Does The Big East Hurt Our KenPom?

[Lafayette]

Who is Ken Pomeroy and what does he have to do with our team, who we play, or really anything affecting the outcome of a game?

 

[RuffRuff]

No. Not being dominant hurts our rating. We are winning close games, which is preferable to losing, but we are not being consistently dominant. Games that should be easy wins are turning into stressful events. On the flip side, a few games that should have been losses turned into wins. Does that mean we tend to play down to our competition? Maybe. But we also tend to play up to our competition and beat the best teams that we play, even if our rating says we shouldn't.

Obviously it would be best to win close games that we shouldn't win AND blow out teams that we should beat. But, if that is not our reality, is it better to play down to weaker teams and win close games that we shouldn't win or blow out weaker teams and lose close games that we aren't supposed to win? I don't know. 2022/2023 was kinda like the latter and this team is the former. I guess we will see how this works out.

Feels like we cannot sustain a full 40 right now. The bench is also getting tighter it seems, so our depth isn’t quite as relevant as it was. I’ll be curious to whether this continues. I feel like Reibe, Malachi and Ross’ leashes have been very short as season has played out.

 

[KingAG]

Feels like we cannot sustain a full 40 right now. The bench is also getting tighter it seems, so our depth isn’t quite as relevant as it was. I’ll be curious to whether this continues. I feel like Reibe, Malachi and Ross’ leashes have been very short as season has played out.

Who in the country can?

 

[Woof]

I don’t care about KenPom so much, just asked the question more as a thought experiment. It’s wild to me that we can beat a top 25 team on the road and drop in the metrics…that just shouldn’t ever really happen. Top 25 road wins are very hard to come by.

It's fine. You are curious and that is good. Ignore the people who hate things they can't understand.

 

[boog204]

I pity the fans who watch a win and go to look at metrics. It’s ruined your fan experience

10000000000%

We're an elite team. We are really good. We are 17-1. It is setting itself up to be a magical season. Stressing over the metrics is robbing you of the joy of probably our only year of Mullins, the last year of Reed, and most importantly, the senior year of the greatest winner in the history of the program. Sit back, crack a Rolling Rock (do they still exist?), eat some pizza, yell inane things at the TV when our "freedom of movement" is impaired, and enjoy the unbelievable fortune and awesomeness of being a UConn fan.

 

[Viet Vet]

Al Davis, "Just win, baby!"

 

[auror]

Who is Ken Pomeroy and what does he have to do with our team, who we play, or really anything affecting the outcome of a game?

He's a retired weatherman.

 

[Inyatkin]

Feels like we cannot sustain a full 40 right now.

Even if you accept that premise, I don't know who in the country is playing a full 40. Duke eking out a Florida State win, Michigan escaping Penn State, Arizona in some struggles. Teams rarely blow everyone out

 

[boog204]

Even if you accept that premise, I don't know who in the country is playing a full 40. Duke eking out a Florida State win, Michigan escaping Penn State, Arizona in some struggles. Teams rarely blow everyone out

^This. I feel like our fanbase forgets that the 23 tourney run or the majority of 24 were an anomaly. Very, very few college basketball teams put together 40 consistent minutes. Very few teams blow people out; the parity is better than it has ever been, especially with the portal and opposing teams having 37-year-olds. Most teams will have those rock-fight games and win.

 

[Inyatkin]

People really should ignore you two.

Yep, every game has played out exactly as one should have expected. Not a point better or worse. You knew we would win every game we have won. Please post a picture of the Lambo that you bought with your winnings in Vegas.

If you're having a hard time finding the ignore function I'm sure you can get someone to help you out

 

[RuffRuff]

Even if you accept that premise, I don't know who in the country is playing a full 40. Duke eking out a Florida State win, Michigan escaping Penn State, Arizona in some struggles. Teams rarely blow everyone out

True - simply an element of the metrics that is weighing us down a bit that should work itself out as we do stretch out longer stretches of playing high caliber. I'm not super worried in games we find ways to get up big, then relent a bit.

 

[businesslawyer]

I don’t care about KenPom so much, just asked the question more as a thought experiment. It’s wild to me that we can beat a top 25 team on the road and drop in the metrics…that just shouldn’t ever really happen. Top 25 road wins are very hard to come by.

In a resume metric, it doesn’t ever happen. In a “pure” performance metric, you will drop every time you win by less than the metric predicted and go up every time you win by more than the metric predicted.

But you have put your thumb directly on why performance metrics are only useful as sorting tools and not for decision making. No win is ever a bad thing. Winning at home against New Haven isn’t/shouldn’t be more than trivially good for you, but it’s certainly not a bad thing. Likewise, losing a game at Michigan shouldn’t be more than immaterially bad for you, but losing a game is never a good thing. (But, if you’re trying to predict the future, as opposed to judging performance, it’s not hard to see why margins would matter.)

 

[JustbrewitMan]

We didn't drop on KenPom though, our rating went from 29.44 to 29.69 after beating Seton Hall

It amazes me that the "basic" thing that KenPom measures just goes so above so many people's heads.

