Model Outputs

[fluoxetine]

I am working on a new model. Currently all I have done is some rather simple stuff that works on scores only but it is still interesting to look at (at least for nerds like me).

Using scores from this year only (not sure if the teams are connected yet.. if not it does use starting ratings to set the avg rating of each connected component):

Rnk	Team	Off	Def	Games	Total
1​	Iowa	29.58749468​	-1.229251233​	6​	28.35824​
2​	Baylor	22.93263378​	3.413697196​	4​	26.34633​
3​	New Mexico	14.62801657​	11.122128​	1​	25.75014​
4​	Arkansas	14.97827769​	10.3868334​	6​	25.36511​
5​	St Louis	14.62793652​	10.27128758​	4​	24.89922​
6​	Syracuse	15.70791898​	8.656654685​	5​	24.36457​
7​	Illinois	16.76512406​	7.443758161​	6​	24.20888​
8​	NC State	17.52388125​	6.391552867​	3​	23.91543​
9​	Houston	7.805380013​	15.8387841​	4​	23.64416​
10​	Gonzaga	27.92547424​	-4.646054559​	3​	23.27942​
11​	Clemson	-0.785993814​	22.75857648​	5​	21.97258​
12​	Missouri	10.10825677​	11.64262038​	5​	21.75088​
13​	Rutgers	10.6763482​	10.58235679​	5​	21.2587​
14​	Florida St	8.8711354​	12.37817911​	3​	21.24931​
15​	Texas	9.049264496​	11.74028266​	6​	20.78955​
16​	Maryland	10.52539294​	10.07498934​	6​	20.60038​
17​	Indiana	6.751515838​	13.53181867​	6​	20.28333​
18​	LSU	16.5913887​	3.261717936​	5​	19.85311​
19​	Florida	12.17872969​	6.752541501​	4​	18.93127​
20​	Stanford	5.973241264​	12.89582121​	4​	18.86906​
21​	North Carolina	6.06836065​	12.45639961​	6​	18.52476​
22​	West Virginia	7.180664525​	11.29951904​	7​	18.48018​
23​	Villanova	7.586416738​	10.65781215​	6​	18.24423​
24​	Virginia	-0.974420895​	18.8884771​	4​	17.91406​
25​	VCU	6.550815444​	11.34808654​	7​	17.8989​

Using the preseason expectations and mean regression components that make it more predictive:

Rnk	Team	Fitted O	Fitted D	Fitted Total
1​	Gonzaga	17.15699681​	1.559420905​	18.71641772​
2​	Iowa	17.84174689​	-0.100003923​	17.74174296​
3​	Baylor	10.1680713​	7.414169116​	17.58224041​
4​	Houston	5.526529707​	10.111769​	15.63829871​
5​	Arkansas	9.812222671​	5.791983797​	15.60420647​
6​	Maryland	7.234213135​	8.301625119​	15.53583825​
7​	Illinois	9.372209353​	5.907686723​	15.27989608​
8​	Kansas	7.505906454​	7.529860857​	15.03576731​
9​	Villanova	6.603651107​	8.327942631​	14.93159374​
10​	Texas Tech	3.621731878​	11.17676102​	14.79849289​
11​	West Virginia	6.196258367​	8.455066833​	14.6513252​
12​	Florida St	7.40183043​	6.922150298​	14.32398073​
13​	Syracuse	8.334309667​	5.873910426​	14.20822009​
14​	Indiana	4.839282115​	9.002567811​	13.84184993​
15​	Louisville	5.854011293​	7.920949454​	13.77496075​
16​	Virginia	-3.812566656​	17.57622344​	13.76365678​
17​	Purdue	4.930595001​	8.662936963​	13.59353196​
18​	Florida	5.832034166​	7.626022017​	13.45805618​
19​	Rutgers	4.793647182​	8.64011635​	13.43376353​
20​	Texas	3.892354264​	9.477535014​	13.36988928​
21​	LSU	12.01931389​	1.340510437​	13.35982432​
22​	Penn St	7.35278597​	5.79325159​	13.14603756​
23​	Michigan	5.670628498​	7.15367295​	12.82430145​
24​	Clemson	-0.098289149​	12.90852689​	12.81023774​
25​	North Carolina	6.769854092​	5.850920522​	12.62077461​

To predict a game:
Home score = League Avg Pts per 40 + Home Off Rating - Away Def Rating + HFA/2
Away score = League Avg Pts per 40 + Away Off Rating - Home Def Rating - HFA/2

Current HFA = 3.281383097
Current Avg Pts per 40 = 70.9805550311497

Cuse / Maryland have performed quite well in their other games..

 

[RUTGERZ_R00LZ]

I'd like to work on a new model too. Heidi Klum would more than suffice.

 

[GoodOl'Rutgers]

SIX of the next 7 games are against Top 20 teams.

That is all.

I'll go out on a limb.. win them all and Rutgers is Number ONE.

OP.. how is Syracuse ahead of Rutgers (and Gonzaga?) in that first table?

 

[Greene Rice FIG]

Takeaway...

don’t schedule NewMexico

 

[fluoxetine]

SIX of the next 7 games are against Top 20 teams.

That is all.

I'll go out on a limb.. win them all and Rutgers is Number ONE.

OP.. how is Syracuse ahead of Rutgers (and Gonzaga?) in that first table?

I'd have to look into it a bit. It's based on raw scoring so their demolition of Boston College helps a lot.

