2021 Offseason Transfers

[House P]

I'd hate for you to lose your bottom dollar. This was recently covered in another thread, but I'll recap here. If you split the last 33 years into 11-year increments (not random, 11 years ago Duke recruited its first OAD; 11 years before that, Duke had its first early entrants to the NBA draft), the summary may be found below (note that for the purposes of this post, I am NOT counting Rasheed Sulaimon as a transfer because he was kicked off the team first):

2010-11 to 2020-21: 9 transfers* (Gbinije, Murphy, Ojeleye, Jeter, D Thornton, Obi, Tucker, O'Connell, Brakefield) out of 52 freshmen (incl. transfers in) recruited during the period, equaling a 17.3% transfer out rate;

1999-2000 to 2009-10: 7 transfers (Sweet, Thompson, Boateng, Boykin, King, E Williams, Czyz) out of 38 freshmen (incl. transfers in) recruited during the period, equaling an 18.4% transfer out rate;

1988-89 to 1998-99: 6 transfers (McCaffrey, Palmer, Ast, Beard, Chappell, Burgess) out of 32 freshmen (incl. transfers in) recruited during the period, equaling an 18.8% transfer out rate.

(* - the 2021 offseason isn't over, so these numbers assume no more transfers in the next few months, which may or may not be accurate.)


So, basically, the rate of Duke players transferring out has remained pretty constant. The rate went down a little but it's basically the same, though if we consider that transfer rates across college basketball have (I believe) gone up a fair amount, Duke's numbers comparatively may have gone down significantly (I don't have the overall college bball numbers for the past 33 years at my fingertips, so I can't say with any certainty).

That said, obviously the number of Duke players recruited in has gone up significantly, in large part due to early departures to the NBA (OAD, 2AD, etc), so our total quantity of transfers out has gone up (though not by that much). Again, I don't know how that absolute increase compares to the overall increase of transfers across college basketball. And if you take out our OADs from the denominator (20 in the most recent period; 2 in the middle period; and 0 in the early period), then the later period has a much higher rate (though if you do that, should guys like O'Connell and Obi count, who transferred with just one year of eligibility remaining? there are nuances).

Overall, Duke has always had transfers under Coach K, mostly because we continually bring in top talent and some players feel like they'll never break through to become starters or stars. It doesn't appear as if this has happened significantly more frequently in the "OAD era."

WARNING: This post contains no meaningful information regarding this year's transfer portal. Please skip if you are tired of the discussion around whether or not transfer rates have increased recently at Duke.


I completely agree with the overall premise that there isn't much evidence that OAD is directly leading to an increase in transfers at Duke. Transfers are increasing across college basketball, even at places which don't have OAD players.

That being said, it does appear - to me at least - that there has been an uptick in transfers from Duke in the past few years. If you include Brakefield and Sulaimon*, but not Goldwire or Tape, Duke has had 9 transfers over the past 8 years. As far as I can tell**, there is no other 8 year period in the Coach K era where Duke had more than 5 transfers. Even if you don't include Sulaimon, 8 transfers in 8 years is still 3 more than any other 8 year period under Coach K.

Going back a bit further, there have only been 3 seasons over the past 14 without a transfer. Prior to 2007, it was rare for Duke to have consecutive years with a transfer.

Here is a plot of Duke's transfers per season since Coach K arrived. It appears that transfers at Duke were less common in the 80s, 90s, and early 00s than they have been more recently.


Transfers Per Year.jpg



*Not including Sulaimon seems like a bit of a technicality to me as a) he did transfer to Maryland, and b) I have no idea what "behind the scenes" conversations may have happened with the other 43 players who transferred under Coach K - it is entirely possible that some other players were not exactly welcomed back.

**This analysis was fairly manual and I wouldn't be shocked if I missed something.

[Kedsy]

Not including Sulaimon seems like a bit of a technicality to me as a) he did transfer to Maryland, and b) I have no idea what "behind the scenes" conversations may have happened with the other 43 players who transferred under Coach K - it is entirely possible that some other players were not exactly welcomed back.

Sulaimon was kicked off the team before he transferred and thus was not on the team when he grad-transferred to Maryland. I'm 100% positive none of the other players were kicked off the team.

Prior to 2007, it was rare for Duke to have consecutive years with a transfer.