Once again, KenPom just ranks teams on "NetRtg"

NetRtg is just the difference between the team's ORtg and DRtg

Both of those are the Points scored (Offense) or points allowed (Defense) indexed to 100 possessions.

The raw averages for those are adjusted for the strength of the opponent, pace of play, "luck" factor (whether a crappy shooting team has a lights-out unexpected performance). KP also has some other secret sauce adjustments, especially early-season.

So, a team can have a win, but if their offense or defense underperforms compared to their previous average, their NetRtg can go down, even in a win. And they may fall in the KP ranks. Conversely, they can win a game, improve their NetRtg slightly, but if other teams change more than them, they can fall in the ranks.

Meanwhile, other teams can do the same thing on days we don't play...All of this metric-ing is going on with multiple games going on on given days.

Its why you can see your team's NetRtg change (slightly) even on days that they are not playing. The other teams that they have played are changing, as well.

The panic some people have over ORtg currently is absurd.

We are at 25th at 121.8 adjusted Pts per 100 possessions.

The #10 offense via KP metrics is Vandy at 124.4 per 100. That's a 2.6 pts per 100 possessions difference.

There are, on average, around 65-70 possessions per game. That's a difference of about 2 pts per game.

If Solo was shooting even 32-34% from 3 this year instead of just below 30%, we'd likely have a Top 10 KP offense.

 

[Inyatkin]

It amazes me that the "basic" thing that KenPom measures just goes so above so many people's heads.

Once again, KenPom just ranks teams on "NetRtg"

NetRtg is just the difference between the team's ORtg and DRtg

Both of those are the Points scored (Offense) or points allowed (Defense) indexed to 100 possessions.

The raw averages for those are adjusted for the strength of the opponent, pace of play, "luck" factor (whether a crappy shooting team has a lights-out unexpected performance). KP also has some other secret sauce adjustments, especially early-season.

So, a team can have a win, but if their offense or defense underperforms compared to their previous average, their NetRtg can go down, even in a win.

Meanwhile, other teams can do the same thing.

As mentioned before, all of this is going on with multiple games going on on given days.

Its why you can see your team's NetRtg change (slightly) even on days that they are not playing. The other teams that they have played are changing, as well.

The panic some people have over ORtg currently is absurd.

We are at 25th at 121.8 adjusted Pts per 100 possessions.

The #10 offense via KP metrics is Vandy at 124.4 per 100. That's a 2.6 pts per 100 possessions difference.

There are, on average, around 65-70 possessions per game. That's a difference of about 2 pts per game.

If Solo was shooting even 32-34% from 3 this year instead of just below 30%, we'd likely have a Top 10 KP offense.

This is all true but at some point there will be a stat that says "Every eventual champion has been top 12 in KP offense and top 16 in KP defense as of xx date since 2002" and if we're not in that range our season ends.

 

[JustbrewitMan]

This is all true but at some point there will be a stat that says "Every eventual champion has been top 12 in KP offense and top 16 in KP defense as of xx date since 2002" and if we're not in that range our season ends.

Yep, thank goodness records and stats are made to be broken...UConn has a history of doing record-breaking things!

 

[NJHusky]

In a resume metric, it doesn’t always happen. In a “pure” performance metric, you will drop every time you win by less than the metric predicted and go up every time you win by more than the metric predicted.

But you have put your thumb directly on why performance metrics are only useful as sorting tools and not for decision making. No win is ever a bad thing. Winning at home against New Haven isn’t/shouldn’t be more than trivially good for you, but it’s certainly not a bad thing. Likewise, losing a game at Michigan shouldn’t be more than immaterially bad for you, but losing a game is never a good thing. (But, if you’re trying to predict the future, as opposed to judging performance, it’s not hard to see why margins would matter.)

It’s interesting. So the only way to keep improving your metric is to keep improving your margin of victory.

 

[Woof]

We didn't drop on KenPom though, our rating went from 29.44 to 29.69 after beating Seton Hall

And that says something, if accurate. I didn't verify your numbers but I assume they are correct. It doesn't say anything definitive or earth moving, but it is something. What it says is that our performance in that game was slightly above our average but that our average fell, relative, to the field and, specifically, the teams around us. It means that at least one team improved a little more than we did.

Power ratings are simple in concept but can be very complex in practice. Simply put, they attempt to RATE a team's performance TO DATE, ON AVERAGE. They can be viewed as a benchmark in time. And that is how most people use them, as a benchmark in time. It is like a mid-term progress report. Where is a team, relative to the field, today? They are also somewhat predictive but there is uncertainty in the predictions and the uncertainty grows the further out you are trying to project.

There are two general approaches. One uses margin of victory (often called predictive or points) and one only cares about whether a team won or lost (Elo is one example). Most systems, like Pomeroy, Greenfield, Omni or Sagarin from back in the day, attempt to mix the two because, really, they both matter. Some publish both their mixed rating and their predictive/points rating. Massey does this. Some also provide their "wins" rating. Sagarin used to do this. Within this there are variations on the algorithms. For example, one can weigh recent results more heavily. Another example would be to weigh competitive games more heavily than blowouts. If the comment on Torvik is accurate, that is interesting but kinda nutty, in my opinion.

