Metrics (3/5)

[auror]

What website did you pull this screenshot from? Would love to see a bunch of metrics in one place

Use this
The committee is looking at something very similar to this.

https://www.warrennolan.com/basketball/2023/net-teamsheets-plus

But he was looking at Torvik's T-ranketology probably

T-Ranketology 2024 - Customizable College Basketball Tempo Free Stats - T-Rank

T-Rank College Basketball Ratings and Rankings. And fun.

barttorvik.com

 

[auror]

I haven’t read the entire thread so forgive me if this has already been said, but I look at Kenpom pretty often, and before this games our defensive rating was something like 21st in the country, and after yesterday’s game we moved up to 12th(!!!). That seems like an ENORMOUS movement for just 1 game, no?

The AdjD didn't go up that much, but it was a cluster of teams in a similar value, so we went up a bunch of ranks.

There are 13 teams between 93.9 and 94.9 for adjD. We're tied at 94.1 right now, but got the highest rank among the teams tied.

 

[CardiacAndre7]

The AdjD didn't go up that much, but it was a cluster of teams in a similar value, so we went up a bunch of ranks.

There are 13 teams between 93.9 and 94.9 for adjD. We're tied at 94.1 right now, but got the highest rank among the teams tied.

Ah I see. Yeah I didn’t do much research as to how it did that and the teams ranked similarly, that increase in rank just caught my eye. 10th ranked offense and 12th ranked defense, pretty cool!

 

[McLovin]

Didn’t read it wrong:

View attachment 84700

Wait, for the committee the conference tournament is a non-factor? Or a non-factor in these metrics?

 

[auror]

Didn’t read it wrong:

View attachment 84700

Wait, for the committee the conference tournament is a non-factor? Or a non-factor in these metrics?

Yep 2 important things here:
1) Conference tourneys are less important than we think, ESPECIALLY games on Saturday/Sunday. Every committee is different, but last year really didn't care about them.

Marquette on Friday would hopefully be early enough to still be fully factored in, but the committee begins its work on Wednesday(!)

2) I know we love the predictive metrics because we look awesome there (and they're better for gambling), but the committee has shown over and over that the resume metrics are more telling. Unfortunately, KPI hates us a bit and we're #18 in that, #12 in SOR. #13 in WAB. Ranking of the average puts us in the 11-13 range. On border of 3/4 like most bracketologists have us.

 

[king conn]

Didn’t read it wrong:

View attachment 84700

You made the statement that the committee bases the seeds by an average of kpi and SOR.

I said that isn't true, it is an average of that average ( called resume) and metrics. That's why on the team sheet, it lists the resume average and the metric average side by side.

You keep showing screenshots cutting off the metric average, I guess to bolster your argument you want to make. As if we can't see the team sheet ourselves.

Read this and get back to me.

Bracketology: Analyzing impact of metrics in NCAA Tournament selection, seeding — Heat Check CBB

Bracketology season is on the horizon. Which metrics warrant major attention and how does the committee evaluate the different types?

heatcheckcbb.com

 

[HuskyHawk]

I’m starting to not care. Beat PC and then we really should be a 3. I know the BET doesn’t matter much, but a win vs Marquette would seem to be pretty significant.

At some point the approach that looks at strong conferences and basically treats every game as a positive, win or lose, becomes a self fulfilling prophecy. Playing all those tough games helps your resume (like Tennessee with 9 losses) but somehow not playing tough games (Houston) doesn’t seem to hurt them. In UConn‘s case, the closeness of the losses and the margin of victories should matter. It tells you something.

 

[auror]

I’m starting to not care. Beat PC and then we really should be a 3. I know the BET doesn’t matter much, but a win vs Marquette would seem to be pretty significant.

At some point the approach that looks at strong conferences and basically treats every game as a positive, win or lose, becomes a self fulfilling prophecy. Playing all those tough games helps your resume (like Tennessee with 9 losses) but somehow not playing tough games (Houston) doesn’t seem to hurt them. In UConn‘s case, the closeness of the losses and the margin of victories should matter. It tells you something.

Marquette game could be more important because they are also in the 3 line mix. If we beat them, especially if the margin is significant, not only are we adding a Q1A win, but also potentially jumping them (won 2/3 against them). Basically, I think we beat them we get a 3 seed, if we lose to Providence or them, I think we get a 4. I'm still team "probably can't get a 2 without a lot of help."

