What is the deal with Pomeroy ?

[SWC75]

Definitely, and it's a useful one. As are all the stats you put together on a regular basis (which are awesome). But the 24-0 isn't the only statistic available and others can also be informative.

By all means. It's a smorgasbord. Look over the whole table. But don't forget the most important stat.

 

[Waltdods]

You might not be but there are plenty of people watching a lot of games. And I don't bother with teams outside of the top 50-100 or so but probably watch 15-20 games a week (in whole or part...part being at least 20 min of game action.

Ok, I can accept that (and envy it). Let's grant this is true - you and Pearl and whoever watch enough games that you can offer some sort of reliable opinion on who's better. Why does that make Pomeroy's efforts useless? (And of course Pomeroy opens up his positions to the public, so we can see what he gets right and wrong.)

 

[pearl31]

Ok, I can accept that (and envy it). Let's grant this is true - you and Pearl and whoever watch enough games that you can offer some sort of reliable opinion on who's better. Why does that make Pomeroy's efforts useless? (And of course Pomeroy opens up his positions to the public, so we can see what he gets right and wrong.)

Definitely not useless. Actually KenPom is one of the useful tools to refer to. Seems, though, that their is an unfortunate growing reliance on it and an unfortunate growing sentiment that he's some sort of guru

 

[Cusefannotindc]

All that he is saying is that our schedule isn't as hard as most think. That is a completely separate argument as to how good we are. That's what these advanced metric guys can't get a handle on. They only deal in the past - what happened. Not "how good is this team" as it portends to future games.

You play who is on your schedule. If you win them all - or if you win your last game in the tourney - KenPom will be forced to lick your boot.

Actually he has a measure of how good we are, and a prediction of all of our future games. Our record of 24-0 is only a measure of what has happened in the past, it does not necessarily hold predictive value or tell us how good we are.

 

[AZOrange]

Ok, I can accept that (and envy it). Let's grant this is true - you and Pearl and whoever watch enough games that you can offer some sort of reliable opinion on who's better. Why does that make Pomeroy's efforts useless? (And of course Pomeroy opens up his positions to the public, so we can see what he gets right and wrong.)

It's useless when talking about us...

 

[Cusefannotindc]

By all means. It's a smorgasbord. Look over the whole table. But don't forget the most important stat.

For seeding in the tournament it might be the most important stat. But if this team is 23-1 because Ennis misses the shot at the end of the Pitt game, is it significantly different team going forward?

 

[Orangezoo]

Actually he has a measure of how good we are, and a prediction of all of our future games. Our record of 24-0 is only a measure of what has happened in the past, it does not necessarily hold predictive value or tell us how good we are.

All predictors are predicated off data and patterns from the past. Without past data there is nothing to use to help predict. 24-0 is absolutely a measure to show how good we are. Factor in quality of opponents and the eye test and you have a holistic analysis.

 

[Cusefannotindc]

All predictors are predicated off data and patterns from the past. Without past data there is nothing to use to help predict. 24-0 is absolutely a measure to show how good we are. Factor in quality of opponents and the eye test and you have a holistic analysis.

Alone a record does not show how good a team is.

 

[Waltdods]

It's useless when talking about us...

Sorry, who's "us" here? I think Syracuse?

Do you really think the Pomeroy numbers are useless when talking about Syracuse? They show, for example, that Syracuse has a top-5 offense and a top-20 defense. That to me is a very useful data point, and it's also something that contradicts what a lot of "eye test" people tend to say about SU.

 

[Orangezoo]

Alone a record does not show how good a team is.

Re read. "A measure" does not imply "the only measure". It still is important. 23-1 vs 24-0 is important when you see the tapes and see late game execution. Its a combination of factors. Model based predictors will always only be a tool not guaranteed money.

 

[AZOrange]

Sorry, who's "us" here? I think Syracuse?

Do you really think the Pomeroy numbers are useless when talking about Syracuse? They show, for example, that Syracuse has a top-5 offense and a top-20 defense. That to me is a very useful data point, and it's also something that contradicts what a lot of "eye test" people tend to say about SU.

Yes, Syracuse...

I agree his efficiency numbers are far better that his other rankings IMO. But again, why do we have to convince anyone how good we are...? If they don't want to believe it oh well...

But pomeroys point re wsu and cuse is goofy...

