Historic ratings of teams in top 25 of Ken Pom AdjO & AdjD...

[A Clockwork Orange]

Since SU is pretty much top 25 in both Adjusted Offensive Efficiency and Adjusted Defensive Efficiency this year, I decided to take a look at what teams ranked in the top 25 in both have done throughout the years (starting with 2003, when Pomeroy started his ratings system).

There is a caveat with all of this -- which is that Ken Pom's historical rankings INCLUDE conference and NCAA tournaments, which means some of these numbers can be skewed. Essentially, some of these teams could be in the top 25 in both BECAUSE they made runs in the tourney, thereby inflating their ratings. I have an e-mail into Pomeroy about whether I could have access into these ratings after conference tournaments for the past 10 years, as I feel that would be a better indicator of success in the tournament.

As I said in a previous post, every national champ in the last ten years has been ranked in the top 25 in both AdjO and AdjD. I decided to go deeper to see how the top 25 teams did in the Sweet 16 and Elite 8. Here are the numbers:

1. In those 10 years, 77% of those teams (58 of 75) in the top 25 made it to the Sweet 16. This represents about 36% (58 of 160) of all teams that made the Sweet 16 in the last ten years.
2. 57% of those teams (43 of 75) in the top 25 made the Elite 8. This represents about 54% (43 of 80) of all teams that made the Elite 8 in the last ten years.
3. 35% of those teams (26 of 75) in the top 25 made the Final Four. This represents about 65% (26 of 40) of all teams that made the Final Four in the last ten years.

The files I've attached show how those top 25 teams fared in each of the last 10 years. You'll notice that many of the teams that lost, lost to other top 25 teams. You'll also notice that only three of these top 25 teams (75 of them) didn't get past the Round of 64.

EDIT: For some reason it's not allowing me to upload the excel sheet I created (transferred to .jpeg) so I'll try to do it in this thread as separate posts.

 

[A Clockwork Orange]

 

[A Clockwork Orange]

 

[Lawrinson14]

 

[Dave85]

Since SU is pretty much top 25 in both Adjusted Offensive Efficiency and Adjusted Defensive Efficiency this year, I decided to take a look at what teams ranked in the top 25 in both have done throughout the years (starting with 2003, when Pomeroy started his ratings system).

There is a caveat with all of this -- which is that Ken Pom's historical rankings INCLUDE conference and NCAA tournaments, which means some of these numbers can be skewed. Essentially, some of these teams could be in the top 25 in both BECAUSE they made runs in the tourney, thereby inflating their ratings. I have an e-mail into Pomeroy about whether I could have access into these ratings after conference tournaments for the past 10 years, as I feel that would be a better indicator of success in the tournament.

Since SU is pretty much top 25 in both Adjusted Offensive Efficiency and Adjusted Defensive Efficiency this year, I decided to take a look at what teams ranked in the top 25 in both have done throughout the years (starting with 2003, when Pomeroy started his ratings system).

There is a caveat with all of this -- which is that Ken Pom's historical rankings INCLUDE conference and NCAA tournaments, which means some of these numbers can be skewed. Essentially, some of these teams could be in the top 25 in both BECAUSE they made runs in the tourney, thereby inflating their ratings. I have an e-mail into Pomeroy about whether I could have access into these ratings after conference tournaments for the past 10 years, as I feel that would be a better indicator of success in the tournament.

As I said in a previous post, every national champ in the last ten years has been ranked in the top 25 in both AdjO and AdjD. I decided to go deeper to see how the top 25 teams did in the Sweet 16 and Elite 8. Here are the numbers:

1. In those 10 years, 77% of those teams (58 of 75) in the top 25 made it to the Sweet 16. This represents about 36% (58 of 160) of all teams that made the Sweet 16 in the last ten years.
2. 57% of those teams (43 of 75) in the top 25 made the Elite 8. This represents about 54% (43 of 80) of all teams that made the Elite 8 in the last ten years.
3. 35% of those teams (26 of 75) in the top 25 made the Final Four. This represents about 65% (26 of 40) of all teams that made the Final Four in the last ten years.