The conference thing is overblown. Plenty of B12 schools, for example, were in position to make the tournament before conference season and now are not. Oklahoma, OK st, Texas Tech were all top 35 in KenPom before conference season, now they're not making the tournament. It wasn't self-fulfilling for them.

 

[DaveHuskiesFan]

You made the statement that the committee bases the seeds by an average of kpi and SOR.

I said that isn't true, it is an average of that average ( called resume) and metrics. That's why on the team sheet, it lists the resume average and the metric average side by side.

You keep showing screenshots cutting off the metric average, I guess to bolster your argument you want to make. As if we can't see the team sheet ourselves.

Read this and get back to me.

Bracketology: Analyzing impact of metrics in NCAA Tournament selection, seeding — Heat Check CBB

Bracketology season is on the horizon. Which metrics warrant major attention and how does the committee evaluate the different types?

heatcheckcbb.com

Never said the base it solely off that. I said last year every team was within 1 seed line of the average of KPI and SOR (which is a fact). This isn’t even an argument? I read it on twitter and was passing it along. You’re arguing with yourself.

 

[king conn]

Didn't see your Twitter post you say you read, my guess is it was closer to what I have been saying, because what I showed you was completely more accurate while claiming the same result.

So I don't know how just figuring out the resume portion would have yielded the same results, it would be impossible. That was my point which you were indeed arguing in a passive aggressive way

 

[UCFan1513]

Any bracketologist who has us as a 4 seed at this point can eat my as$

 

[ctchamps]

Marquette game could be more important because they are also in the 3 line mix. If we beat them, especially if the margin is significant, not only are we adding a Q1A win, but also potentially jumping them (won 2/3 against them). Basically, I think we beat them we get a 3 seed, if we lose to Providence or them, I think we get a 4. I'm still team "probably can't get a 2 without a lot of help."

The conference thing is overblown. Plenty of B12 schools, for example, were in position to make the tournament before conference season and now are not. Oklahoma, OK st, Texas Tech were all top 35 in KenPom before conference season, now they're not making the tournament. It wasn't self-fulfilling for them.

Overblown yes but wouldn't it still be a factor? How much was the OOC play of Creighton and Nova an impact on our SOR?

 

[auror]

Overblown yes but wouldn't it still be a factor? How much was the OOC play of Creighton and Nova an impact on our SOR?

It's 4 games out of 31. A factor? Sure. But a small one.

 

[AzHusky]

Something else to look at:
ESPN's computer model, the Basketball Power Index, has ranked the country's top 25 teams as we head into the final day of the regular season.

Here's the top 25:

1. Houston
2. Tennessee
3. Alabama
4. UCLA
5. UConn
6. Texas
7. Purdue
8. Gonzaga
9. Creighton
10. Kansas
11. Baylor
12. Arizona
13. Kentucky
14. Arkansas
15. Saint Mary's
16. Marquette
17. Maryland
18. Indiana
19. Texas A&M
20. Xavier
21. West Virginia
22. Iowa State
23. Rutgers
24. San Diego State
25. Duke

 

[UCFan1513]

Something else to look at:
ESPN's computer model, the Basketball Power Index, has ranked the country's top 25 teams as we head into the final day of the regular season.

Here's the top 25:

1. Houston
2. Tennessee
3. Alabama
4. UCLA
5. UConn
6. Texas
7. Purdue
8. Gonzaga
9. Creighton
10. Kansas
11. Baylor
12. Arizona
13. Kentucky
14. Arkansas
15. Saint Mary's
16. Marquette
17. Maryland
18. Indiana
19. Texas A&M
20. Xavier
21. West Virginia
22. Iowa State
23. Rutgers
24. San Diego State
25. Duke

It really hates Marquette

 

[businesslawyer]

It's because the algorithms for these ranking systems take into account margin of victory and margin of loss, but the selection committee doesn't take these things into account very much. UConn has won every game by at least 6 points and 23/24 games by at least 8. UConn also has 3 losses by 3 points or less.

This really shouldn’t be hard to understand. If you judge us by how Vegas would view our chances to win a game on a neutral court against an opponent, which is what all the metrics you’re using take into account, we’re a two seed. If, on the other hand, you only look at who we beat, who we lost to and where, we don’t deserve any better than a 4 seed. It is not unreasonable to not care a lot about how much you are winning and losing by, since we tell athletes that the goal is to win, not to win by as many points as possible.

 

[businesslawyer]

The Big XII was incredible in the OOC, which means that when they beat each other up in conference it "launders" everyone's resume and they end up with 4-5 teams on the top 3 seed lines. Whether they deserve it remains to be seen.

There were years where that was the old Big East. Most of the conference would win their holiday tournaments, and whatever your record was the committee viewed the losses as good losses.

 

[Iluvsuebird]

NET: 6 (moved up a spot)
KenPom: 4 (also moved up)
BPI: 5

I would be very mad as a fan of a #1 seeded team having to play Uconn as a 4 seed.

 

[aburks41]

Tennessee lost to Colorado and Vanderbilt, so I’m not sure where this “no bad losses” thing is coming from. Those are on par (probably worse) than our losses to Seton Hall and St Johns. We also have 2 fewer losses than them. I don’t see an argument for having them seeded higher than us at all

And Tenn has lost 6 of their last 10 games. They are trending in the wrong direction.

 

[ToolzieFan]

 

[Hey Adrien!]

There's a ton of parity this year, but still blows my mind that Houston and UConn are the only two teams in the nation that rank in the top-12 in both KenPom offense and defensive efficiency.

The next closest teams to that level of efficiency: Texas (16th in O, 18th in D).

 

[McLovin]

Yep 2 important things here:
1) Conference tourneys are less important than we think, ESPECIALLY games on Saturday/Sunday. Every committee is different, but last year really didn't care about them.

Marquette on Friday would hopefully be early enough to still be fully factored in, but the committee begins its work on Wednesday(!)

2) I know we love the predictive metrics because we look awesome there (and they're better for gambling), but the committee has shown over and over that the resume metrics are more telling. Unfortunately, KPI hates us a bit and we're #18 in that, #12 in SOR. #13 in WAB. Ranking of the average puts us in the 11-13 range. On border of 3/4 like most bracketologists have us.

Appreciate this. I get that a loss in the conference tournament shouldn’t really hurt a team, especially if it’s against a good opponent.

But wild that they would not put much (or at least, equal) value on a few very good wins the weekend before the actual event.

 

[UConNick]

Appreciate this. I get that a loss in the conference tournament shouldn’t really hurt a team, especially if it’s against a good opponent.

But wild that they would not put much (or at least, equal) value on a few very good wins the weekend before the actual event.

Yep, I don’t get it. Reeks of laziness to me. They want to get it done as soon as possible and don’t feel like revising or adjusting later in the week.

If it’s a time issue, then have all the conference title games finish up on Sat. so they then have all day Sunday to look at the whole national picture and finalize the bracket.

 

[tzznandrew]

Tennessee lost to Colorado and Vanderbilt, so I’m not sure where this “no bad losses” thing is coming from. Those are on par (probably worse) than our losses to Seton Hall and St Johns. We also have 2 fewer losses than them. I don’t see an argument for having them seeded higher than us at all

And Tenn has lost 6 of their last 10 games. They are trending in the wrong direction.

Colorado: 56 in Kenpom. 72 in NET. It was a Neutral game.
Vanderbilt: 82 in NET. 81 in KenPom. It was an away game. They lost by 1

Each of these has a chance of being Q1 by the time of the tournament! Colorado is 6 away from making it that, Vanderbilt 7.

By any standard those are both better losses than:

St. John's: 97 NET, 88 KenPom. Home. We lost by 11.

It is solidly Q3.

And the Seton Hall game is maybe better than the two above but it's in the same ballpark of the Vanderbilt one.

Seton Hall: 74 NET, 63 KenPom. At least this could become a Q1 loss if they win a game (or two!) in the BET.

 

[McLovin]

Yep, I don’t get it. Reeks of laziness to me. They want to get it done as soon as possible and don’t feel like revising or adjusting later in the week.

If it’s a time issue, then have all the conference title games finish up on Sat. so they then have all day Sunday to look at the whole national picture and finalize the bracket.

Agreed. Or do the job they are contracted to do and work 16 hours a day for 4 days a year. No one is putting a gun to their head to be on the committee…