Duke had a transfer out in 1998, 1999, 2001, 2004, 2006, 2007, 2008, 2009, and 2010 (all before we "joined" the OAD era). Unless there's some special reason for using 8-year periods, your use of a somewhat unusual number to group by* and a seemingly arbitrary cutoff of 2007 would appear to me to skew the analysis.

I always enjoy your analysis and I understand you agree the numbers don't show evidence that any increase in transfers were caused by OAD. And I don't wish to relitigate whether or not calculating a transfer rate (# of transfers over # of rostered players) is appropriate. But without that and/or a comparison of Duke's transfer rate to that of college basketball as a whole, I'm not sure how much this tells us.


*(yes, I know, I used 11-year periods, but I did so because certain milestones (i.e., recruiting our first expected OAD in 2011 (11 seasons ago) and our first early entrants in 1999 (11 seasons before that)); if we used 10 year periods, I believe the transfer curve would look even smoother, or at least the same)

[Truth&Justise]

UNC losing a second big-man named "Walker"--senior walk-on Walker Miller will transfer.

[FerryFor50]

Wow... I was really impressed With Richmond this season and I know plenty of Orange fans were eager for Boehiem to give him more time (at the expense of Joe Girard). For a while during the season, Richmond had the highest steal percentage of any player in all of college basketball. He ended the year at #13 in that category, but that still makes him an incredibly dangerous defender.

I expect him to be a really good player wherever he lands. He does not need to look to go to a smaller program. He was a top 100 recruit a year ago and had offers from Fla St, UConn, Georgetown, and Cincy when he picked Syracuse. I suspect we will see some even bigger programs take an interest in him now.

This is similar to how Boeheim used Dion Waiters. Limited minutes, coming off the bench. And similarly, they both should have been starting and playing more.

[amusedcupcake]

UNC losing a second big-man named "Walker"--senior walk-on Walker Miller will transfer.

Bet he goes to play for his big brother at UNC-G.

[UrinalCake]

Cole Swider from Villanova is in the portal. The great Jay Wright loses a player. Is the world coming to an end?!?

[CDu]

Cole Swider from Villanova is in the portal. The great Jay Wright loses a player. Is the world coming to an end?!?

He joins Quinerly and Painter has transfers from Villanova in the past 3 years. Gotta wonder if Bryan Antoine (former top-20 recruit who has been injured a lot and has rarely played even when healthy) will join them.

[Skydog]

I understand you agree the numbers don't show evidence that any increase in transfers were caused by OAD. And I don't wish to relitigate whether or not calculating a transfer rate (# of transfers over # of rostered players) is appropriate. But without that and/or a comparison of Duke's transfer rate to that of college basketball as a whole, I'm not sure how much this tells us.
...

Defining the question clearly is vital or else the arguments go in circles. Regarding changes in rates of transfers there are a lot of interrelated, but different questions. Are more people transferring out of Duke recently? Yes, clearly. But that gives no information about the why or whether its just a statistical artifact.

So lets focus on one question that does give us some insight. Start with this one:
Does recruiting OAD's, in and of itself, make it more likely that a random player chosen from the non-OAD's* will transfer? I'm limiting the player pool to the non-OAD's because - by definition - OAD's can't transfer.

That question is answered by comparing two probabilities: P(random non-OAD player will transfer) in the OAD era and P(random non-OAD player will transfer) in the pre-OAD era.
Also P(player will transfer) = # of players who transfer / total # of players eligible to transfer.

So we just need 4 numbers to answer this particular question:
pre-OAD era:
a)# of non-OAD players who entered Duke (same as # of players who entered Duke)
b)# of those players who transferred out

OAD era:
c)# of non-OAD players who entered Duke
d)# of those players who transferred out

P(random non-OAD player will transfer) in the pre-OAD era = a/b
P(random non-OAD player will transfer) in the OAD era = c/d

So if c/d > a/b then yes, the likelihood a player will transfer during the OAD era is higher. If indeed P(transfer) is significantly higher under OAD then it is suggestive (but not proof) that OAD is the cause. But it doesn't answer the question of [I]how[I] OAD is causal. Could be a few explanations, each of which would need to be explored.

What am I missing?

Note: Players who transfer in should be completely left out of these analyses because they have fewer opportunities (seasons) to transfer out.