The files I've attached show how those top 25 teams fared in each of the last 10 years. You'll notice that many of the teams that lost, lost to other top 25 teams. You'll also notice that only three of these top 25 teams (75 of them) didn't get past the Round of 64.

EDIT: For some reason it's not allowing me to upload the excel sheet I created (transferred to .jpeg) so I'll try to do it in this thread as separate posts.

As I said in a previous post, every national champ in the last ten years has been ranked in the top 25 in both AdjO and AdjD. I decided to go deeper to see how the top 25 teams did in the Sweet 16 and Elite 8. Here are the numbers:

1. In those 10 years, 77% of those teams (58 of 75) in the top 25 made it to the Sweet 16. This represents about 36% (58 of 160) of all teams that made the Sweet 16 in the last ten years.
2. 57% of those teams (43 of 75) in the top 25 made the Elite 8. This represents about 54% (43 of 80) of all teams that made the Elite 8 in the last ten years.
3. 35% of those teams (26 of 75) in the top 25 made the Final Four. This represents about 65% (26 of 40) of all teams that made the Final Four in the last ten years.

The files I've attached show how those top 25 teams fared in each of the last 10 years. You'll notice that many of the teams that lost, lost to other top 25 teams. You'll also notice that only three of these top 25 teams (75 of them) didn't get past the Round of 64.

EDIT: For some reason it's not allowing me to upload the excel sheet I created (transferred to .jpeg) so I'll try to do it in this thread as separate posts.

I usually create two brackets. One where SU wins a national championship. And one that takes into account numbers like these.

 

[A Clockwork Orange]

Essentially it means in the last ten years if you're top 25 in both you have a pretty darn good shot of making it deep into the tournament. There's the tl;dr for you Lawrinson14

 

[Lawrinson14]

Essentially it means in the last ten years if you're top 25 in both you have a pretty darn good shot of making it deep into the tournament. There's the tl;dr for you.

I actually appreciate this. Anything with numbers and my brain shuts down immediately.

 

[Waltdods]

1. In those 10 years, 77% of those teams (58 of 75) in the top 25 made it to the Sweet 16. This represents about 36% (58 of 160) of all teams that made the Sweet 16 in the last ten years.
2. 57% of those teams (43 of 75) in the top 25 made the Elite 8. This represents about 54% (43 of 80) of all teams that made the Elite 8 in the last ten years.
3. 35% of those teams (26 of 75) in the top 25 made the Final Four. This represents about 65% (26 of 40) of all teams that made the Final Four in the last ten years.

This is interesting stuff. 25% of tournament teams make the Sweet 16, versus 77% of your top-25 teams; 12.5% make the Elite 8 versus 57%; and 6% make the Final 4 versus 35%. So teams in this top-25 bucket perform on average about 4ish times better than teams that aren't. (Which is a really big number.)

Thinking about it a bit more, I'm not sure this is the right comparison just because there are so many ungood teams in the tournament. Given that there are about 8 of these teams a year (75/10), it'd be interesting to see how these teams perform compared to the 1 and 2 seeds. If the performance of these teams is better than that of the top 2 seeds, that would be a strong case for its predictive power. (My guess just eyeballing the numbers is that it is going to be close.)

Maybe the better way to do this comparison is see whether these teams outperform their seeds. Probably, I would think.

 

[dollarbill44]

This is interesting stuff. 25% of tournament teams make the Sweet 16, versus 77% of your top-25 teams; 12.5% make the Elite 8 versus 57%; and 6% make the Final 4 versus 35%. So teams in this top-25 bucket perform on average about 4ish times better than teams that aren't. (Which is a really big number.)

Thinking about it a bit more, I'm not sure this is the right comparison just because there are so many ungood teams in the tournament. Given that there are about 8 of these teams a year (75/10), it'd be interesting to see how these teams perform compared to the 1 and 2 seeds. If the performance of these teams is better than that of the top 2 seeds, that would be a strong case for its predictive power. (My guess just eyeballing the numbers is that it is going to be close.)

Maybe the better way to do this comparison is see whether these teams outperform their seeds. Probably, I would think.

I would bet that a lot of those 75 teams had high seeds. It would be nice to see those figures.

