mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-11-15 06:08 PM - Post#197194
Just dusted off the old code to run game scores. Game scores are essentially a "Pythagorean Win Percentage" times 100 for a given game based on the outcome, quality of opponent and venue.
100 is the best and 0 is the worst. The way to think about it is that if you played exactly like this over and over again against a nationally average opponent, you would win X% of those games.
What we're looking for here is the quality of a team that you are on a usual basis, so the important metrics here are 25-75th percentile, means and medians. Knowing the outliers are nice to get a sense of how inconsistent a team can be as well, but they shouldn't be remotely considered to be measures of true quality.
RecordNum Year Date Venue Opp PA Ivy PF WL GS
681 2016 2015-11-25 H Lafayette 52 princeton 104 W 99.0
1084 2016 2015-11-22 A Southern Methodist 71 yale 69 L 97.2
161 2016 2015-12-02 H Bryant 40 yale 79 W 96.5
648 2016 2015-12-05 A Kansas 75 harvard 69 L 94.5
1083 2016 2015-11-29 A Southern Methodist 77 brown 69 L 93.5
486 2016 2015-11-30 H Fairleigh Dickinson 61 princeton 91 W 91.3
162 2016 2015-11-25 H Bryant 45 harvard 80 W 90.4
1156 2016 2015-12-05 H Vermont 54 yale 72 W 89.5
689 2016 2015-11-22 H Lehigh 61 columbia 88 W 87.8
479 2016 2015-11-13 N Fairfield 57 yale 70 W 85.8
823 2016 2015-11-20 A Northwestern 83 columbia 80 L 84.0
178 2016 2015-12-02 A Bucknell 61 columbia 72 W 82.9
696 2016 2015-11-29 H Long Island University 56 dartmouth 79 W 82.8
72 2016 2015-12-08 A Boston University 69 harvard 75 W 76.5
688 2016 2015-11-19 A Lehigh 67 yale 79 W 74.1
630 2016 2015-12-09 A Illinois 69 yale 65 L 72.4
649 2016 2015-11-16 A Kansas State 81 columbia 71 L 71.9
1023 2016 2015-11-13 A Rider 56 princeton 64 W 70.2
998 2016 2015-11-14 A Providence 76 harvard 64 L 65.8
664 2016 2015-11-25 H La Salle 64 penn 80 W 65.1
620 2016 2015-11-16 H Holy Cross 55 brown 71 W 64.9
1055 2016 2015-12-04 H Saint Joseph's 80 columbia 78 L 63.9
465 2016 2015-11-25 A Duke 80 yale 61 L 63.7
1184 2016 2015-11-24 H Wofford 59 columbia 70 W 62.0
452 2016 2015-12-06 H Delaware 69 columbia 82 W 56.7
1043 2016 2015-11-16 H Sacred Heart 77 yale 99 W 54.1
480 2016 2015-11-27 A Fairfield 82 columbia 81 L 53.1
55 2016 2015-11-18 H Binghamton 59 cornell 76 W 50.7
1111 2016 2015-12-05 A Stony Brook 91 princeton 77 L 47.6
1056 2016 2015-12-08 A Saint Joseph's 62 princeton 50 L 46.8
682 2016 2015-12-05 H Lafayette 67 cornell 85 W 45.9
997 2016 2015-11-21 A Providence 94 brown 73 L 44.0
724 2016 2015-12-09 A Maine 69 dartmouth 79 W 41.8
817 2016 2015-12-02 A Northeastern 80 harvard 71 L 41.8
1080 2016 2015-12-01 H Siena 80 cornell 81 W 41.1
516 2016 2015-12-07 A Georgetown 74 brown 57 L 38.4
749 2016 2015-11-17 H Massachusetts 69 harvard 63 L 36.1
1075 2016 2015-11-13 A Seton Hall 84 dartmouth 67 L 36.0
1125 2016 2015-12-09 H Temple 77 penn 73 L 34.9
530 2016 2015-12-05 H Hartford 65 dartmouth 74 W 30.8
1066 2016 2015-11-21 H Saint Peter's 72 princeton 75 W 29.0
621 2016 2015-11-29 A Holy Cross 50 harvard 49 L 28.8
732 2016 2015-12-09 A Manhattan 71 columbia 72 W 26.9
219 2016 2015-11-16 A Colgate 98 cornell 101 W 23.7
163 2016 2015-12-05 A Bryant 68 brown 76 W 21.8
198 2016 2015-12-02 H Central Connecticut State 64 brown 82 W 21.4
454 2016 2015-11-17 A Delaware State 54 penn 60 W 19.1
517 2016 2015-11-13 A Georgia Tech 116 cornell 81 L 19.0
199 2016 2015-11-15 H Central Connecticut State 61 penn 77 W 18.7