 

[fluoxetine]

Game: Predicted Spread (Predicted Score) - Rutgers Win %

vs Illinois: Rutgers -1.5 (Rutgers 72, Illinois 70) - 56%

at Ohio St: Ohio St -1.5 (Ohio St 69, Rutgers 68) - 44%

vs Purdue: Rutgers -3 (Rutgers 69, Purdue 66) - 63%

vs Iowa: Iowa -1 (Iowa 79, Rutgers 78) - 46%

at Michigan St: Michigan St -1.5 (Michigan St 73, Rutgers 72) - 44%

vs Ohio St: Rutgers -5 (Rutgers 71, Ohio St 66) - 70%

at Penn St: Penn St -3 (Penn St 71, Rutgers 68) - 38%

vs Wisconsin: Rutgers -5 (Rutgers 67, Wisconsin 62) - 69%

at Indiana: Indiana -3.5 (Indiana 69, Rutgers 65) - 35%

vs Michigan St: Rutgers -5 (Rutgers 75, Michigan St 70) - 69%

at Northwestern: Rutgers -4 (Rutgers 71, Northwestern 67) - 65%

vs Minnesota: Rutgers -7.5 (Rutgers 74, Minnesota 67) - 77%

at Iowa: Iowa -7.5 (Iowa 82, Rutgers 74) - 22%

vs Northwestern: Rutgers -10.5 (Rutgers 74, Northwestern 64) - 84%

at Michigan: Michigan -2.5 (Michigan 70, Rutgers 67) - 39%

vs Maryland: Rutgers -1 (Rutgers 69, Maryland 68) - 55%

vs Indiana: Rutgers -3 (Rutgers 68, Indiana 66) - 62%

at Nebraska: Rutgers -7.5 (Rutgers 74, Nebraska 66) - 78%

at Minnesota: Rutgers -1 (Rutgers 71, Minnesota 70) - 53%

Predicted Record: 16 - 8 (12 - 8)

 

[fluoxetine]

Results So Far

Date	Opponent	Predicted Spread	Predicted Score	Model Win %	Vegas Spread	Vegas Implied Win %	Actual Result
11/25/2020​	vs Sacred Heart	Rutgers -16	Rutgers 76, Sacred Heart 60	93%​	Rutgers -27.5	99%​	Rutgers 86, Sacred Heart 63 (Rutgers by 23)
11/27/2020​	vs F Dickinson	Rutgers -19.5	Rutgers 79, F Dickinson 60	96%​	Rutgers -25	99%​	Rutgers 96, F Dickinson 75 (Rutgers by 21)
11/29/2020​	vs Hofstra	Rutgers -10	Rutgers 75, Hofstra 65	84%​	Rutgers -14	92%​	Rutgers 70, Hofstra 56 (Rutgers by 14)
12/8/2020​	vs Syracuse	Rutgers -5.5	Rutgers 68, Syracuse 63	72%​	Rutgers -5	68%​	Rutgers 79, Syracuse 69 (Rutgers by 10)
12/14/2020​	at Maryland	Maryland -10	Maryland 73, Rutgers 63	16%​	Maryland -2.5	42%​	Rutgers 74, Maryland 60 (Rutgers by 14)

 

[GoodOl'Rutgers]

Results So Far

Date	Opponent	Predicted Spread	Predicted Score	Model Win %	Vegas Spread	Vegas Implied Win %	Actual Result
11/25/2020​	vs Sacred Heart	Rutgers -16	Rutgers 76, Sacred Heart 60	93%​	Rutgers -27.5	99%​	Rutgers 86, Sacred Heart 63 (Rutgers by 23)
11/27/2020​	vs F Dickinson	Rutgers -19.5	Rutgers 79, F Dickinson 60	96%​	Rutgers -25	99%​	Rutgers 96, F Dickinson 75 (Rutgers by 21)
11/29/2020​	vs Hofstra	Rutgers -10	Rutgers 75, Hofstra 65	84%​	Rutgers -14	92%​	Rutgers 70, Hofstra 56 (Rutgers by 14)
12/8/2020​	vs Syracuse	Rutgers -5.5	Rutgers 68, Syracuse 63	72%​	Rutgers -5	68%​	Rutgers 79, Syracuse 69 (Rutgers by 10)
12/14/2020​	at Maryland	Maryland -10	Maryland 73, Rutgers 63	16%​	Maryland -2.5	42%​	Rutgers 74, Maryland 60 (Maryland by 14)

can you explain the Maryland game "Actual Result" number "Maryland by 14"? If Maryland was expected to win by 10 and we win by 14.. shouldn't that number be "Rutgers by 14"? And Rutgers outperformed expectation by 24 and outperformed in every game

 

[fluoxetine]

It was a typo (I fixed it).

Rutgers outperformed the model every game thus far. The model really likes Maryland (especially so before our game).

 

[littlenis5]

Good stuff. To me, all college basketball analytics are with a grain of salt because of the vast range of competition, especially early in the season, b it still like it.

 

[GoodOl'Rutgers]

Good stuff. To me, all college basketball analytics are with a grain of salt because of the vast range of competition, especially early in the season, b it still like it.

I think it is very hard with college ball.. so much of it is specific matchups and then you got foul calls.... with so few players per team.. it is very easy for things to go horribly wrong for one team.

 

[RUDieHard]

Like any other model, the more games, the more accurate It will be.

 

[Knights 1212]

I cannot believe Syracuse is rated at # 6 in that first listing. They just barely beat Northeastern just 2 days ago. Also New Mexico has only played one game. I doubt they will stay in a Top 20. I just hope we can continue to stay healthy and continue to do well.I give you credit for working on all of this.

 

[Rhuarc]

One thing that should be noted from the model is that Rutgers hasn't played a similar amount of games so far as other peer schools. We really didn't need another OOC game unless it was for tune up purposes. With our success level, it clearly wasn't needed.