[Kedsy]

Defining the question clearly is vital or else the arguments go in circles. Regarding changes in rates of transfers there are a lot of interrelated, but different questions. Are more people transferring out of Duke recently? Yes, clearly. But that gives very little information about whats going.

So lets focus on one question that does give us some insight. Start with this one:
Does recruiting OAD's, in and of itself, make it more likely that a random player chosen from the non-OAD's* will transfer? I'm limiting the player pool to the non-OAD's because - by definition - OAD's can't transfer.

That question is answered by comparing two probabilities: P(random non-OAD player will transfer) in the OAD era and P(random non-OAD player will transfer) in the pre-OAD era.
Also P(player will transfer) = # of players who transfer / total # of players eligible to transfer.

So we just need 4 numbers to answer this particular question:
pre-OAD era:
a)# of non-OAD players who entered Duke (same as # of players who entered Duke)
b)# of those players who transferred out

OAD era:
c)# of non-OAD players who entered Duke
d)# of those players who transferred out

P(random non-OAD player will transfer) in the pre-OAD era = a/b
P(random non-OAD player will transfer) in the OAD era = c/d

So if c/d > a/b then yes, the likelihood a player will transfer during the OAD era is higher.

What am I missing?

Note: Players who transfer in should be completely left out of these analyses because they have fewer opportunities (seasons) to transfer out.

You may be on the right track, but I suspect it's more complicated than this.

First of all, we did have OADs in the "pre-OAD era" (Maggette and Deng). We also had several early entrants who left after two or three seasons in that era. Second, the fact that, e.g., Seth Curry only had three seasons at Duke instead of four, doesn't mean they should be "completely left out of these analyses" -- that makes no sense to me. Third, players transfer at different times (e.g., one semester, one year, 1.5 years to 2 years, 3 years, even four years (Justin Goldwire, Sean Obi, Patrick Tapé); how does that effect the calculus? Fourth, players leave to go pro at different times (one, two, or three years). Fifth, how do we define "recruiting OADs" in your initial premise? A guy like Cassius Stanley as #33 in RSCI, did anyone realistically think he was an OAD when he was recruited? As a recruiting matter, how would we distinguish between, e.g., OADs Winslow (#13), Jackson (#14), Trent (#14), and Johnson (#11) from non-OADs Sulaimon (#12), Bolden (#11), Tre Jones (#13), and Hurt (#12)? Finally, wouldn't it be b/a and d/c, rather than vice versa?

[Kedsy]

You may be on the right track, but I suspect it's more complicated than this.

First of all, we did have OADs in the "pre-OAD era" (Maggette and Deng). We also had several early entrants who left after two or three seasons in that era. Second, the fact that, e.g., Seth Curry only had three seasons at Duke instead of four, doesn't mean they should be "completely left out of these analyses" -- that makes no sense to me. Third, players transfer at different times (e.g., one semester, one year, 1.5 years to 2 years, 3 years, even four years (Justin Goldwire, Sean Obi, Patrick Tapé); how does that effect the calculus? Fourth, players leave to go pro at different times (one, two, or three years). Fifth, how do we define "recruiting OADs" in your initial premise? A guy like Cassius Stanley as #33 in RSCI, did anyone realistically think he was an OAD when he was recruited? As a recruiting matter, how would we distinguish between, e.g., OADs Winslow (#13), Jackson (#14), Trent (#14), and Johnson (#11) from non-OADs Sulaimon (#12), Bolden (#11), Tre Jones (#13), and Hurt (#12)? Finally, wouldn't it be b/a and d/c, rather than vice versa?

Also, like I mentioned when I responded to House P, this analysis can't prove anything unless the respective transfer rates at Duke are compared to transfer rates in the rest of college basketball. If transfers from 2011 to 2021 have gone up (compared to 2000 to 2010 and 1989 to 1999) the same or more amount as/than other college teams than they have at OAD schools like Duke and Kentucky, then that would suggest that OAD has little if anything to do with the transfers. If transfers have gone up by a lot more at, e.g., Duke and Kentucky (something I'm pretty sure is not true but I don't know for certain) then that would suggest OADs have something to do with the transfers.

[mkirsh]

You may be on the right track, but I suspect it's more complicated than this.