 

[Cusefannotindc]

Re read. "A measure" does not imply "the only measure". It still is important. 23-1 vs 24-0 is important when you see the tapes and see late game execution. Its a combination of factors. Model based predictors will always only be a tool not guaranteed money.

Is passed "clutch performance" indicative of future?

I don't think anyone here who defends Pomeroy believes his predictions are guaranteed money, hence him putting probabilities on all games. When considering the entirety of D1 college basketball is his model superior to humans? Without a doubt.

 

[Waltdods]

Yes, Syracuse...

I agree his efficiency numbers are far better that his other rankings IMO. But again, why do we have to convince anyone how good we are...? If they don't want to believe it oh well...

But pomeroys point re wsu and cuse is goofy...

(I'm not sure what the point about convincing anyone how good we are is directed at - I agree that that's silly.)

Here's the operative WSU paragraph from Pomeroy:

I don’t bring this up to make the case that Wichita State’s schedule has been tougher than Syracuse’s. It hasn’t. But I think if you asked most basketball people, certainly talking heads on TV or people with a poll votes, you’d get something close to unanimity that winning at home against Duke is better than winning at Saint Louis, and again, that’s wrong, and wrong by a lot.


First, he says - explicitly - that WSU's schedule hasn't been tougher than SU's. I assume everyone agrees with that. Then he says that beating Duke at home is easier than beating SLU away. I take it many disagree with that. I'm not so sure myself. But the basis of his opinion is no secret. And it has nothing to do with animosity towards Syracuse or, per the OP, "arrogance." (Do you even disagree? My sense has been that folks who follow college basketball the closest are the ones with the strongest appreciation for the importance of home court advantage.)

 

[MGP]

I watch 100 percent of the games Syracuse plays and roughly 0.01 percent of the games every other D-1 team plays. With one notable, uconn hating exception, I think that's true of just about everyone on this board.

Care to comment on this, Marsh?

 

[orange2win]

Sorry, who's "us" here? I think Syracuse?

Do you really think the Pomeroy numbers are useless when talking about Syracuse? They show, for example, that Syracuse has a top-5 offense and a top-20 defense. That to me is a very useful data point, and it's also something that contradicts what a lot of "eye test" people tend to say about SU.

We are 24-0, ranked #1 in the two "official" polls and you need Pomeroy to assure you we are top 5 in offense and top 20 in defense?? Seriously?

Let me review your "eye test." Did you watch Arizona lose last night? Think they are the best prospective team in the nation? And Duke, that we have beaten twice, is still the third best team going forward?

He crunches a lot of data, which is always useful, but his conclusions are warped, because his formula is flawed.

 

[Orangezoo]

Is passed "clutch performance" indicative of future?

I don't think anyone here who defends Pomeroy believes his predictions are guaranteed money, hence him putting probabilities on all games. When considering the entirety of D1 college basketball is his model superior to humans? Without a doubt.

Lol you like to move fulcrum when trying to argue your point. Im not bashing the guy. Hell props to him as a fellow math nerd. But it is ridiculous to think you can use models to predict a champion. Any student of college basketball could give you the top 15 teams most likely to win the title. It would be congruent with much of what kenpom would say. Ken pom is beneficial in seeing where the dropoffs are in terms of who is for real and who is not. It is not any better than a human in saying who will win the whole thing. Comparing championship contenders requires qualitative analysis. Kenpom may back into a similar conclusion but its not ahead of the game. We over value these things.

 

[PoppyHart]

We are 24-0, ranked #1 in the two "official" polls and you need Pomeroy to assure you we are top 5 in offense and top 20 in defense?? Seriously?

Let me review your "eye test." Did you watch Arizona lose last night? Think they are the best prospective team in the nation? And Duke, that we have beaten twice, is still the third best team going forward?

He crunches a lot of data, which is always useful, but his conclusions are warped, because his formula is flawed.

Find me a formula or poll that isn't flawed. Nobody is saying that any of these are perfect.

 

[Knicks411]

Yes, Syracuse...

I agree his efficiency numbers are far better that his other rankings IMO. But again, why do we have to convince anyone how good we are...? If they don't want to believe it oh well...

But pomeroys point re wsu and cuse is goofy...

What is goofy about his point?