 

[Darth]

Really enjoyed this analysis. I too am thinking about weighting teams in my bracket which are good on both sides of the ball. I'd love to somehow see the data that isn't skewed by tournament game data. Also agree with Waltdods - would be cool to see relatively how being good at both offense and defense predicts success relative to seeding.

 

[HoustonCuse]

Thanks for this. I look forward to hearing what Mr. Pomeroy has to say about your request.

 

[HoustonCuse]

Welp.. so much for that. While I was looking at it tonight, SU moved down to #26 in AdjO... We'll need an improved offensive performance to get back into the Top 25. Cooney getting hot from 3 like Maui would be fine.

 

[A Clockwork Orange]

Welp.. so much for that. While I was looking at it tonight, SU moved down to #26 in AdjO... We'll need an improved offensive performance to get back into the Top 25. Cooney getting hot from 3 like Maui would be fine.

We were already 26 in AdjO.

 

[Knicks411]

I think the FSU performance moved us from 31 to 26 or so.

I wonder if Pomeroy has access to the data you asked him for. The last few years, I for sure think he does. (for the last few years you can see each team's overall rating as of each game, though not the offense and defense. But I would assume that means he has it somewhere). Going back historically, I wouldn't be surprised if he doesn't have all of it.

 

[SWC75]

I would bet that a lot of those 75 teams had high seeds. It would be nice to see those figures.

Of course being in the top 25 offensively and defensively can result in a won-lost record that gets you a high seed.

 

[A Clockwork Orange]

I think the FSU performance moved us from 31 to 26 or so.

I wonder if Pomeroy has access to the data you asked him for. The last few years, I for sure think he does. (for the last few years you can see each team's overall rating as of each game, though not the offense and defense. But I would assume that means he has it somewhere). Going back historically, I wouldn't be surprised if he doesn't have all of it.

I'd be surprised if he does. I'm going to start keeping the data myself starting with this year. Hopefully he'll get back to me, but I wouldn't expect to hear anything from him until after the tourney is over.

 

[jncuse]

You can use archive.org to get data right before the tourney for some years.

Personally, I think you are focusing too much on balance in terms of the National champion. You need a good defensive team (top 20 or so) but you also need a more elite offensive team (top 10 or close to it). That has been the more typical profile of the national champ.

Yes, the champs of the last 11 years are all 25/25 teams. But 9 of the 11 champs have had top 10 offenses (most top 5) the other 2 (uconn and syracuse) were top 15. Quite a few more had sub 10 defenses (6 vs 2).

I would say balance is important, but top 10 offence is also needed.

Also using archive.org we can find that uconn did not fit this profile entering 2011. They were 12 in offence, and 27 in defense entering the tourney. Another team weighted to offence.

There have also been more final 4 teams that are offensive;y oriented rather than defensively oriented (top 10 in one, sub 40 in the other). Theory is that teams can play harder defence in the tourney. Not that either is optimal... but two teams ranked the same in KP, the team with the better offence will tend to do better.

 

[jncuse]

Interestingly we lack that elite offensive team with a good enough top 25 d right now. Top 7 offensive teams are outside of the top 25 on d.

Florida and wichita st appear to have the best mix.

 

[SWC75]

From a much longer post I did after the 2008-09 season:

So, how do we determine what’s important and what isn’t, (or what’s less important, anyway)? The method I decided to use was something I decided to call “ranking differentials”. (I’m sure somebody else must have already invented this and called it something else.) The idea is to rank the teams 1-23 in one statistic and then record the ranking they had in the statistic I wished to compare it to. (Ties get the lower number: If two teams are tied for 3rd, they both get a “3” and the next team gets a “5”.) Then I noted the numerical difference in the ranking. I added up the numerical differences and divided by 23. The result is the correlation between the two stats. The lower the resulting number, the stronger the correlation. Let’s look at how this works by comparing efficiency ratings to winning percentage:


Offensive Efficiency vs. Winning Percentage
1) 1988-89 1.217 vs. .789 (6th ) differential: 5
2) 1987-88 1.191 vs. .743 (10th) differential: 8
3) 1986-87 1.116 vs. .816 (2nd) differential: 1
4) 1989-90 1.135 vs. .787 (7th) differential: 3
5) 2004-05 1.104 vs. .794 (5th) differential: 0
6) 1990-91 1.102 vs. .813 (3rd) differential: 3
7) 2002-03 1.096 vs. .857 (1st) differential: 6
8) 1999-00 1.092 vs. .813 (3rd) differential: 5
9) 2008-09 1.089 vs. .737 (13th) differential: 4
10) 1993-94 1.085 vs. .767 (8th) differential: 2
11) 2007-08 1.080 vs. .600 (22nd) differential:11
12) 2003-04 1.070 vs. .742 (12th) differential:10
13) 1991-92 1.062 vs. .688 (16th) differential:3
14) (13)1995-96 1.062 vs. .763 (9th) differential: 4
15) 2006-07 1.055 vs. .686 (17th) differential: 2
16) 1992-93 1.054 vs. .690 (15th) differential: 1
17) 1994-95 1.037 vs. .667 (18th) differential: 1
18) 2000-01 1.034 vs. .735 (14th) differential: 4
19) 2005-06 1.026 vs. .657 (19th) differential: 0
20) 1998-99 1.021 vs. .636 (21st) differential: 1
21) 1996-97 1.020 vs. .594 (23rd) differential: 2
22) 2001-02 1.016 vs. .639 (20th) differential:2
23) 1997-98 1.014 vs. .743 (10th) differential: 13

The total of the differentials are 91. Divide by 23 teams: 3.96

Defensive Efficiency vs. Winning Percentage
1) 1998-99 0.887 vs. .636 (21th) differential: 20
2) 1996-97 0.930 vs. .594 (23nd) differential: 21
3) 1999-00 0.935 vs. .813 (3rd) differential: 0
4) 1997-98 0.949 vs. .743 (10th) differential: 6
5) 2006-07 0.951 vs. .686 (17th) differential: 12
6) 1994-95 0.957 vs. .667 (18th) differential: 12
7) (6) 1995-96 0.957 vs. .763 (9th) differential: 3
8) 2005-06 0.961 vs. .657 (19th) differential: 11
9) 2004-05 0.964 vs. .794 (5th) differential: 4
10) 1987-88 0.966 vs. .743 (10th) differential: 0
11) 2000-01 0.968 vs. .735 (14th) differential: 3
12) 2001-02 0.976 vs. .639 (20th) differential: 8
13) 1990-91 0.979 vs. .813 (3rd) differential: 10
14) 2002-03 0.980 vs. .857 (1st) differential: 13
15) 2008-09 0.980 vs. .737 (13th) differential: 1
16) 1993-94 0.983 vs. .767 (8th) differential: 8
17) 1989-90 0.983 vs. .787 (7th) differential: 10
18) 1992-93 0.988 vs. .690 (15th) differential: 3
19) 2003-04 0.994 vs. .742 (12th) differential: 7
20) 1986-87 1.001 vs. .816 (2nd) differential: 18
21) 1988-89 1.001 vs. .789 (6th) differential: 15
22) 1991-92 1.016 vs. .688 (16th) differential: 6
23) 2007-08 1.017 vs. .600 (22ndt) differential: 1

The total differentials are 192. Divide by 23 and you get 8.35, indicating that offensive efficiency is more important- in fact a lot more important- to winning than defensive efficiency. When I remember disappointing SU performances, it seems to me that the problems started out on the offensive end. The other team then takes advantage of offensive mistakes to get their own offense going. Players tend to sulk when they aren’t scoring and their defensive performance then suffers. So the offense goes bad before the defense does and the result is that offensive efficiency correlates more directly to winning than defensive efficiency.

The difference between offensive and defensive efficiency ought to make a pretty good “power” rating:

1) 1987-88 0.225 vs. .743 (10th) differential: 9
2) 1988-89 0.216 vs. .789 (6th) differential: 4
3) 1999-00 0.157 vs. .813 (3rd) differential: 0
4) 1989-90 0.152 vs. .787 (7th) differential: 3
5) 2004-05 0.140 vs. .794 (5th) differential: 0
6) 1998-99 0.134 vs. .636 (21st) differential: 15
7) 1990-91 0.123 vs. .813 (3rd) differential: 4
8) 2002-03 0.116 vs. .857 (1st) differential: 7
9) 1986-87 0.115 vs. .816 (2nd) differential: 7
10) 2008-09 0.109 vs. .737 (13th) differential: 3
11) 2006-07 0.104 vs. .686 (17th) differential: 6
12) 1995-96 0.105 vs. .763 (9th) differential: 3
13) 1993-94 0.102 vs. .767 (8th) differential: 5
14) 1996-97 0.090 vs. .594 (23rd) differential: 9
15) 1994-95 0.080 vs. .667 (18th) differential: 3
16) 2003-04 0.075 vs. .742 (12th) differential: 4
17) 1992-93 0.066 vs. .690 (15th) differential: 2
18) (16) 2000-01 0.066 vs. .735 (14th) differential: 4
19) 1997-98 0.065 vs. .743 (10th) differential: 9
20) (18) 2005-06 0.065 vs. .657 (19th) differential: 1
21) 2007-08 0.063 vs. .600 (22nd) differential: 1
22) 1991-92 0.046 vs. .688 (16th) differential: 6
23) 2001-02 0.040 vs. .639 (20th) differential: 3

The total of the differentials is 101. Divide by 23 and you get 4.39. Apparently offensive efficiency is more important than even the difference between offensive and defensive efficiency to winning, although here the difference is much smaller and thus less significant.

I won’t reproduce all the charts here but I used the same system to relate 2 point field goal percentage, 3 point field goal percentage, percentage of shots that are three pointers

The percentage of made field goals that are assisted, offensive vs. defensive rebounding, turnovers, steals and blocks to both offensive and defensive efficiency to see how those things contribute to a team’s efficiency on either end. Here is the bottom line on each:

- 2 point field goal percentage had an average differential ranking of 4.25
- Rebounding had an average differential ranking of 4.70
- Percentage of field goal attempts from three point range had an average differential ranking of 4.98
- Turnovers had an average differential ranking of 5.41
- Steals had an average differential ranking of 5.93
- 3 point field goal percentage had an average differential ranking of 6.14
- Blocks had an average differential ranking of 6.84
- Percentage of made field goals that were assisted had an average differential ranking of 8.27

In other words that’s the order of importance of these things to playing efficient basketball. If you want to win, take the ball inside, (but don’t forget the medium range jump shot) and prevent the other team from doing so. When a shot misses, get the damn ball! Take care of the ball and see if you can generate some turnovers. The rate at which three pointers are made is not as important as how many you, (or they) are taking. Blocks are overrated and assists are even more overrated.

 

[A Clockwork Orange]

Interestingly we lack that elite offensive team with a good enough top 25 d right now. Top 7 offensive teams are outside of the top 25 on d.

Florida and wichita st appear to have the best mix.

There have been many outstanding offensive teams in the past ten years that have not made noise in the tournament -- likewise many outstanding defensive teams that have not made noise. The combination of the two sets these teams apart for success in the tournament (or so it seems, with incomplete data).

 

[jncuse]

There have been many outstanding offensive teams in the past ten years that have not made noise in the tournament -- likewise many outstanding defensive teams that have not made noise. The combination of the two sets these teams apart for success in the tournament (or so it seems, with incomplete data).

You need balance.. absolutely. Your going to have to be good at both

But the minor problem I have is that you are presenting a team with an offensive efficiency between 15-25 as a topnational title contender.

Not one team in that range has won the national title since 2003. There have been defences in that 15-25 range. But not offences.

 

[jncuse]

1. In those 10 years, 77% of those teams (58 of 75) in the top 25 made it to the Sweet 16. This represents about 36% (58 of 160) of all teams that made the Sweet 16 in the last ten years.
2. 57% of those teams (43 of 75) in the top 25 made the Elite 8. This represents about 54% (43 of 80) of all teams that made the Elite 8 in the last ten years.
3. 35% of those teams (26 of 75) in the top 25 made the Final Four. This represents about 65% (26 of 40) of all teams that made the Final Four in the last ten years.

