Adjusted Offensive & Defensive Efficiency...

[HOFCeluck]

I don't understand your comment or what you are implying.

Let me ask you something. At the highest level of play, and you have two teams of equal caliber, what do you think are the chances of winning for each team?

Sorry, let me clarify.

When your success rate in the sweet sixteen is as "good" as JB's, of course you're going to say it's a crapshoot.

 

[Dave85]

Sorry, let me clarify.

When your success rate in the sweet sixteen is as "good" as JB's, of course you're going to say it's a crapshoot.

I don't know his rate. Are you saying JB's success rate in the sweet sixteen is good or bad.

What I meant by crap shoot, is when you have two equally matched teams, it's a 50/50 bet. Sometimes the shots fall sometimes they don't. From the sweet 16 winning the national championship is like throwing 4 coins and having them all come up heads.

I'm not sure what you are saying. I'm not saying JB or the Orange are one way or the other. I just was latching on JB's parity in CBB comment.

 

[SoBeCuse]

Looking for that right now, can't seem to find any. It will be interesting to see how much they move after the conference tournaments.

Good stuff here. Thanks for posting! As of a few days ago I have penciled in 3 of those 4 teams to the F4, I think. We'll see how the brackets look when they come out.

 

[690West]

AdjO - Adjusted offensive efficiency - An estimate of the offensive efficiency (points scored per 100 possessions) a team would have against the average D-I defense.

AdjD - Adjusted defensive efficiency - An estimate of the defensive efficiency (points allowed per 100 possessions) a team would have against the average D-I offense.

Because we play 2-3 zone which creates a different tempo of play, do these measure apply? Maybe they are not accurate for the Orange.

KenPom adjusts for Tempo. The guy thinks of everything!!

 

[HOFCeluck]

I don't know his rate. Are you saying JB's success rate in the sweet sixteen is good or bad.

What I meant by crap shoot, is when you have two equally matched teams, it's a 50/50 bet. Sometimes the shots fall sometimes they don't. From the sweet 16 winning the national championship is like throwing 4 coins and having them all come up heads.

I'm not sure what you are saying. I'm not saying JB or the Orange are one way or the other. I just was latching on JB's parity in CBB comment.

Seven of the last ten champions have been one seeds. Another was a two seed. How does that reconcile at all with the bolded text?

 

[Dave85]

Seven of the last ten champions have been one seeds. Another was a two seed. How does that reconcile at all with the bolded text?

I guess you are right. You would think brackets would not be busted so badly.

 

[Dave85]

KenPom adjusts for Tempo. The guy thinks of everything!!

You don't have to be snide about it. That's what I love about Syracuse fans. They are so friendly.

 

[690West]

You don't have to be snide about it. That's what I love about Syracuse fans. They are so friendly.

Lol. Dude you need to walk away for awhile. The "guy"I'm talking about is Kenpom. Sheesh

 

[Dave85]

Lol. Dude you need to walk away for awhile. The "guy"I'm talking about is Kenpom. Sheesh

4 out 3 people don't understand statistics.

If the calculated statistical measures were that accurate there would not be as many busted brackets and they would not call it "march madness".

It seems like we are always on the short end of the rankings. I can't believe we dropped to 11. I guess SOS and everything.

 

[690West]

4 out 3 people don't understand statistics.

If the calculated statistical measures were that accurate there would not be as many busted brackets and they would not call it "march madness".

It seems like we are always on the short end of the rankings. I can't believe we dropped to 11. I guess SOS and everything.

You are spot on. I just didn't get how I was being snide. Lol

 

[Cusefan0307]

Kenpom is a good indicator, but it's not the bible. Teams like pittsburgh and Gonzaga have consistently underperformed their kenpom ranking year after year. That's why brackets become a mess. People are fooled yearly on Pitt's computer numbers. Tennessee is a team this year that has a high Kenpom ranking with 11 losses. I guess the jury is still out on Louisville this year. We beat Indiana last year, and at the time they were one or two in kenpom if I recall. My best guess is there really isn't much difference in the top twenty teams based on the model. We had one bad game against BC, and our ranking tanked after that game.

 

[Waltdods]

If the calculated statistical measures were that accurate there would not be as many busted brackets and they would not call it "march madness".

This isn't right, and I think it's wrong in a way that gets to the misunderstanding a lot of people have with Pomeroy and other statistical measures. (And that misunderstanding seems to inspire a lot of weird anger.)

Imagine I flip a coin 100 times. I can tell you that on average it should come up heads about 50 times. But if it actually comes up heads 40 times or 52 times or even 70 times, that doesn't mean my original statement was wrong. And saying the coin will come up heads 50 percent of the time doesn't mean I'm wrong if it comes up tails on a particular flip (or that I'm right if it comes up heads).

This is all the Pomeroy numbers can tell you - on average this is what we would expect. The Pomeroy stats might tell you, for example, that each 4 seed has at least a 75% chance of winning its first game. But that means the chances of all the four seeds winning is only about 30 percent. So despite all this predicting, there's a ton of room for madness. Because the tournament doesn't turn on what happens "on average"; it turns on what happens once.

