remember that a loss by 1 and a win by 1 are nearly the same thing. Obviously, as a fan, that's not the case as those two things couldn't be more different, but as far as your efficiency profile, both would affect them about the same. Tempo has nothing to do with it.
Here are Houston's win probability charts in their 3 losses. It's extremely bad luck that they lost all 3 of those games (which is why they rank 336 out 364 in Kenpom's luck metric).
And here's their statistical profile game-by-game. Compare their points per possession on offense to what they give up on defense. Other than their 3 close losses, there's not a single close game in there. If you score 1.2-1.3 PPP in a game and give up 0.7-0.8 PPP on defense, you're going to be in the top 10 of every single predictive metric (Kenpom, Torvik, BPI, NET).
It's not Houston's 3 losses that I get hung up on. It's that they lost to San Diego State on a neutral floor. I get that they blow out bad teams, and their 3 losses are all close. I think their ranking is skewed because of those blow outs, and they aren't as good as the computers think they are.
Those aren't road losses for Houston. They were two neutral site games, and a pseudo home game against Auburn. I don't care that they beat Hofstra by 36.
Those aren't road losses for Houston. They were two neutral site games, and a pseudo home game against Auburn. I don't care that they beat Hofstra by 36.
The good news is that ultimately none of this really matters until march. Houston will be extremely well tested by the 11 Q1 opportunities (7 Q1a!) they have over the next 2 months, and either they'll succeed and verify their current metrics or they won't and their metrics will drop to reflect it.
This is correct. However one thing that has come up with tempo in the past is defense-minded, slow paced teams often win blowouts by a higher percentage of points per possession.
You were more likely to see 60-40 scores versus 90-60 scores. Often after a 20 point lead teams call off the dogs and it ends up 90-70. However, TJ and other coaches have realized this and you are seeing less subbing even when up by 20.
Yeah, on some level looking too deeply into metrics right now is like watching the sausage while its being made.
It'll come together and make sense in march. Either teams will have a resume or they won't.
ISU was tied with currently #58 in Kenpom Dayton on a neutral floor with 1-minute remaining and won.
Houston was up 2 on currently #34 in Kenpom San Diego State on a neutral floor with 1-minute remaining and lost.
ISU won their close game. Houston lost theirs. Now ultimately if you wanted to say ISU is more likely to win a close game due to having better guard play leading to a higher likelihood to get a good shot and not having to rely on grabbing an offensive rebound to score on a key possession than Houston, that makes sense and I'd agree. But as far as how advanced metrics view those games, I don't know how you can differentiate them that much.
I'm sure others are curious, our luck rating is .002 (#187/364). Basically, we have been equally lucky/unlucky in the win/loss column. We could have easily seen a flip in the result from TT and Auburnfor example