756 2016 2015-11-29 A Massachusetts-Lowell 80 cornell 77 L 17.7
1155 2016 2015-12-02 H Vermont 68 dartmouth 63 L 16.3
780 2016 2015-12-02 H Navy 65 penn 59 L 15.5
1065 2016 2015-11-13 A Saint Peter's 77 brown 65 L 14.0
1176 2016 2015-11-21 A Washington 104 penn 67 L 14.0
63 2016 2015-11-22 A Boston College 69 harvard 56 L 11.6
802 2016 2015-11-18 A Niagara 75 brown 66 L 11.2
680 2016 2015-11-29 A Lafayette 92 penn 86 L 10.4
1025 2016 2015-11-13 H Robert Morris 75 penn 76 W 6.0
705 2016 2015-11-28 H Longwood 70 columbia 69 L 5.7
911 2016 2015-11-25 A Pittsburgh 93 cornell 49 L 5.5
740 2016 2015-11-17 A Marist 73 dartmouth 63 L 5.1
190 2016 2015-11-21 A Canisius 87 cornell 62 L 5.1
513 2016 2015-12-05 A George Mason 63 penn 44 L 3.4
795 2016 2015-11-24 H New Hampshire 88 brown 77 L 2.1
13 2016 2015-11-29 A Albany (NY) 88 yale 54 L 2.0
Rec Team Avg
1 brown 34.58
2 columbia 59.49
3 cornell 26.08
4 dartmouth 35.46
5 harvard 55.68
6 penn 20.78
7 princeton 63.98
8 yale 70.58
|
20Penn14
Senior
Posts: 364
Reg: 02-26-12
|
Re: Ivy Game Scores Thus Far...
12-11-15 06:33 PM - Post#197195
In response to mrjames
I see that Penn has the 3 lowest games scores that occurred during wins. What does this mean exactly?
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-11-15 07:28 PM - Post#197200
In response to 20Penn14
Great question.
Since game scores include both opponent quality and venue, if you play bad teams at home, the national average team would win comfortably. And if you play extremely awful teams at home, the national average team would win big (or comfortably on the road).
Penn's three wins with low GS were essentially a site-adj loss to a bad team and not terribly large margin wins over a couple of the worst D-I teams.
|
20Penn14
Senior
Posts: 364
Reg: 02-26-12
|
12-12-15 03:22 PM - Post#197220
In response to mrjames
Thanks!
Another question. What does the first number mean?
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-12-15 05:33 PM - Post#197222
In response to 20Penn14
I'll pull that out in the future, but it's just the row ID that R assigns. No meaning to that whatsoever.
|
20Penn14
Senior
Posts: 364
Reg: 02-26-12
|
12-12-15 06:08 PM - Post#197223
In response to mrjames
Cool. Thanks!
|
penn nation
Professor
Posts: 21193
Reg: 12-02-04
|
Re: Ivy Game Scores Thus Far...
12-12-15 07:32 PM - Post#197227
In response to mrjames
The way to think about it is that if you played exactly like this over and over again against a nationally average opponent, you would win X% of those games.
Isn't that a bit of a misnomer? This formula doesn't care how you got to this outcome or "how you played"--rather only what the final score is.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-12-15 08:56 PM - Post#197236
In response to penn nation
Over the long run the final scores will equal how you played. But on an individual game basis, there are corrections that can be made to get a better sense of how you played than the final score can provide. Those corrections involve regressing the more chance elements of the game back to the mean while maintaining the less chance elements at their manifested levels. It worked well last year on an ad hoc basis, but it's one of the things I haven't had a chance to code into my system given my shifted priorities this season.
|
SRP
Postdoc
Posts: 4910
Reg: 02-04-06
|
12-12-15 11:26 PM - Post#197247
In response to mrjames
This is very interesting and the numbers have the right relative "feel" for me with respect to the PU games.