First of all, we did have OADs in the "pre-OAD era" (Maggette and Deng). We also had several early entrants who left after two or three seasons in that era. Second, the fact that, e.g., Seth Curry only had three seasons at Duke instead of four, doesn't mean they should be "completely left out of these analyses" -- that makes no sense to me. Third, players transfer at different times (e.g., one semester, one year, 1.5 years to 2 years, 3 years, even four years (Justin Goldwire, Sean Obi, Patrick Tapé); how does that effect the calculus? Fourth, players leave to go pro at different times (one, two, or three years). Fifth, how do we define "recruiting OADs" in your initial premise? A guy like Cassius Stanley as #33 in RSCI, did anyone realistically think he was an OAD when he was recruited? As a recruiting matter, how would we distinguish between, e.g., OADs Winslow (#13), Jackson (#14), Trent (#14), and Johnson (#11) from non-OADs Sulaimon (#12), Bolden (#11), Tre Jones (#13), and Hurt (#12)? Finally, wouldn't it be b/a and d/c, rather than vice versa?

A more simplistic way would be looking at "3 true outcomes" (to steal a baseball phrase): 1) players that use all of their college eligibility at Duke; 2) players who leave Duke to play at another school other than Duke; 3) players who leave Duke to play professionally. Without looking at the data I can surmise that group #1 is way down as a % and group #3 is way up, but I don't know how #2 would net out. However, even this would be flawed because a) Duke is not in a vacuum and the % in each category in college basketball as a whole has changed drastically over time, so we would need a benchmark to compare Duke to, and b) I think the thesis being proposed is that OAD has "caused" more transfers by recruiting over players, which can't be determined by this sort of analysis anyway.

[House P]

You may be on the right track, but I suspect it's more complicated than this.

First of all, we did have OADs in the "pre-OAD era" (Maggette and Deng). We also had several early entrants who left after two or three seasons in that era. Second, the fact that, e.g., Seth Curry only had three seasons at Duke instead of four, doesn't mean they should be "completely left out of these analyses" -- that makes no sense to me. Third, players transfer at different times (e.g., one semester, one year, 1.5 years to 2 years, 3 years, even four years (Justin Goldwire, Sean Obi, Patrick Tapé); how does that effect the calculus? Fourth, players leave to go pro at different times (one, two, or three years). Fifth, how do we define "recruiting OADs" in your initial premise? A guy like Cassius Stanley as #33 in RSCI, did anyone realistically think he was an OAD when he was recruited? As a recruiting matter, how would we distinguish between, e.g., OADs Winslow (#13), Jackson (#14), Trent (#14), and Johnson (#11) from non-OADs Sulaimon (#12), Bolden (#11), Tre Jones (#13), and Hurt (#12)? Finally, wouldn't it be b/a and d/c, rather than vice versa?

When I created the spreadsheet I used for the analysis I posted last night, I included a column labeled "transfer opportunities". In short, I looked at each season's roster and determined how many scholarship players had the opportunity to transfer out that year. This is basically the number of scholarship players each year minus those that left for the pros (OAD, 2AD, 3AD) and seniors who had exhausted their eligibility (as much as Greg Newton may have wished to transfer after his senior year, he didn't have the chance).

For example, the 2016-17 team started the season with 13 scholarship players. Three players (Tatum, Giles, Jackson), went pro early and two Seniors (Jones, Jefferson) exhausted their eligibility, so the team had 8 "transfer opportunities". One player (Jeter) transferred, so the "transfer rate" was 12.5% (1/8).

This got somewhat complicated (how to handle guys like Justin Robinson, Reggie Love, and JD Simpson?, what to do about incoming transfers in the pre-grad transfer era?), but the general trends seemed to roughly mimic the trend in the raw number of transfers - low transfer rates in the 80s, a moderate rates in the 90s and early 00s and slightly/somewhat increased rates starting some time around 2006.

I can post the table or graphs if anyone would like, but I didn't spend much time quality checking, so there are likely a few errors.


That being said, I am not sure what this says about OAD as transfer rates have been increasing across the country since 2006 (I can't find any nationwide figures before 2006).

[Skydog]

You may be on the right track, but I suspect it's more complicated than this.

