Recruiting ranking, experience, and playing time at Duke

[nocilla]

First of all, great work Kedsy.

The only caveat is the changes that Coach K alluded to recently which Billy Dat referenced as well. Will the changes be grand enough to throw off this formula? I doubt there would be drastic enough changes to effect minutes of the top level players. I think the changes are more geared toward defense and playing style than minutes but I could see some changes to depth. Although we see reasons for playing deeper almost every year it seems and they haven't played out yet. Is this the year?

[Billy Dat]

The fact that Kedsy's work has been featured on the front page of DBR is another inspiration for posters everywhere! I know he writes official stuff for other Duke sites, so he's an experienced internet journalist, but it is still a day for all of us to be proud of one of our own!

I wonder if this formula holds true for other schools, or is it unique to Duke? That would provide interesting insights into different coaching styles and philosophies, and also substantiate tommy's points about the quality of the recruiting ranking services.

[lotusland]

First of all, great work Kedsy.

The only caveat is the changes that Coach K alluded to recently which Billy Dat referenced as well. Will the changes be grand enough to throw off this formula? I doubt there would be drastic enough changes to effect minutes of the top level players. I think the changes are more geared toward defense and playing style than minutes but I could see some changes to depth. Although we see reasons for playing deeper almost every year it seems and they haven't played out yet. Is this the year?

Maybe K's undefined proposed changes should be separate thread but I can't help but wonder if, considering the rule changes and OAD dynamic, K wants to simplify the D with perhaps less rotation, showing and hedging or, heaven forbid, more zone.

[Class of '94]

This is tremendous work Kedsy! Thank-you for taking the time to write this post and do the research to derive this formula. I think the rotation for 2014-15 season is right on point although personally I would love to see Matt Jones and Semi consistently get meaningful minutes on the floor (but I'm not sure that it will happen).

For me, it raises the age-old question of whether or not going with a 7-8 man rotation over the course of an entire season is more advantageous from a recruiting and success during the season compared to a longer rotation where a 9-10+ rotation is used (like Roy tries to do). As others have noted, it will be interesting to see what kinds of changes Coach K will bring to next season's team and if lengthening the rotation is a part of those changes (I highly doubt it but again will be interesting to see).

[dukenilnil]

Thought I'd extend the ranking, minutes prediction system and see if it added any insight to predictions of team success.
For a rough calculation, I added the rankings of the top 7 players even in years with 5 perimeter players (4 perimeter and 3 bigs) and compared it to season success. A more exacting calculation would weight the rankings of each player based on minutes/total team minutes.

99-00: 11* Sweet 16

00-01: 6.5** NATIONAL CHAMPIONS

01-02: 8.5 Sweet 16

02-03: 10.5 Sweet 16

03-04: 8.5 Final 4

04-05: 10.5 Sweet 16

05-06: 8.5 Sweet 16

06-07: 12 1st Round

07-08: 9 2nd Round

08-09: 7.5 Sweet 16

09-10: 9.5 NATIONAL CHAMPIONS (a year where weighting minutes might explain results as Dawkins 3.0 weighs down rating and he played less than 500 minutes)

10-11: 9.5*** Sweet 16

11-12: 12.5 1st round

12-13: 12 Elite 8

13-14: 11.5 1st Round

14-15: 11

* (99-00) Final number may be off as Kedsy had to estimate some player numbers
** (00-01) Really only played 6 men this year with Casey getting most of his 373 minutes while Boozer was hurt making the team have a crazy low 5 rating)
*** (10-11) Actual healthy top 7 roster equals 11 due to Kyrie injury)


Will need someone smarter than me to extrapolate useful trends from this (or with more time to do proper weighting). My 2 takeaways are that (1) our talent + experience on paper these last 4 years is significantly down from other 4 year periods which helps explain the 1st round exits better than seeding/ranking/w-l heading into tournament and (2) it looks like a Sweet 16 would be just slightly exceeding expectations

[Li_Duke]

Thought I'd extend the ranking, minutes prediction system and see if it added any insight to predictions of team success.
For a rough calculation, I added the rankings of the top 7 players even in years with 5 perimeter players (4 perimeter and 3 bigs) and compared it to season success. A more exacting calculation would weight the rankings of each player based on minutes/total team minutes.

