Untitled Document
Brown Columbia Cornell Dartmouth Harvard Penn Princeton Yale



Username Post: Way too early 2017-18 Continuity Model Rankings        (Topic#20174)
mrjames 
Professor
Posts: 6062

Loc: Montclair, NJ
Reg: 11-21-04
03-21-17 03:42 PM - Post#227855    

Much like last year, I've produced here the outputs of what my continuity model would predict for next year's finish. Essentially the continuity model is trained on about 10 years of Pomeroy data to analyze the changes in offensive and defensive ratings YOY based on past "house" effects for each team and the continuity of minutes.

This will evolve as we get closer to next year, but this should be a decent first blush estimate of where the teams will finish.

Team Adj Eff Marg ImpRank
Harvard +10.02 67
Yale +8.67 83
Princeton +8.37 85
Penn +2.33 133
Columbia -2.19 181
Cornell -6.27 223
Dartmouth -6.62 231
Brown -9.37 263


 
penn nation 
Professor
Posts: 21085

Reg: 12-02-04
Re: Way too early 2017-18 Continuity Model Rankings
03-21-17 06:09 PM - Post#227861    
    In response to mrjames

Hopefully you'll have underestimated Penn for a second straight year.



 
Old Bear 
Postdoc
Posts: 3988

Reg: 11-23-04
Re: Way too early 2017-18 Continuity Model Rankings
03-21-17 07:31 PM - Post#227862    
    In response to mrjames

Sometime I have a glass of wine when it's "way too early". Usually it results in an afternoon nap.

 
bradley 
PhD Student
Posts: 1842

Age: 74
Reg: 01-15-16
Re: Way too early 2017-18 Continuity Model Rankings
03-21-17 08:24 PM - Post#227864    
    In response to mrjames

Seems reasonable to me as an initial projection -- needless to say, a host of wild cards i.e. freshmen, injuries, etc. Probably a good reflection as well of what the preseason Ivy Poll will be next season. Looks like four teams with maybe a 5th team in play.

Biggest wildcard from my perspective will be the play of Makai Mason -- will he fully recover from his injury, etc. If so, he strikes me as the difference maker when it comes down to crunch time. Yale will have Mason, Copeland and Phils in the backcourt and probably play with Oni as 3 with Bruner and Reynolds. They will be tough.

 
SomeGuy 
Professor
Posts: 6391

Reg: 11-22-04
Re: Way too early 2017-18 Continuity Model Rankings
03-21-17 09:05 PM - Post#227866    
    In response to bradley

Certainly puts the top 3 as too close to call. I'm glad I saw this --I was feeling a little too bullish about Penn for next year.


 
SomeGuy 
Professor
Posts: 6391

Reg: 11-22-04
Re: Way too early 2017-18 Continuity Model Rankings
03-21-17 09:07 PM - Post#227867    
    In response to bradley

Certainly puts the top 3 as too close to call. I'm glad I saw this --I was feeling a little too bullish about Penn for next year.


 
Bryan 
Junior
Posts: 231

Loc: Philadelphia
Reg: 11-21-04
Re: Way too early 2017-18 Continuity Model Rankings
03-21-17 10:45 PM - Post#227871    
    In response to mrjames

Please let me know if I'm interpreting your figures correctly. Princeton is currently listed as +12.61 in Pomeroy. Next year you project them as +8.37, a difference of 4.24 points per 100 possessions. Princeton played at an adjusted tempo of 63.0 possessions per game this year so this equates to a projection of them being 2.68 PPG worse next year than this year.

This seems like a smaller decrease than I would have expected for a team that lost 2 first-team all-Ivy performers (including POY) and a 3rd starter who was 6th on the team in minutes. Are you projecting one or two freshmen to come in and play well right away or for players who weren't regulars this year to largely offset the loss of Cook and Weisz?




 
mrjames 
Professor
Posts: 6062

Loc: Montclair, NJ
Reg: 11-21-04
03-22-17 07:53 AM - Post#227876    
    In response to Bryan

You're thinking about it correctly.

Some of it is the players coming in - in the absence of a proper freshman weighting system, I do not penalize fully in continuity minutes that I believe have a logical replacement, either internally or from freshmen. So, for instance, the loss of Cook and Weisz are huge, but with Desrosiers, Schweiger and Much coming in, plus the possibility of Bell to take on more minutes, I don't penalize Princeton for all of the continuity loss there.

Also working in Princeton's favor is that continuity seems to have less effect on defense than offense (at least in the absence of wholesale lineup changes). Given that the Tigers were better defensively than offensively last year, the loss of players won't hit them as hard as if they were an offensive juggernaut with okay defense.