 

[HoustonCuse]

I don't know his rate. Are you saying JB's success rate in the sweet sixteen is good or bad.

What I meant by crap shoot, is when you have two equally matched teams, it's a 50/50 bet. Sometimes the shots fall sometimes they don't. From the sweet 16 winning the national championship is like throwing 4 coins and having them all come up heads.

I'm not sure what you are saying. I'm not saying JB or the Orange are one way or the other. I just was latching on JB's parity in CBB comment.

I'm trying to say this without being snide, but really you don't know JB's success rate in the Sweet 16 is good or bad? It is catastrophically bad in case you seriously don't know.

Conceptually, I do not disagree with your theory that teams are relatively evenly matched at those levels, but it's not really true. If there was really only a 50% chance of any given team winning those games, you wouldn't have perennial powers who win championships regularly or get to Final Fours, etc. None of these games are coin flips and these teams are always evolving, plus game planning and strategy can take a team far off from its kenpom profile in a one-off game (either on purpose or as a result of the other team's actions). These deviations can be in a positive or negative direction.

 

[HoustonCuse]

This isn't right, and I think it's wrong in a way that gets to the misunderstanding a lot of people have with Pomeroy and other statistical measures. (And that misunderstanding seems to inspire a lot of weird anger.)

Imagine I flip a coin 100 times. I can tell you that on average it should come up heads about 50 times. But if it actually comes up heads 40 times or 52 times or even 70 times, that doesn't mean my original statement was wrong. And saying the coin will come up heads 50 percent of the time doesn't mean I'm wrong if it comes up tails on a particular flip (or that I'm right if it comes up heads).

This is all the Pomeroy numbers can tell you - on average this is what we would expect. The Pomeroy stats might tell you, for example, that each 4 seed has at least a 75% chance of winning its first game. But that means the chances of all the four seeds winning is only about 30 percent. So despite all this predicting, there's a ton of room for madness. Because the tournament doesn't turn on what happens "on average"; it turns on what happens once.

Yep, this is what I was saying in an earlier post. Now if you flip that coin 10,000,000 times you'll have 50% heads to probably 3 or 4 decimal places. But in college basketball, you get one "coin flip" and the outcome of any game is just the next event that the probability distribution cranks out that day - possibly equally influenced by butterfly wing flaps in the Maldives as it is by the Monte Carlo that kenpom is running.

 

[690West]

Yep, this is what I was saying in an earlier post. Now if you flip that coin 10,000,000 times you'll have 50% heads to probably 3 or 4 decimal places. But in college basketball, you get one "coin flip" and the outcome of any game is just the next event that the probability distribution cranks out that day - possibly equally influenced by butterfly wing flaps in the Maldives as it is by the Monte Carlo that kenpom is running.

knocking the cover off the ball this morning.

 

[BuffOrange]

Kenpom is a good indicator, but it's not the bible. Teams like pittsburgh and Gonzaga have consistently underperformed their kenpom ranking year after year. That's why brackets become a mess. People are fooled yearly on Pitt's computer numbers. Tennessee is a team this year that has a high Kenpom ranking with 11 losses. I guess the jury is still out on Louisville this year. We beat Indiana last year, and at the time they were one or two in kenpom if I recall. My best guess is there really isn't much difference in the top twenty teams based on the model. We had one bad game against BC, and our ranking tanked after that game.

Here we go again with the anecdotal examples & impossible standards that nobody applies to any other system/ranking/whatever. As if picking by seed would've yielded much more accurate results w/ Indiana? His rankings predict better than seeds - would yours?

I really don't think brackets become a mess because of the 1% of people filling out brackets who even know What kp is, much less treat it as bible. They become a mess because it's a single elimination tournament.

 

[Dave85]

I'm trying to say this without being snide, but really you don't know JB's success rate in the Sweet 16 is good or bad? It is catastrophically bad in case you seriously don't know.

Conceptually, I do not disagree with your theory that teams are relatively evenly matched at those levels, but it's not really true. If there was really only a 50% chance of any given team winning those games, you wouldn't have perennial powers who win championships regularly or get to Final Fours, etc. None of these games are coin flips and these teams are always evolving, plus game planning and strategy can take a team far off from its kenpom profile in a one-off game (either on purpose or as a result of the other team's actions). These deviations can be in a positive or negative direction.

I think we are in agreement for the most part. I guess my 50% comment has more to do with the unpredictable nature of these matchups. Many times the team you think is going to win doesn't. There are just so many human factors involved. An objective and predictive analysis is probably not possible (except for games like a 1 seed versus 16 seed).

 

[CU44SE]

I think we are in agreement for the most part. I guess my 50% comment has more to do with the unpredictable nature of these matchups. Many times the team you think is going to win doesn't. There are just so many human factors involved. An objective and predictive analysis is probably not possible (except for games like a 1 seed versus 16 seed).

