I don't think it was the margin of victory as much as it was scoring 75 against OU. Our offensive efficiency is bad so a performance like that really gives it a boost. I think if we had won 75 - 70 we still would have seen a similar boost.
NET Rankings are Flawed
- Thread starter BillBrasky4Cy
- Start date
same thing though, for a resume a 65-60 win shouldn’t be worse than a 75-70 win. It’s not gambling.
That's fine. I'm just pointing out that NET doesn’t use margin of victory any more. I disagree that wins vs SOS is all that matters. There are many other performance metrics that should be considered in a power ranking. The people who follow this most closely, bettors and handicappers, have known this for decades. The RPI flat out sucks. If you want to argue otherwise that's fine, but you aren't going to convince me.
i think it makes all kinds of sense for betting and coaches and prognosticators.
it makes no sense for evaluating a resume of games that have already occurred. Especially with how stupid different these games are officiated in different conferences and regions and the different styles of play that results in.
I never said the RPI was perfect or without flaw but it’s interesting that’s what people wanted to read.
Did I say you thought it was perfect? I must have missed that. I think it's interesting that's how you want to read my comments.
I think this is the key to what you're saying - performance metric is helpful in early part of season to evaluate with a low set of data and establishing some kind of general ranking -- then it should become less of a factor the more "connection" there is among teams and 20-plus games ... going from early speculation to concrete results.
people fall in love with predictive metrics (to an almost cult like degree at times), but unless I’m mistaken the committee has never been tasked with predicting the future.
Again anyone can correct me if I’m wrong but I thought they were to evaluate and compare resumes of games in the books.
The committee put out their first 16 seeded teams and those chosen teams fall in line with what you’re saying. Houston wasn’t even listed and they’re #4 in the NET.
Yeah that falls in line with “it’s a tool”. They traded in a flawed tool for a differently flawed tool.
I’m sure part of the reason for focus on quad wins is that they know it offsets the outliers much more than actually focusing on one team’s ranking.
Why is a team with a record of 1-7 vs Quad 1 games be rated in the AP top 25?? Answer, there are some real lazy sports writers living in Big 10 country.
Margin of victory is 100% factored in with the offensive efficiency ratings. I would argue it has more of an impact now than it did two years ago.
How exactly? Offensive efficiency is just points per possession as far as I understand it. I guess it's a little more complicated then that as they then modify that offensive efficiency by your opponents "team value index". Where does margin of victory come into play?
Wins vs SOS is pretty much all Strength of Record is, and I'd be fine if they wanted to seed strictly off a version of that. It's much better than RPI (or NET) at measuring resume. The official selection procedures say they want the "best teams" though, not best resume. I don't really agree with it, but that's where they're adding efficiency to the NET equation.
And while ranking by SoR would by my preferred approach, we'd probably still want two wins from here to be safe. 19-13 probably puts us well into the 40s.
PPP and margin of victory kind of go hand in hand. In order to put up high totals you are going to have to be efficient. Someone a lot smarter than me broke it down on titter a couple weeks back.
A rose by any other name... . .
Well a turd by any other name. ..
It's just amusing that too many people are too dense, too stupid, or too lazy to see that with a little deriving that it is the same info, breaking it out to offense and defense is just as bad if not worse.
Breaking it into 2 categories is just putting a sad and very crooked mustache on MOV.
Well a turd by any other name. ..
It's just amusing that too many people are too dense, too stupid, or too lazy to see that with a little deriving that it is the same info, breaking it out to offense and defense is just as bad if not worse.
Breaking it into 2 categories is just putting a sad and very crooked mustache on MOV.
Personally, I think wins vs SOS should be all the matters. It may not be the best prediction but I think the point of sports is to win the game. Taken to the extreme, should the 4 teams that go to the Final 4 be the teams that won their games or the teams with the best performance metric through the first 4 rounds of the tournament? The RPI can be improved but I still believe the system should be based on earning your way in by winning games, not by having better offensive/defensive efficiencies, point differentials, etc.
It’s calmed down a ton lately but 5-10 years ago I bet a small minority of fanatics would have welcomed using KenPom to select a final four.
Five factors in NET Rating:
Win %
Adjusted Win % (based on home / road / neutral)
Team Value Index (the modeled analytic metric)
Scoring Margin (capped at 10 points)
Net Efficiency (points per possession minus points allowed per possession)
Win %
Adjusted Win % (based on home / road / neutral)
Team Value Index (the modeled analytic metric)
Scoring Margin (capped at 10 points)
Net Efficiency (points per possession minus points allowed per possession)
I'd hazard that the reason we jumped so much after Saturday was less that there are major flaws with NET Ranking and more because there are a lot of closely ranked teams in the 30-50 range.
It's at large selection and seeding criteria for the tournament that arguably already has way too many teams. No one is proposing these be used to decide a champion. The biggest impact is what teams at the margins get left out. Generally the same teams are going to be at the top and the same teams are going to be at the bottom ni matter what criteria you use with a few exceptions. Using SOS is just another factor that can be determined in various ways and manipulated to benefit certain teams. Why don't you just propose straight up record as the criteria then? All at larges are filled by best records. Wins and losses are all that matters.
Last edited:
