3/12 Metrics

[businesslawyer]

If you just have one, yes that matters. But a team with lots of losses to good teams isn't showing me anything. You don't get a pass for that. Especially when the measure of how "good" those teams are (in the case of the SEC this year) is inflated by some strong OOC play.

To @businesslawyer's valid point, I understand they have to try to be objective. But if you're really picking a bracket, I think all of us long time basketball watchers know that sometimes teams are more or less than they seem. Injuries, problems that have since been fixed (like 2022-23 UConn) or a style of play/matchup situation. Were all your losses by 2-4 points? Or did you, like Georgia, get blown out regularly.

Which is why picking a bracket totally based on Ken Pom, or some other metric that weighs games in November and March equally, is not as good as picking in a manner that has a recency bias. But I don’t want the Committee to be selecting teams and seeding based on purely “what they think.” I want them to be largely objective, because madness lies in letting them act totally subjectively.

 

[businesslawyer]

Looking at ESPN's BPI, UConn is 11-6 against the top 50, ranked #15, and also 4-2 against the top 20.

Despite some rumblings about the schedule this year (further exacerbated by the Maui flameout), the fact is that UConn played 17 games against the top 50 of the BPI. That's a good schedule.

So are any of the numbers geeks here able to explain why the BPI differs so greatly from other metrics? Not asking if it’s better or worse, just how the recipe differs.

 

[nelsonmuntz]

If you just have one, yes that matters. But a team with lots of losses to good teams isn't showing me anything. You don't get a pass for that. Especially when the measure of how "good" those teams are (in the case of the SEC this year) is inflated by some strong OOC play.

To @businesslawyer's valid point, I understand they have to try to be objective. But if you're really picking a bracket, I think all of us long time basketball watchers know that sometimes teams are more or less than they seem. Injuries, problems that have since been fixed (like 2022-23 UConn) or a style of play/matchup situation. Were all your losses by 2-4 points? Or did you, like Georgia, get blown out regularly.

I have always felt that the OOC games were really important in assessing the conferences to compare teams with similar records that play in different leagues. The challenge now is that it is a really unusual year in a lot of ways, with the SEC so dominant and the ACC pretty bad. I believe this year's SEC is the strongest a conference has ever been in KenPom going into the Tournament.

I guess we will find out in a week if the SEC really is that good.

 

[RuffRuff]

I have always felt that the OOC games were really important in assessing the conferences to compare teams with similar records that play in different leagues. The challenge now is that it is a really unusual year in a lot of ways, with the SEC so dominant and the ACC pretty bad. I believe this year's SEC is the strongest a conference has ever been in KenPom going into the Tournament.

I guess we will find out in a week if the SEC really is that good.

What will be crazy is that if they get 14 teams in, they would effectively be owning 22% of the field.

 

[king conn]

So are any of the numbers geeks here able to explain why the BPI differs so greatly from other metrics? Not asking if it’s better or worse, just how the recipe differs.

The big outlier for BPI is it takes into account injuries.
Seems like it should be a no brainer to weigh this metric heavier than the others but I don't think they do. But at least it helps with McNeeley this year

 

[auror]

So are any of the numbers geeks here able to explain why the BPI differs so greatly from other metrics? Not asking if it’s better or worse, just how the recipe differs.

Supposedly BPI builds in more preseason prior than other metrics and it lasts the whole year (though does diminish over time, just not to 0), so that benefits us (and teams like Kansas).

Historically it has undervalued the MWC because of having an altitude adjustment in its metrics.

The big outlier for BPI is it takes into account injuries.
Seems like it should be a no brainer to weigh this metric heavier than the others but I don't think they do. But at least it helps with McNeeley this year

BPI doesn't account for injuries anymore in college

 

[businesslawyer]

Supposedly BPI builds in more preseason prior than other metrics and it lasts the whole year (though does diminish over time, just not to 0), so that benefits us (and teams like Kansas).

Historically it has undervalued the MWC because of having an altitude adjustment in its metrics.

BPI doesn't account for injuries anymore in college

That’s interesting. As a predictive tool, it is probably not totally irrelevant how the program has done in the recent past. As a matter of fairness, however, you can’t punish one team vis a vis another because of something that happened in a prior year with prior players.

 

[JohnFSilver]

Looking at ESPN's BPI, UConn is 11-6 against the top 50, ranked #15, and also 4-2 against the top 20.

Despite some rumblings about the schedule this year (further exacerbated by the Maui flameout), the fact is that UConn played 17 games against the top 50 of the BPI. That's a good schedule.

If we never saw the KenPom and Net ratings, does UConn look like an 8 seed?

Think about the 8 seeds they have been. The Andr Drummond team, the Rodney Purvis team. Even the Shabazz title team.

This team is better than those teams, imo. More talent than the 2014 title team (7 seed), more chemistry than that loaded 2012 team that underperformed all year.

Just food for thought.

 

[JohnFSilver]

That’s interesting. As a predictive tool, it is probably not totally irrelevant how the program has done in the recent past. As a matter of fairness, however, you can’t punish one team vis a vis another because of something that happened in a prior year with prior players.

Flip side. If everyone felt a team like UConn was a top 20 program on eye test, like no one disputes that, but metrics say 8 because of weak conference schedule…why would you put a 5 seed caliber team playing well and sandbag the 1 seed?

I dont know how this is gonna turn out. Seems the sec and big 10 super conferences have a massive advantage that even if they lose they win. You are going to get a weird 4,5,6 seeds who have some ugly records.

 

[RuffRuff]

Flip side. If everyone felt a team like UConn was a top 20 program on eye test, like no one disputes that, but metrics say 8 because of weak conference schedule…why would you put a 5 seed caliber team playing well and sandbag the 1 seed?

I dont know how this is gonna turn out. Seems the sec and big 10 super conferences have a massive advantage that even if they lose they win. You are going to get a weird 4,5,6 seeds who have some ugly records.

The strength of those leagues was built in the early season OOC games - important as a league to go out and win as many of those games as you can. What kills us is that our very best team had a pretty poor OOC showing, losing to a lower tier SEC team and a lower tier B12 team on a neutral court, with their only win against the field being in the Pitino-festival at MSG (UNM). The bottom of our league being so bad doesn't help as they really don't pull any wins to help the cause.

 

[auror]

If we never saw the KenPom and Net ratings, does UConn look like an 8 seed?

Think about the 8 seeds they have been. The Andr Drummond team, the Rodney Purvis team. Even the Shabazz title team.

This team is better than those teams, imo. More talent than the 2014 title team (7 seed), more chemistry than that loaded 2012 team that underperformed all year.

Just food for thought.

5-8 seeds this year are MUCH better than they've been in the recent past. Last year the 7 seeds were all around 10-10 (around .500) in Q1+2. This year most of them are like 12-8.

 

[HuskyHawk]

I have always felt that the OOC games were really important in assessing the conferences to compare teams with similar records that play in different leagues. The challenge now is that it is a really unusual year in a lot of ways, with the SEC so dominant and the ACC pretty bad. I believe this year's SEC is the strongest a conference has ever been in KenPom going into the Tournament.

I guess we will find out in a week if the SEC really is that good.

They won't be. The top 4 are very good. Beyond that? None of them would be rated so highly in a different conference. Put Xavier in the SEC and they do just as well as A&M, Miss State or Missouri. They are better than Arkansas, UGA or Vandy. Ole Miss might be pretty good. Butler beat Miss State on a neutral court. Kentucky is Jeckyl and Hyde, good wins and bad losses.

A few of them will advance simply because if you put 8+ mediocre teams in, somebody is going to win. Texas and OU don't belong in the field. Arkansas should be the last team in. Texas did eek out a 4 point win over Syracuse, so there's that.

 

[king conn]

BPI doesn't account for injuries anymore in college

Well that figures

The one factor that gives a more accurate and trustworthy picture and they get rid of it.

Of course it's ESPN. They probably did a deep dive and found out it helped mid majors more than their p4 clients and ditched it

 

[UCLJ0211]

UCLA and Ole Miss losses may open the door to move up in the NET with a win tonight