One message: run up the score against the bad teams if you want to look good on this index. In the Lafayette game the Tigers had their backups in early and often and still kept pulling away, so that one isn't too misleading, but I wonder if there isn't an adjustment to be made for bench usage in games where big leads are attained.
|
penn nation
Professor
Posts: 21193
Reg: 12-02-04
|
12-12-15 11:55 PM - Post#197257
In response to SRP
Or, for that matter, if you lose to a good/very good team but don't get blown out you'll pretty much get a high score on here.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-13-15 12:38 AM - Post#197264
In response to penn nation
KenPom did a great job with this topic in a blog post last year (or maybe the year before). In the face of concerns about margin of victory, and that winning by different margins beyond a certain point are meaningless, he very clearly demonstrated that incremental MOV was predictive of future outcomes even at very high MOV levels.
So, teams that win by 30 would be expected to outperform teams that only win by 20 in a similar game.
While intuitively it would seem that once a game is a blowout the ultimate margin doesn't matter, that study clearly showed that the ultimate MOV did indeed matter.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-17-15 12:38 PM - Post#197461
In response to mrjames
UPDATED Game Scores:
Two notes, as well: 1) No row IDs this time; 2) I've added a final column labeled GSNoMOV. That column shows what the game score would have been for beating that caliber of opponent at the given site but with no accounting for the margin of victory.
Year Date Venue Opp PA Ivy PF WL GS GSNoMOV
2016 2015-11-25 H Lafayette 52 princeton 104 W 99.1 7.8
2016 2015-11-22 A Southern Methodist 71 yale 69 L 97.6 NA
2016 2015-12-02 H Bryant 40 yale 79 W 97.6 3.5
2016 2015-12-05 A Kansas 75 harvard 69 L 95.4 NA
2016 2015-11-29 A Southern Methodist 77 brown 69 L 94.3 NA
2016 2015-11-25 H Bryant 45 harvard 80 W 93.2 3.5
2016 2015-11-30 H Fairleigh Dickinson 61 princeton 91 W 91.6 14.9
2016 2015-12-05 H Vermont 54 yale 72 W 90.5 33.1
2016 2015-12-12 A Lipscomb 64 princeton 78 W 89.7 51.5
2016 2015-11-20 A Northwestern 83 columbia 80 L 88.5 NA
2016 2015-11-22 H Lehigh 61 columbia 88 W 87.9 13.9
2016 2015-11-13 N Fairfield 57 yale 70 W 86.0 41.6
2016 2015-11-29 H Long Island University 56 dartmouth 79 W 84.0 13.0
2016 2015-12-02 A Bucknell 61 columbia 72 W 83.4 46.1
2016 2015-12-08 A Boston University 69 harvard 75 W 77.0 58.1
2016 2015-11-19 A Lehigh 67 yale 79 W 74.4 33.3
2016 2015-12-13 A Southern California 68 yale 56 L 74.1 NA
2016 2015-11-13 A Rider 56 princeton 64 W 72.6 39.0
2016 2015-11-16 H Holy Cross 55 brown 71 W 70.7 14.8
2016 2015-11-16 A Kansas State 81 columbia 71 L 69.6 NA
2016 2015-12-12 H NJIT 56 columbia 65 W 69.0 32.4
2016 2015-11-25 H La Salle 64 penn 80 W 64.7 15.3
2016 2015-12-09 A Illinois 69 yale 65 L 64.7 NA
2016 2015-11-25 A Duke 80 yale 61 L 64.4 NA