99-00: 11* Sweet 16

00-01: 6.5** NATIONAL CHAMPIONS

01-02: 8.5 Sweet 16

02-03: 10.5 Sweet 16

03-04: 8.5 Final 4

04-05: 10.5 Sweet 16

05-06: 8.5 Sweet 16

06-07: 12 1st Round

07-08: 9 2nd Round

08-09: 7.5 Sweet 16

09-10: 9.5 NATIONAL CHAMPIONS (a year where weighting minutes might explain results as Dawkins 3.0 weighs down rating and he played less than 500 minutes)

10-11: 9.5*** Sweet 16

11-12: 12.5 1st round

12-13: 12 Elite 8

13-14: 11.5 1st Round

14-15: 11

* (99-00) Final number may be off as Kedsy had to estimate some player numbers
** (00-01) Really only played 6 men this year with Casey getting most of his 373 minutes while Boozer was hurt making the team have a crazy low 5 rating)
*** (10-11) Actual healthy top 7 roster equals 11 due to Kyrie injury)


Will need someone smarter than me to extrapolate useful trends from this (or with more time to do proper weighting). My 2 takeaways are that (1) our talent + experience on paper these last 4 years is significantly down from other 4 year periods which helps explain the 1st round exits better than seeding/ranking/w-l heading into tournament and (2) it looks like a Sweet 16 would be just slightly exceeding expectations

Talented players rarely stay to be seniors anymore, so there may be an overall trend of top-7 rankings increasing over tournament bound teams. Any extrapolation of USEFUL trends, would have to account for that, making it very difficult (unless someone on the board decides to spend his time calculating top-7 rankings for tournament bound teams the past decade and a half.

[Des Esseintes]

Talented players rarely stay to be seniors anymore, so there may be an overall trend of top-7 rankings increasing over tournament bound teams. Any extrapolation of USEFUL trends, would have to account for that, making it very difficult (unless someone on the board decides to spend his time calculating top-7 rankings for tournament bound teams the past decade and a half.

Yeah, nilnil's data doesn't tell us much unless we know whether Duke's talent/experience has slipped compared to the nation as a whole. (I strongly doubt it has.) That Duke teams have less elite experience these days is already commonly understood.

[sagegrouse]

Will need someone smarter than me to extrapolate useful trends from this (or with more time to do proper weighting). My 2 takeaways are that (1) our talent + experience on paper these last 4 years is significantly down from other 4 year periods which helps explain the 1st round exits better than seeding/ranking/w-l heading into tournament and (2) it looks like a Sweet 16 would be just slightly exceeding expectations

I did a simple regression analysis, which produced the following formula:

No of NCAA wins = 8.3 - 0.61*Index

It says that every reduction in Kedsy's index by one point reduces the expected no. of tournament wins by 0.6.

This is not surprising: least squares analysis is frequently dominated by extreme values. The 2001 team had an index of 6.5, by far the lowest, and won the national championship. At the other end, the three teams with 0 wins had indices of 12.5, 12.0, and 11.5 -- three of the worst four years. Our team with the other NC was 2010 and had a score of 9.5, just slightly better than the mean value of the sample.

I would tell you the statistics, but I am having trouble getting Excel to display them.

[MarkD83]

I did a simple regression analysis, which produced the following formula:

No of NCAA wins = 8.3 - 0.61*Index

So Duke needs an index of less than 3.8 to be assured of a Nat'l championship (6 wins). This means with a 7 man rotation the average player in the rotation has to be a 0.54 or less. You can't get this with a team of all top 10 freshman. Their average index would be 1. However, players that are ranked in the 11-20 range that stay around for 3 years would have an index per player of 0.5.

See I knew I was right in the other thread when I said we should be more excited about recruits that are ranked in the 11-20 range that stay for 3 years . Numbers do not lie. (Just don't perform an error analysis on my calculations.)

[budwom]

Fine job by Kedsy. Interesting how most of us ignore the occasional claims by K that "THIS years we're going to play a lot of guys."
Seeing is believing.

Unless I see differently this year (and I do not expect to, "big changes" or no big changes) the Kedsy formula seems to hold.

So I guess it kind of comes down to whether Semi or Jones joins the rotation. Jones obviously needs to find his shot...

