Select "print" from your browser's "File" menu.

Back to Post
Username Post: Midseason Efficiency Rankings
whitakk
Junior
Posts 214
02-12-18 12:03 PM - Post#247526    

Deep dive on each team and why this is the weirdest Ivy season in a while: http://www.nycbuckets.com/2018/02/ivy-league-we ekl...

Brown's record is more fragile than I realized, regardless of Cambridge's status. Columbia is dangerous if it can keep playing well at home. I have no idea what to make of Princeton either, but remember that "momentum" has meant nothing over the last couple seasons.
mrjames
Professor
Posts 5237
02-12-18 06:37 PM - Post#247581    

Great read.

Also important to look at can be Game Scripts. Essentially, a Game Script is the average lead that a team holds throughout the game (think of it as the "area under the curve"). Game Scripts correlate very well with which team wins the overall game (it's harder to win games you don't lead for extended periods of time, and incrementally harder to win games where you dig yourself a massive hole).

Here are the total Game Script scores through 8 Ivy games:

Harvard: 42
Penn: 24
Princeton: 20
Yale: 10
Columbia: -5
Brown: -21
Dartmouth: -32
Cornell: -38

Through 32 Ivy games, Ivy teams that win the game script battle are 24-8. The two biggest game script upsets were Harvard-Columbia (Columbia won despite a -2.8 game script) and Harvard-Cornell (Harvard won despite a -2.2 game script). No other teams have won with a game script of -2.0 or less.

Tightest game thus far has been Columbia-Brown: Brown had a +0.2 game script.
Old Bear
Postdoc
Posts 3440
02-12-18 08:13 PM - Post#247586    

Brown has played 3 OT's in 4 games.
PennFan10
PhD Student
Posts 1785
02-12-18 10:52 PM - Post#247617    

Very interesting. Is this calculated by time or by score margin? Probably both? I wouldn’t be surprised to see the top four here in the ILT.

Do you have results for last year?
mrjames
Professor
Posts 5237
02-12-18 11:06 PM - Post#247621    

It’s basically if we took the margin at all 2400 seconds and averaged them together - so it’s a combination of how long you hold a lead and the magnitude of the lead.

I can pull any of the years, and I probably should just to see how this holds. Knowing the win pcts by game script, I have to imagine that it holds across years (might be minor deviations in rank if the averages are close). But it’s worth checking for sure.
whitakk
Junior
Posts 214
02-13-18 12:05 AM - Post#247630    

Is game script more predictive than (or incrementally predictive over) efficiency margin?
mrjames
Professor
Posts 5237
02-13-18 08:50 AM - Post#247645    

I don't know if it's more predictive of the future - something to test, for sure - but it seems better descriptive of the past. What I mean by that is that the efficiency margins, as you know, are merely the final score divided by pace. But you can get to that final score in myriad different ways (some ways are good representations of teams playing hard and at their best and some, umm, not so much), and this does a better job describing how we got there in one number.

Here are the past Game Scripts by year back to 2015 (note that by 2015, there are games here and there that are missing, but nothing major):

2018
Harvard: 42
Penn: 24
Princeton: 20
Yale: 10
Columbia: -5
Brown: -21
Dartmouth: -32
Cornell: -38

2017
Princeton: 110
Harvard: 41
Penn: 14
Yale: 6
Columbia: -18
Cornell: -38
Brown: -48
Dartmouth: -67

2016
Yale: 114
Princeton: 105
Columbia: 29
Dartmouth: -16
Harvard: -17
Penn: -51
Brown: -79
Cornell: -85

2015
Harvard: 57
Yale: 49
Princeton: 36
Columbia: 15
Dartmouth: -25
Brown: -32
Cornell: -33
Penn: -67
mrjames
Professor
Posts 5237
02-13-18 09:00 AM - Post#247646    

Oh, and some fun numbers:

Since 2015 in Ivy play, the team winning the Game Script was 160-36.

Overcoming Game Scripts is pretty random. Here are some of the biggest ones over the past four seasons of Ivy play:

2016 - Harvard 76, CORNELL 74 (-13.1 Game Script)
2015 - DARTMOUTH 75, Brown 69 (-11.5)
2016 - Columbia 55, HARVARD 54 (-8.8)
2015 - BROWN 57, Cornell 56 (-6.4)
2015 - Penn 71, CORNELL 69 (-6.0)
2016 - Princeton 88, COLUMBIA 83 (-5.9)
2017 - Yale 75, BROWN 74 (-4.1)

All the rest were under four. So, as you can see, the ability to win a game script that is more than four points against you is very hard (Ivy teams are 7-111 in those games in league play over the past four seasons).
PennFan10
PhD Student
Posts 1785
02-13-18 02:58 PM - Post#247712    

So basically this says if you trail in a game for a long time it’s nit likely you will come back and win? And teams that lead most of the game win most of the time? Is there more to it than that?
mrjames
Professor
Posts 5237
02-13-18 04:36 PM - Post#247732    

No doubt - the point is not supposed to be at all surprising. Very intuitive. It’s more about the description of the game. If all I see is the final score, seeing the game script can be very useful context for understanding how the game went. I won’t know if a 10-point game was a tight one where a handful of free throws in the final minute pushed it from competitive to comfortable or if the game was healthy double-digits throughout but the losing team finished on a run in garbage time to close it to 10.

My hypothesis too (untested) is that the final stretch of the game can get unrepresentative of the normal course of the game. For instance, if a team down double-digits throughout hits a bunch of threes and gets the leading team to miss 1-and-1s and parts of double bonuses and miraculously comes all the way back, that may not be as useful a signal of that team’s true quality as the bulk of the game played more normally. So, it might be that adding game script info to the efficiency margin info (and the expected points per shot info) could enhance the predictive nature of team performance moving forward.
PennFan10
PhD Student
Posts 1785
02-13-18 08:59 PM - Post#247768    

Got it. Thanks. Very interesting.



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: 0.132 seconds.   Total Queries: 15   Zlib Compression is on.
All times are (GMT -0500) Eastern. Current time is 10:23 AM
Top