How the "wins" rating is mixed with the "points" rating accounts for a lot of the variation you see between systems. Omni seems to lean toward the "points" side, Pomeroy seems to be evenly balanced. And Massey seems to lean toward the "wins" side.

When you can see both from a single system, you get more insight. Our current team is really high in the "wins" rating because they are both winning at a high percentage but they have also beaten some really good teams. Our predictive rating is quite a bit lower because our MOVs have been low, on average, relative to the strength of our opponents. And that lowish average includes the fact that we had a postive MOV over some really good teams. That means the MOV has been low against teams we are supposed to beat.

In an ideal world, the team is rated highly in both types of approaches. Being low in both is obviously bad. If one is quite a bit higher than the other, it is up to debate as to which one is preferable.

 

[nelsonmuntz]

It’s interesting. So the only way to keep improving your metric is to keep improving your margin of victory.

The SEC and Big 12 just lay waste to the low majors on their schedules. The SEC in particular is dominating KenPom, and this year's conference currently has one of the highest conference ratings of all time. This despite the fact that the SEC has a losing record against the other majors (Big 10, Big 12, Big East, ACC) and is #4 in RPI, which just measures wins and losses.

Pomeroy has questioned whether his ratings should be used by the tournament selection committee because he did not build his models for the purpose they are being used for.

 

[Woof]

By the way, the uncertainty that I mentioned above is why a team can beat a higher rated team, even quite a bit higher, on any given day. Outcomes are affected by injury, illness, choice of refs, bad decisions the night before, GF dumped a guy, GF just found out she got pregnant, failed an exam and on and on.

Having said that, the probability that a higher rated team will win is still the reality. And the infamous Pomeroy offense/defense ratings metric to predict what teams have the best chance of winning it all, is still true. Nothing is 100% and no nerd ever said so. It is all about probabilities and expectations.

 

[auror]

It’s interesting. So the only way to keep improving your metric is to keep improving your margin of victory.

The raw margin of victory is much less important than the expected margin based on the opponent and location. Theoretically you could win your first game by 35 and win every game by 1 point less (34, then 33, etc.) and still improve your rating every game if your opponent got more difficult enough every game.

 

[freescooter]

Who is Ken Pomeroy and what does he have to do with our team, who we play, or really anything affecting the outcome of a game?

I’d be worried about Ken Pomeroy if he was a 40% 3-point shooter for Creighton or somebody. Otherwise look at Ws and Ls.

 

[RuffRuff]

The SEC and Big 12 just lay waste to the low majors on their schedules. The SEC in particular is dominating KenPom, and this year's conference currently has one of the highest conference ratings of all time. This despite the fact that the SEC has a losing record against the other majors (Big 10, Big 12, Big East, ACC) and is #4 in RPI, which just measures wins and losses.

Pomeroy has questioned whether his ratings should be used by the tournament selection committee because he did not build his models for the purpose they are being used for.

I would start looking at this as which of these leagues produces the most high quality tournament teams. You slice and dice this so many ways to try to squeeze out how it's a conspiracy. Simplify it. Which leagues produce the most top 25 teams, and then you'll see the true quality of a league. No one cares if St John's beat Ole Miss. No one cares if one league has a better quality lower and middle half non tourney teams.

The Big12 has 4 teams that are appearing in Final 4 picks. The SEC has 6 teams in the top 25. They aren't very top heavy this year like last, but stocked with teams in that next tier and compete at a fairly high level, with very good talent.

The continued beating of the drum in terms of the BE being a good conference, when it's apparent to anyone in the world with eye balls it's not this year, is weird. It's almost as if you're attempting to create a conspiracy this year that the BE is secretly good.

No one uses the RIP anymore, surprised it's still published TBH.

 

[NJHusky]

The raw margin of victory is much less important than the expected margin based on the opponent and location. Theoretically you could win your first game by 35 and win every game by 1 point less (34, then 33, etc.) and still improve your rating every game if your opponent got more difficult enough every game.

Right but if you are expected to win by 10 and y win by 12 your rating goes up. If y are then expected to win by 12 and y win by 15 your rating goes up. Etc. so even if y keep winning but not by what the algorithm thinks y should it goes down. So the only way to increase you rating is to keep winning by more than expected…a ponzi scheme

 

[superjohn]

If you're having a hard time finding the ignore function I'm sure you can get someone to help you out

What a strange post he/she made. All I did was ask which games? Instead of answering he/she chose to say we should be ignored.

 

[auror]

Right but if you are expected to win by 10 and y win by 12 your rating goes up. If y are then expected to win by 12 and y win by 15 your rating goes up. Etc. so even if y keep winning but not by what the algorithm thinks y should it goes down. So the only way to increase you rating is to keep winning by more than expected…a ponzi scheme

That's not what a ponzi scheme is.