[COYS]

I did a simple regression analysis, which produced the following formula:

No of NCAA wins = 8.3 - 0.61*Index

It says that every reduction in Kedsy's index by one point reduces the expected no. of tournament wins by 0.6.

This is not surprising: least squares analysis is frequently dominated by extreme values. The 2001 team had an index of 6.5, by far the lowest, and won the national championship. At the other end, the three teams with 0 wins had indices of 12.5, 12.0, and 11.5 -- three of the worst four years. Our team with the other NC was 2010 and had a score of 9.5, just slightly better than the mean value of the sample.

I would tell you the statistics, but I am having trouble getting Excel to display them.

And, as I mentioned earlier, if Rivals had not been completely off the mark by ranking Scheyer in the 70's and thus messing up his RSCI, I think 2010 would have had an 8.5 as its value, thus making your regression even more accurate (although perhaps skewing the results for previous years). Add in the fact that Dawkins' 3.0 at fewer than 500 minutes is also deceptive and that team looks even better. I know it's not fair to pick and choose which recruiting rankings we actually include, but Rivals ranking of Scheyer was so terribly off base and out of sync with every other recruiting service, it's worth at least keeping in mind.

[MarkD83]

R-sq 0.34
The real outlier is the Elite 8 in 2012-13.


Or maybe this is why I love the 2010 team even more than I did before. They really exceeded expectations.

[Kedsy]

I've been wondering how likely it is that Tyus Jones and Justise Winslow are one-and-done. So I looked over the RSCI since the one-and-done rule has been in place, and this is what I saw:

Tyus Jones, #7

Guys in the #7 to #11 range from 2006 to 2013: 40
One-and-dones in the #7 to #10 range from 2006 to 2013: 10 (Noah Vonleh #8, James Young #9, Archie Goodwin #10, Marquis Teague #7, Enes Kanter #7, Lance Stephenson #8, Tiny Gallon #10, BJ Mullens #8, Jerryd Bayless #7, JJ Hickson #8)

Justise Winslow, #13

Guys in the #12 to #16 range from 2006 to 2013: 40
One-and-dones in the #12 to #16 range from 2006 to 2013: 9 (Joel Embiid #16, Ricardo Ledo #13, Cory Joseph #13, Fab Melo #14, Kosto Koufos #12, DeAndre Jordan #13, Anthony Randolph #15, Javaris Crittenton #12, Daequan Cook #13)

Just for completeness sake:

Guys in the #17 to #25 range from 2006 to 2013: 72
One-and-dones in the #17 to #21 range from 2006 to 2013: 4 (Tyler Ennis #23, Tony Wroten #18, Daniel Orton #19, Mike Conley #21)

It's possible I missed one or two, but for the moment let's assume I didn't.


So, based on history it looks like there's around a 1 in 4 chance (25%) that each of Tyus and Justise might leave school after one season. The chance of one of the two going pro after one season would appear to be almost 42%. The chance of both coming out would be around 5.6%, and the chance of both staying in school for their sophomore year would be about 58%.

Looking at it a bit closer, four of the ten in the first group went to Kentucky, a program that encourages one-and-dones, and five of the nine in the second group were big men. Since our two guys are neither going to Kentucky nor big men, I think that improves the odds at least a little bit in Duke fans' favor.

We can hope, anyway.

[rocketeli]

Great post by Kedsy. Does anyone have the time/inclination to do the analysis for UNC or similar teams and see if it holds true for them too? That would be interesting.

I'd almost forgotten about Casey S. , darn it. Watching him play I would think of an extremely talented dancer in a Broadway show involving a lot of simulated basketball action (perhaps to be written by Andrew Lloyd Webber?) It was as if he had learned to imitate basketball moves, but his actions didn't seem to produce any tangible results on the court.