Not saying its entirely right or wrong, but last year for the first time I filled out my bracket using a combination of Kenpom, BMI, and my own equations with the advanced metrics and my bracket was in the 98.9% on ESPN. There is a reason schools use these advanced metrics though, and according to those advanced metrics, Indiana was actually expected to lose to Syracuse. Again, not saying these stats are god almighty but they definitely have some truth to them.

 

[CU44SE]

It's interesting, I'm going through the teams that have been in the top 25 in both (hoping SU will get there after the tournament) and 76% currently have made it to at least the Sweet 16. Exactly 50% made it to the Elite 8. As Knicks 411 rightly notes, some of this is definitely skewed by how far they made it into the tournament/who they played.

I may just e-mail Pomeroy to see if he has these numbers right after conference tournaments for all of those years, to get a better idea.

Sports-refrence.com/cbb has very similar stats to Kenpoms and you can look at individual games and season game logs. Obviously you would need to average them out and it would be a lot of work but you could calculate the teams averages prior to the NCAA tournament. I have a lot of work to do, but I may check some out and post them if I have the chance.

 

[OttoinGrotto]

Efficient market theory will probably slap me in the face though.

So slap it back.

 

[Cusefan0307]

Here we go again with the anecdotal examples & impossible standards that nobody applies to any other system/ranking/whatever. As if picking by seed would've yielded much more accurate results w/ Indiana? His rankings predict better than seeds - would yours?

I really don't think brackets become a mess because of the 1% of people filling out brackets who even know What kp is, much less treat it as bible. They become a mess because it's a single elimination tournament.

I follow Kenpom as much as anyone myself and i'm not saying its not a good model, because it is (I'm a Risk Analyst myself), BUT you still can't ignore the fact that certain teams (Gonzaga, Pitt, Georgetown) consistently have GREAT computer numbers, but are terrible in the tournament and consistently get beat by teams with lower Kenpom numbers.

EDIT: What I mean to say is maybe even good teams can't overcome Bad Coaching.

 

[Knicks411]

There is a reason schools use these advanced metrics though, and according to those advanced metrics, Indiana was actually expected to lose to Syracuse. Again, not saying these stats are god almighty but they definitely have some truth to them.

Is this true? IU was #3 in KP when we played them; it wasn't played at much of a favorable site for us, so what metrics had us favored?

Not saying its entirely right or wrong, but last year for the first time I filled out my bracket using a combination of Kenpom, BMI, and my own equations with the advanced metrics and my bracket was in the 98.9% on ESPN.

Assuming you meant BPI, but I like this typo; like just pick the fattest teams?

I follow Kenpom as much as anyone myself and i'm not saying its not a good model, because it is (I'm a Risk Analyst myself), BUT you still can't ignore the fact that certain teams (Gonzaga, Pitt, Georgetown) consistently have GREAT computer numbers, but are terrible in the tournament and consistently get beat by teams with lower Kenpom numbers.

Yeah but I don't think anyone takes KP as the bible, nor should they. It's a system, it's not perfect, nothing is.

On Gonzaga; looking at KP, since 2003, it looks like they have lost to a team lower ranked than they are in KP 4 times. 2 of those were in 2004 and 2005. So just twice int he last 8 tournaments. Assuming I added them right

 

[Cusefannotindc]

I think we are in agreement for the most part. I guess my 50% comment has more to do with the unpredictable nature of these matchups. Many times the team you think is going to win doesn't. There are just so many human factors involved. An objective and predictive analysis is probably not possible (except for games like a 1 seed versus 16 seed).

You are thinking about it wrong. No model is going to make perfect predictions. No one is making the argument, that one should. But can we improve on picking at random? Can we improve on picking by an informed fan? Those are the goals, not to be perfect, but to make better picks.

Even 1-16 game isn't solid, a loss just hasn't happened yet. The 16 seeds are actually probably around 2.5% to win, rather than zero.

 

[CU44SE]

Is this true? IU was #3 in KP when we played them; it wasn't played at much of a favorable site for us, so what metrics had us favored?

Using the average pace, and calculations based on adjusted O and Adjusted D compared to the opponents adjusted O and opponents Adjusted D, which I used to calculate the 98.9 percentile bracket (subtle brag), Syracuse was projected to win by 3 points. Obviously not that big of a deal or much of a projection but we were expected to win none the less. The final score was 61-50, and the projections I had were 67-64 Cuse and 72-69 Cuse. Its obviously not full proof, its just a little tool that helped me look at match-ups a little differently.

 

[Knicks411]

Using the average pace, and calculations based on adjusted O and Adjusted D compared to the opponents adjusted O and opponents Adjusted D, which I used to calculate the 98.9 percentile bracket (subtle brag), Syracuse was projected to win by 3 points. Obviously not that big of a deal or much of a projection but we were expected to win none the less. The final score was 61-50, and the projections I had were 67-64 Cuse and 72-69 Cuse. Its obviously not full proof, its just a little tool that helped me look at match-ups a little differently.

Interesting, can you explain how you did the calcs? Just curious. I know Pomeroy would have projected IU to win because they had a higher rating.