And no one is saying we need to convince anyone of anything about how good we are. The record speaks for itself and whether or not we are first or second or third is irrelevant in March.

 

[Knicks411]

We are 24-0, ranked #1 in the two "official" polls and you need Pomeroy to assure you we are top 5 in offense and top 20 in defense?? Seriously?

Well, yeah. Considering you had Doris Burke talk about how we don't score enough on Wednesday, and you have people on here all the time talk about how our offense is better than our defense.
If you used the "official" NCAA stats, we'd be what, 70th in the country in offense?

 

[Waltdods]

We are 24-0, ranked #1 in the two "official" polls and you need Pomeroy to assure you we are top 5 in offense and top 20 in defense?? Seriously?

Let me review your "eye test." Did you watch Arizona lose last night? Think they are the best prospective team in the nation? And Duke, that we have beaten twice, is still the third best team going forward?

He crunches a lot of data, which is always useful, but his conclusions are warped, because his formula is flawed.

My point with the top-5/offense and top-20/defense is that folks very, very often say that Syracuse is a better defensive team than offensive team, while Pomeroy's numbers show SU to be an elite offensive team and a very good defensive team. To me, that's an interesting point.

I'm fairly certain we have not yet beaten Duke twice. But actually, I think Duke is a good argument for the strength of the system - the human polls still have Duke at 8, but they've been playing very well lately and it seems like Pomeroy captures this better than the voters.

To revert to my earlier posts, I did not watch Arizona lose last night (I miss a lot of the seven gazillion college basketball games!). And there's nothing in Pomeroy's formula to adjust for the injury, so they're getting credit for performance by a team that, in a sense, no longer exists. So he's likely to over-rate Arizona for a while. I think we can all recognize that and adjust our thoughts accordingly.

But again, no one is arguing these ratings are right all the time. We're arguing that they're informative and transparent.

 

[pearl31]

you’d get something close to unanimity that winning at home against Duke is better than winning at Saint Louis, and again, that’s wrong, and wrong by a lot.

The bolded part is a ridiculous thing to say, ESPECIALLY for a purported objective measurement guy — and is an example of why I have always found kenpom and other college hoops numbers geeks to be insufferable.

 

[Knicks411]

The Arizona thing is a perfect example of what I was talking about before; Arizona is #1 in Pomeroy, they lose, and there is a comment like

Let me review your "eye test." Did you watch Arizona lose last night? Think they are the best prospective team in the nation?

Arizona was #2 in the AP Poll too. But no one knocks the AP poll for having Arizona #2.

 

[orange2win]

My point with the top-5/offense and top-20/defense is that folks very, very often say that Syracuse is a better defensive team than offensive team, while Pomeroy's numbers show SU to be an elite offensive team and a very good defensive team. To me, that's an interesting point.

But again, no one is arguing these ratings are right all the time. We're arguing that they're informative and transparent.

Agree entirely with your summation. I would be a KenPom fan too if he didn't rank the teams and simply presented the data. Some of his data is indeed interesting, like the point you make about our offense.

Glad you acknowledge my Duke forecast which is based on my own built-in computer algorithm.

 

[Orangezoo]

Iyiyi... all that matters is we know who the top 20 teams in the country are. Kenpom, college coaches, students of the game can all get to a similar conclusion. Those of us who enjoy math certainly like to look at these things. We also often look to create algorithms that can possibly help us understand how certain teams are as good as they are. But that approach is not better. It is simply another good way to analyze teams. Its helpful for a selection committee because it allows for more information to help process everything and make a decision even if the math ends up having zero influence.

Finally- its a media thing. Its the information age. Anything that helps sports nuts compare their team to others is going to garner attention. This is especially the case for a guy like pomeroy who is a smart dude and has a neat ratings system be it flawed or not. I think he is a bit overconfident with his model but its more a guy defending his work than anything.

 

[Waltdods]

The bolded part is a ridiculous thing to say, ESPECIALLY for a purported objective measurement guy — and is an example of why I have always found kenpom and other college hoops numbers geeks to be insufferable.

Syracuse was favored at home against Duke and then, despite being #1 in the country, an underdog at #25 Pitt (which is probably a worse team than SLU this year). I think it's pretty clear that a whole lot of people share his opinion, including those who put their money on it, even if they wouldn't put it quite as strongly as Pomeroy did.