[Kedsy]

I've been wondering how likely it is that Tyus Jones and Justise Winslow are one-and-done. So I looked over the RSCI since the one-and-done rule has been in place, and this is what I saw:

Tyus Jones, #7

Guys in the #7 to #11 range from 2006 to 2013: 40
One-and-dones in the #7 to #10 range from 2006 to 2013: 10 (Noah Vonleh #8, James Young #9, Archie Goodwin #10, Marquis Teague #7, Enes Kanter #7, Lance Stephenson #8, Tiny Gallon #10, BJ Mullens #8, Jerryd Bayless #7, JJ Hickson #8)

Well, I decided to check #5 and #6:

Guys in #5 to #6 range from 2006 to 2013: 16
One-and-dones in the #5 to #6 range from 2006 to 2013: 9 (Steven Adams #5, Anthony Bennett #6, Quincy Miller #5, Tobias Harris #5, Josh Selby #6, Xavier Henry #6, DeMar DeRozan #5, Derrick Rose #5, Thaddeus Young #6)

So if you look at the #5 to #8 range (instead of #7 to #11), the odds for Tyus would look like 50% (16 of 32) instead of 25% (10 of 40). I guess the question is whether #7 is more like #5 than #9. And I have no idea what the answer is to that question. I guess it's probably somewhere in between.

But even 50% isn't a sure thing, right? Right?

[flyingdutchdevil]

Well, I decided to check #5 and #6:

Guys in #5 to #6 range from 2006 to 2013: 16
One-and-dones in the #5 to #6 range from 2006 to 2013: 9 (Steven Adams #5, Anthony Bennett #6, Quincy Miller #5, Tobias Harris #5, Josh Selby #6, Xavier Henry #6, DeMar DeRozan #5, Derrick Rose #5, Thaddeus Young #6)

So if you look at the #5 to #8 range (instead of #7 to #11), the odds for Tyus would look like 50% (16 of 32) instead of 25% (10 of 40). I guess the question is whether #7 is more like #5 than #9. And I have no idea what the answer is to that question. I guess it's probably somewhere in between.

But even 50% isn't a sure thing, right? Right?

Firstly, great stuff Kedsy. Really, really nice job overall.

Stepping away from stats for a while, I would be be surprised if Tyus is a one-and-done. I give the following reasons:

1) Physically not mature (especially compared to one-and-done PGs like Kyrie and Ennis)
2) Defensive is a huge question mark. Kyrie, Ennis, MCW and other one-and-dones didn't have as big question marks about defense (scouts got that wrong about Kyrie, huh?)
3) Tyus has the reputation of being the best play maker in his class and potentially in the whole NCAA. But he isn't known as a scorer. If he doesn't look to score this year (and with Cook, Sulaimon, Okafor, and Jefferson on board, he may not seek his own shot as much), he may not impress scouts on that facet of the game. They may see Tyus as a smaller version of Kendall Marshall, which isn't not true nor fair to Tyus.

Thus, I really don't see Tyus as a one-and-done. Two-and-done? Absolutely.

[Kedsy]

Tyus has the reputation of being the best play maker in his class and potentially in the whole NCAA. But he isn't known as a scorer.

I hope you're right about Tyus not being one-and-done, and I know high school scoring stats aren't worth all that much, but from what I can gather, Tyus averaged 28 ppg for his high school team as a sophomore, 20+ as a junior and 30+ as a senior. So, he at least scores a little. Maybe when he has better options (like in AAU or national team) he doesn't look to score so much, but he certainly seems to be able to do it.

[flyingdutchdevil]

I hope you're right about Tyus not being one-and-done, and I know high school scoring stats aren't worth all that much, but from what I can gather, Tyus averaged 28 ppg for his high school team as a sophomore, 20+ as a junior and 30+ as a senior. So, he at least scores a little. Maybe when he has better options (like in AAU or national team) he doesn't look to score so much, but he certainly seems to be able to do it.

Impressive stats. But I would take high school numbers with a grain of salt. Sean Dockery averaged 28 ppg as a senior in high school. Mason Plumlee? Only 16. Even if Tyus can score, I'm not sure he's expected to be a first, second, or third option this year. And that isn't to say he can't score but rather he'd be better as a distributor who scores opportunistically (like Cook, only with much better distribution and no hero ball).

Once Cook and Okafor leave (and potentially Sulaimon), Tyus quickly becomes a top option to score (along with distribute). I think that's why he stays at least two years: year 1 - facilitate. Year 2 - score.

[SupaDave]

Once Cook and Okafor leave (and potentially Sulaimon), Tyus quickly becomes a top option to score (along with distribute). I think that's why he stays at least two years: year 1 - facilitate. Year 2 - score.

He may not have a choice but to stay two years...

[Kedsy]

He may not have a choice but to stay two years...

You think that rule will come in that quickly and bind 2014-15 freshmen?