2016 2015-12-04 H Saint Joseph's 80 columbia 78 L 60.6 NA
2016 2015-11-14 A Providence 76 harvard 64 L 59.3 NA
2016 2015-11-24 H Wofford 59 columbia 70 W 59.1 19.8
2016 2015-11-27 A Fairfield 82 columbia 81 L 53.5 NA
2016 2015-12-06 H Delaware 69 columbia 82 W 53.5 16.2
2016 2015-11-18 H Binghamton 59 cornell 76 W 50.6 6.9
2016 2015-11-16 H Sacred Heart 77 yale 99 W 50.5 7.0
2016 2015-12-02 A Northeastern 80 harvard 71 L 50.4 NA
2016 2015-12-05 H Lafayette 67 cornell 85 W 47.9 7.3
2016 2015-12-12 A Stanford 64 dartmouth 50 L 46.0 NA
2016 2015-12-01 H Siena 80 cornell 81 W 45.9 42.9
2016 2015-11-13 A Seton Hall 84 dartmouth 67 L 44.0 NA
2016 2015-12-14 A Cal State Bakersfield 69 dartmouth 62 L 43.8 NA
2016 2015-12-08 A Saint Joseph's 62 princeton 50 L 43.7 NA
2016 2015-12-05 A Stony Brook 91 princeton 77 L 39.9 NA
2016 2015-12-09 A Maine 69 dartmouth 79 W 39.3 13.1
2016 2015-11-21 A Providence 94 brown 73 L 37.3 NA
2016 2015-12-09 H Temple 77 penn 73 L 36.0 NA
2016 2015-11-29 A Holy Cross 50 harvard 49 L 34.5 NA
2016 2015-12-07 A Georgetown 74 brown 57 L 34.0 NA
2016 2015-11-17 H Massachusetts 69 harvard 63 L 33.2 NA
2016 2015-12-09 A Manhattan 71 columbia 72 W 30.4 27.4
2016 2015-11-16 A Colgate 98 cornell 101 W 29.6 23.5
2016 2015-12-05 A Bryant 68 brown 76 W 29.0 11.1
2016 2015-12-05 H Hartford 65 dartmouth 74 W 26.5 8.6
2016 2015-11-21 H Saint Peter's 72 princeton 75 W 24.0 17.1
2016 2015-11-13 A Georgia Tech 116 cornell 81 L 22.4 NA
2016 2015-12-02 H Central Connecticut State 64 brown 82 W 21.1 2.0
2016 2015-11-29 A Massachusetts-Lowell 80 cornell 77 L 18.8 NA
2016 2015-11-15 H Central Connecticut State 61 penn 77 W 18.4 2.0
2016 2015-12-02 H Vermont 68 dartmouth 63 L 18.0 NA
2016 2015-12-14 H Robert Morris 71 columbia 78 W 14.4 6.0
2016 2015-11-21 A Washington 104 penn 67 L 14.3 NA
2016 2015-11-22 A Boston College 69 harvard 56 L 13.8 NA
2016 2015-11-17 A Delaware State 54 penn 60 W 12.7 4.5
2016 2015-11-13 A Saint Peter's 77 brown 65 L 11.2 NA
2016 2015-11-18 A Niagara 75 brown 66 L 11.2 NA
2016 2015-11-29 A Lafayette 92 penn 86 L 11.1 NA
2016 2015-12-02 H Navy 65 penn 59 L 10.9 NA
2016 2015-11-13 H Robert Morris 75 penn 76 W 6.6 5.8
2016 2015-11-28 H Longwood 70 columbia 69 L 6.2 NA
2016 2015-11-25 A Pittsburgh 93 cornell 49 L 6.1 NA
2016 2015-11-17 A Marist 73 dartmouth 63 L 6.0 NA
2016 2015-11-21 A Canisius 87 cornell 62 L 5.2 NA
2016 2015-12-05 A George Mason 63 penn 44 L 2.9 NA
2016 2015-11-29 A Albany (NY) 88 yale 54 L 1.8 NA
2016 2015-11-24 H New Hampshire 88 brown 77 L 1.7 NA
2016 2015-11-13 H Kean 62 columbia 107 W NA NA
2016 2015-11-13 H MIT 39 harvard 59 W NA NA
2016 2015-11-23 H 47 cornell 76 W NA NA
|
SRP
Postdoc
Posts: 4910
Reg: 02-04-06
|
12-17-15 10:27 PM - Post#197471
In response to mrjames
Not sure I understand that last stat. The GS stat was defined above by: "The way to think about it is that if you played exactly like this over and over again against a nationally average opponent, you would win X% of those games."