First of all, we did have OADs in the "pre-OAD era" (Maggette and Deng). We also had several early entrants who left after two or three seasons in that era. Second, the fact that, e.g., Seth Curry only had three seasons at Duke instead of four, doesn't mean they should be "completely left out of these analyses" -- that makes no sense to me. Third, players transfer at different times (e.g., one semester, one year, 1.5 years to 2 years, 3 years, even four years (Justin Goldwire, Sean Obi, Patrick Tapé); how does that effect the calculus? Fourth, players leave to go pro at different times (one, two, or three years). Fifth, how do we define "recruiting OADs" in your initial premise? A guy like Cassius Stanley as #33 in RSCI, did anyone realistically think he was an OAD when he was recruited? As a recruiting matter, how would we distinguish between, e.g., OADs Winslow (#13), Jackson (#14), Trent (#14), and Johnson (#11) from non-OADs Sulaimon (#12), Bolden (#11), Tre Jones (#13), and Hurt (#12)? Finally, wouldn't it be b/a and d/c, rather than vice versa?

I guess the way to handle the surprise OADs, 2AD’s, etc is like a disease exposure: transfers per player year of transfer exposure. At the end of each year - players who aren’t drafted or run out of eligibility have the choice of transferring. So at end of each season divide the # of duke players who transfer by # of players who have that choice. Oops - looks like that is what House P essentially did.

[JasonEvans]

Well, here's an interesting name: https://twitter.com/jeffborzello/sta...72730404839425

Memphis guard Boogie Ellis has entered the transfer portal, sources told ESPN. Averaged 10.2 points this season.

Now that Cassius Stanley isn't here, perhaps Boogie wants to come back to his first love

[Mtn.Devil.91.92.01.10.15]

Well, here's an interesting name: https://twitter.com/jeffborzello/sta...72730404839425



Now that Cassius Stanley isn't here, perhaps Boogie wants to come back to his first love

Funny how that thread reemerged this week. Perhaps someone had inside information?

Hrmmm

[Kedsy]

Well, here's an interesting name: https://twitter.com/jeffborzello/sta...72730404839425

Now that Cassius Stanley isn't here, perhaps Boogie wants to come back to his first love

He would seem to be a little redundant on next year's Duke team, don't you think?

[JasonEvans]

He would seem to be a little redundant on next year's Duke team, don't you think?

I could see this as a backup plan to Keels. We all seem to agree that Duke needs another perimeter player. If we don't get Keels then Boogie (who hit nearly 40% of his threes this season) could be a very nice option.

[brevity]

Now that Cassius Stanley isn't here, perhaps Boogie wants to come back to his first love

Funny how that thread reemerged this week. Perhaps someone had inside information?

I brought back that thread, and let me tell you, this is pure coincidence. I saw a story about Memphis beating Mississippi State in the NIT final, and saw an opportunity to mention four familiar names (head coaches Penny Hardaway and Ben Howland, Memphis leading scorer Boogie Ellis, and Memphis Most Outstanding Player Landers Nolley II).

I don't know anything about Memphis, but by all traditional measures, an NIT champion with a roster of 12 underclassmen would have a bright future. Ellis ranks third in scoring average and fourth in overall minutes, and would appear to be at the top of the backcourt depth chart. I don't know the recruiting situation, or whether having Jayden Hardaway (the coach's son) on the bench creates more tension. Maybe Ellis' reasons are personal and have nothing to do with basketball.

[Bay Area Duke Fan]

I could see this as a backup plan to Keels. We all seem to agree that Duke needs another perimeter player. If we don't get Keels then Boogie (who hit nearly 40% of his threes this season) could be a very nice option.

I think Duke needs help at the point guard position. Can either Keels or Ellis provide that? What other options are available?

[Mtn.Devil.91.92.01.10.15]

I brought back that thread, and let me tell you, this is pure coincidence. I saw a story about Memphis beating Mississippi State in the NIT final, and saw an opportunity to mention four familiar names (head coaches Penny Hardaway and Ben Howland, Memphis leading scorer Boogie Ellis, and Memphis Most Outstanding Player Landers Nolley II).

I don't know anything about Memphis, but by all traditional measures, an NIT champion with a roster of 12 underclassmen would have a bright future. Ellis ranks third in scoring average and fourth in overall minutes, and would appear to be at the top of the backcourt depth chart. I don't know the recruiting situation, or whether having Jayden Hardaway (the coach's son) on the bench creates more tension. Maybe Ellis' reasons are personal and have nothing to do with basketball.

Funny how much more quickly we tend to recognize the possibility of non-basketball reasons for transfers who aren't wearing our jerseys!

To be fair, I wholly agree. Was making this point on the Goldwire thread last week.