These numbers are just for fun for now. Once I get good intel on how the holes are being plugged, I can get to a better continuity model that will probably be more accurate. Since Ken and Bart Torvik have been teasing theirs, I figured I'd provide mine as well!

 
TigerFan 
PhD Student
Posts: 1871

Reg: 11-21-04
03-22-17 08:42 AM - Post#227880    
    In response to mrjames

I too am hopeful that, despite losing three solid starters (two of whom were very special players), the Tigers won't drop off too far. In addition to Bell, Cannady, Stephens, Gladson and Brennan, I believe there are some solid returning players that haven't seen a lot of playing time but will be ready to step up next year and that the incoming class has tremendous potential to make an immediate impact (and don't overlook Elija Barnes, the fourth member of the new class). The bIg question in my mind is how quickly the young Harvard, Yale and Penn teams improve.

 
Go Green 
PhD Student
Posts: 1124

Age: 52
Reg: 04-22-10
03-22-17 08:47 AM - Post#227881    
    In response to TigerFan

  • TigerFan Said:
I too am hopeful that, despite losing three solid starters (two of whom were very special players), the Tigers won't drop off too far. In addition to Bell, Cannady, Stephens, Gladson and Brennan, I believe there are some solid returning players that haven't seen a lot of playing time but will be ready to step up next year and that the incoming class has tremendous potential to make an immediate impact (and don't overlook Elija Barnes, the fourth member of the new class).



This is pretty much the same arguments the Dartmouth football fans were saying last offseason.

Hope it works out better for you guys than it did for us.




 
Bryan 
Junior
Posts: 231

Loc: Philadelphia
Reg: 11-21-04
Re: Way too early 2017-18 Continuity Model Rankings
03-22-17 11:10 AM - Post#227890    
    In response to Bryan

Thanks for the additional explanation. Another question, this about normal year-to-year improvement. If a team had all players returning from the prior year roughly how much of an increase in PPG (or Pomeroy points) would be expected? Does it matter if the team was good or bad? Does it matter if the team had mainly freshmen, sophomore or juniors in the earlier season?

Penn has everyone in their rotation returning except for Howard and appears to have a likely replacement for him in Simmons. Harvard loses an all-Ivy player in Chambers and another good player in Z but also had 5 freshmen in their rotation. Based on your projected Pomeroy gain of about 5.3 for Harvard and 2.5 for Penn I'd guess that your model projects larger improvements between freshman and sophomore years than for later years.

 
mrjames 
Professor
Posts: 6062

Loc: Montclair, NJ
Reg: 11-21-04
03-22-17 11:53 AM - Post#227894    
    In response to Bryan

Just a quick update - looks like Pomeroy changed the way his data downloads, so I was pulling in Raw rather than Adjusted ORAT/DRAT as the basis for last year's numbers. That has a minor impact on the expected numbers:

Team Adj Eff Marg ImpRank
Harvard +9.81 70
Princeton +8.37 85
Yale +8.44 86
Penn +1.78 140
Columbia -3.27 188
Cornell -5.77 219
Dartmouth -7.13 236
Brown -10.88 275

There's a little more to what's going on here than just continuity. The model also pulls teams down to or up to their prior year averages and league averages. So, for a team like Penn, it's going to have a little more drag on it than Harvard will (as well as Princeton and Yale, which both have strong historical numbers to buoy them).

Beyond that though, there is zero complexity to this model. I used to do a far more accurate version of this, looking through at actual players and projecting them out based on growth curves. That's just beyond the scope of what I have time for these days, but this model serves a nifty 80/20 purpose.

 
Columbia Alum 
Junior
Posts: 247

Age: 38
Reg: 11-15-11
03-22-17 12:40 PM - Post#227901    
    In response to mrjames

Does the model account for Makai Mason's return? If not Yale definitely seems high. Princeton might also drop off more than that and Harvard and Penn will probably be slightly higher.

I hate to say it, but the model pegs Columbia decently in my mind. A slightly improvement, but not enough to push us into the top 4. Hope the players and coaches prove me wrong in the off-season!





 
Icon Legend Permissions Topic Options
Report Post

Quote Post

Quick Reply

Print Topic

Email Topic

1805 Views




Copyright © 2004-2012 Basketball U. Terms of Use for our Site and Privacy Policy are applicable to you. All rights reserved.
Basketball U. and its subsidiaries are not affiliated in any way with any NCAA athletic conference or member institution.
FusionBB™ Version 2.1 | ©2003-2007 InteractivePHP, Inc.
Execution time: 0.017 seconds.   Total Queries: 8   Zlib Compression is on.
All times are (GMT -0500) Eastern. Current time is 04:15 PM
Top