But now the GSnoMOV is defined in terms of a projected game score instead of a win probability and it often comes in at very low numbers (e.g. 7.8 for the first game on the list). I get that it is computed by ignoring the margin of victory, but what exactly is being computed?
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-17-15 11:20 PM - Post#197474
In response to SRP
Yeah, I don't like the metric much either, because I don't think what it explains is very useful, but here's how to interpret it.
The GSNoMOV can be read as the level to which you need to play to beat this team expressed as a percent of a time that you'd beat a nationally-average opponent on a neutral floor. It accomplishes this by showing what the GS would be if the game ended in a tie (thus establishing the hurdle GS you'd have to put up to win the game).
So, if the GSNoMOV is a 7.8, that means that you'd only have to play to a level that would beat a nationally-average team on a neutral floor 7.8% of the time in order to beat this team. Another way to think of it is that it can identify the most impressive wins irrespective of margin.
Not a huge fan of the metric, but it does have a useful application in identifying the most important wins in a binary sense of did you win or did you lose.
|
SRP
Postdoc
Posts: 4910
Reg: 02-04-06
|
12-18-15 02:24 AM - Post#197477
In response to mrjames
Thanks for the explanation. Much clearer.
|
mrjames
Professor
Posts: 6062
Loc: Montclair, NJ
Reg: 11-21-04
|
12-24-15 03:51 PM - Post#197928
In response to SRP
Another couple updates...
First the Average Game Scores by team and month:
MthYr Ivy mean N
2015/12 harvard 81.12000 5
2015/12 yale 80.92500 4
2015/11 princeton 73.15000 4
2015/11 yale 60.48333 6
2015/11 columbia 58.50000 6
2015/12 columbia 53.33333 6
2015/11 harvard 46.96000 5
2015/12 princeton 44.46667 6
2015/11 brown 38.21667 6
2015/12 dartmouth 35.71429 7
2015/12 cornell 34.85000 4
2015/12 brown 24.22500 4
2015/12 penn 23.35000 4
2015/11 cornell 20.40000 6
2015/11 penn 20.33333 6
Next, all games with their GS:
Date Venue Opp PA Ivy PF WL GS
2015-11-25 H Lafayette 52 princeton 104 W 98.9
2015-12-23 N Auburn 51 harvard 69 W 98.1
2015-12-02 H Bryant 40 yale 79 W 97.8
2015-11-22 A Southern Methodist 71 yale 69 L 97.4
2015-12-05 A Kansas 75 harvard 69 L 96.3
2015-11-29 A Southern Methodist 77 brown 69 L 93.9
2015-11-25 H Bryant 45 harvard 80 W 93.7
2015-11-30 H Fairleigh Dickinson 61 princeton 91 W 90.3
2015-12-05 H Vermont 54 yale 72 W 89.1
2015-11-20 A Northwestern 83 columbia 80 L 88.9
2015-12-12 A Lipscomb 64 princeton 78 W 87.7
2015-11-22 H Lehigh 61 columbia 88 W 87.2
2015-12-22 N Brigham Young 82 harvard 85 W 85.5
2015-11-29 H Long Island University 56 dartmouth 79 W 85.4
2015-12-02 A Bucknell 61 columbia 72 W 83.8
2015-11-13 N Fairfield 57 yale 70 W 81.8
2015-11-13 A Rider 56 princeton 64 W 78.1
2015-12-22 H Longwood 54 dartmouth 78 W 74.4
2015-11-19 A Lehigh 67 yale 79 W 73.0