The files I've attached show how those top 25 teams fared in each of the last 10 years. You'll notice that many of the teams that lost, lost to other top 25 teams. You'll also notice that only three of these top 25 teams (75 of them) didn't get past the Round of 64.

.

25/25 teams are also going to have high seeds. This could be the main reason they are doing well. How different are the numbers from teams on same seed lines with one outlier on offence or defence?

I really like the analysis, but without an outlier comparison by seed line , its hard to reach a conclusion.

You are also correct on year end KP mumbers (I have had the same issues before)... they will tend to prop up the teams that did well in the tourney, which also make an analysis very difficult.

Archive.org has some data pre-tournament (but not all years, especially not the older ones when KP had a different format on his main page)

 

[A Clockwork Orange]

25/25 teams are also going to have high seeds. This could be the main reason they are doing well. How different are the numbers from teams on same seed lines with one outlier on offence or defence?

I really like the analysis, but without an outlier comparison by seed line , its hard to reach a conclusion.

You are also correct on year end KP mumbers (I have had the same issues before)... they will tend to prop up the teams that did well in the tourney, which also make an analysis very difficult.

Archive.org has some data pre-tournament (but not all years, especially not the older ones when KP had a different format on his main page)

I
I have time I'll go back and look at the 25/25 teams I've listed above and check out seeding and overall rank in AdjO and AdjD. Really what I'm looking for here is if you can fundamentally use this system to fill out brackets more or less accurately then however we all do it now. Obviously I'm selecting a small sub set that fits the needs of what I'm trying to look at (25/25 - since that's around where SU is this year).

I could expand it out more, or look at offensive efficiency only, and how that effects overall competitiveness in the tournament, but I'm not doing this as a dissertation (been there done that!) but just as some down time fun. I was looking specifically at where SU is this year, and what that has meant for other similar teams throughout the past ten years. I appreciate the feedback though, and I may expand it out if I have some more time today and tomorrow (spring break for a professor, what a tough life!).

Thanks for the tips, if I can get some stuff together I'll post. Of course, the biggest issue with all of this is that it is post tournament numbers, which skews a lot. If I start keeping my own database for future years, I may be able to bear these numbers out a bit more three of four years down the road (assuming Pomeroy doesn't get back to me).

 

[jncuse]

I
I have time I'll go back and look at the 25/25 teams I've listed above and check out seeding and overall rank in AdjO and AdjD. Really what I'm looking for here is if you can fundamentally use this system to fill out brackets more or less accurately then however we all do it now. Obviously I'm selecting a small sub set that fits the needs of what I'm trying to look at (25/25 - since that's around where SU is this year).

I could expand it out more, or look at offensive efficiency only, and how that effects overall competitiveness in the tournament, but I'm not doing this as a dissertation (been there done that!) but just as some down time fun. I was looking specifically at where SU is this year, and what that has meant for other similar teams throughout the past ten years. I appreciate the feedback though, and I may expand it out if I have some more time today and tomorrow (spring break for a professor, what a tough life!).

Thanks for the tips, if I can get some stuff together I'll post. Of course, the biggest issue with all of this is that it is post tournament numbers, which skews a lot. If I start keeping my own database for future years, I may be able to bear these numbers out a bit more three of four years down the road (assuming Pomeroy doesn't get back to me).

LOL, yes that little thing called time.

You put a lot of effort into this already, and thank you for that.

 

[Waltdods]

Here's a quick-and-dirty on the seed performance.

According to these guys (http://bracketscience.com/articles/_navAnalysis.asp?i=13) the average 1 seed wins 3.32 games and the average 2 seed wins 2.41 games.

By my math, the average team in Clockwork's sample wins almost exactly 3 games. (I got slightly different numbers the two times I ran this and then got lazy, but I'm pretty confident this is close enough.) So considerably better than a 2 seed. In fact, even if you assume every team is a 1 or a 2 seed, the teams in the sample will have outperformed, since the average of a 1 and 2 seed is about 2.9 wins. Obviously, that's not the case so the actual out-performance must be higher (although still confounded by the timing issue folks have raised.)