2015-12-08 A Boston University 69 harvard 75 W 72.5
2015-12-13 A Southern California 68 yale 56 L 72.4
2015-11-16 H Holy Cross 55 brown 71 W 70.3
2015-11-16 A Kansas State 81 columbia 71 L 68.9
2015-12-12 H NJIT 56 columbia 65 W 68.8
2015-12-09 A Illinois 69 yale 65 L 64.4
2015-12-04 H Saint Joseph's 80 columbia 78 L 64.2
2015-11-25 A Duke 80 yale 61 L 63.6
2015-11-14 A Providence 76 harvard 64 L 62.3
2015-11-25 H La Salle 64 penn 80 W 57.5
2015-12-22 H Bucknell 77 princeton 89 W 55.4
2015-11-24 H Wofford 59 columbia 70 W 55.2
2015-12-02 A Northeastern 80 harvard 71 L 53.2
2015-12-06 H Delaware 69 columbia 82 W 51.9
2015-11-13 A Seton Hall 84 dartmouth 67 L 47.9
2015-12-08 A Saint Joseph's 62 princeton 50 L 47.5
2015-12-12 A Stanford 64 dartmouth 50 L 47.4
2015-12-22 A Drexel 53 penn 52 L 46.6
2015-12-01 H Siena 80 cornell 81 W 46.3
2015-11-18 H Binghamton 59 cornell 76 W 45.9
2015-11-27 A Fairfield 82 columbia 81 L 45.6
2015-11-16 H Sacred Heart 77 yale 99 W 45.5
2015-12-14 A Cal State Bakersfield 69 dartmouth 62 L 43.8
2015-12-09 A Maine 69 dartmouth 79 W 43.0
2015-12-05 H Lafayette 67 cornell 85 W 42.7
2015-11-21 A Providence 94 brown 73 L 40.2
2015-12-23 H Monmouth 78 cornell 69 L 35.5
2015-11-29 A Holy Cross 50 harvard 49 L 34.2
2015-12-05 A Stony Brook 91 princeton 77 L 34.1
2015-12-19 N Maryland 82 princeton 61 L 33.5
2015-12-09 A Manhattan 71 columbia 72 W 32.8
2015-12-09 H Temple 77 penn 73 L 31.9
2015-12-05 A Bryant 68 brown 76 W 30.2
2015-11-16 A Colgate 98 cornell 101 W 30.1
2015-12-07 A Georgetown 74 brown 57 L 27.2
2015-11-17 H Massachusetts 69 harvard 63 L 26.0
2015-11-21 H Saint Peter's 72 princeton 75 W 25.3
2015-12-02 H Central Connecticut State 64 brown 82 W 24.3
2015-12-05 H Hartford 65 dartmouth 74 W 24.1
2015-11-15 H Central Connecticut State 61 penn 77 W 21.3
2015-11-29 A Massachusetts-Lowell 80 cornell 77 L 20.5
2015-11-22 A Boston College 69 harvard 56 L 18.6
2015-12-14 H Robert Morris 71 columbia 78 W 18.5
2015-11-17 A Delaware State 54 penn 60 W 15.8
2015-12-02 H Vermont 68 dartmouth 63 L 15.8
2015-11-13 A Georgia Tech 116 cornell 81 L 15.6
2015-12-22 N Marist 84 brown 83 L 15.2
2015-12-19 A Syracuse 67 cornell 46 L 14.9
2015-11-13 A Saint Peter's 77 brown 65 L 12.0
2015-12-02 H Navy 65 penn 59 L 11.6
2015-11-18 A Niagara 75 brown 66 L 10.2
2015-11-21 A Washington 104 penn 67 L 9.5
2015-11-29 A Lafayette 92 penn 86 L 9.1
2015-11-13 H Robert Morris 75 penn 76 W 8.8
2015-12-17 H Liberty 72 princeton 77 W 8.6
2015-11-17 A Marist 73 dartmouth 63 L 7.8
2015-11-21 A Canisius 87 cornell 62 L 5.3
2015-11-28 H Longwood 70 columbia 69 L 5.2
2015-11-25 A Pittsburgh 93 cornell 49 L 5.0
2015-12-05 A George Mason 63 penn 44 L 3.3
2015-11-24 H New Hampshire 88 brown 77 L 2.7
2015-11-29 A Albany (NY) 88 yale 54 L 1.6
2015-12-19 A New Hampshire 76 dartmouth 56 L 1.5
2015-11-13 H Kean 62 columbia 107 W NA
2015-11-13 H MIT 39 harvard 59 W NA
2015-11-23 H 47 cornell 76 W NA
2015-12-19 H Ursinus 66 penn 73 W NA
|
