Thursday, November 27, 2014

Early Season Defensive Frauds

I am using the term "fraud" loosely here as most of these teams will be good defensively, just not as good as they currently look.  In the past 5 seasons 2 teams have managed to hold their opponents to a sub 32% mark from 3 pt land with the absolute best mark being the 11/12 Celtics with a 30.8% mark.  Five teams this season are currently on pace to beat that and a sixth (Sacramento) would set the second best season over the past 5 years if they maintain their current rate.  Those 5 teams with their 3 pt% against and current defensive rank are

Rockets  27.5%  #2
Trailblazers  29.0%  #5
Spurs  29.7%  #3
Warriors  30.2%  #1
Thunder  30.7%  #8

I think it is pretty obvious that this is early season variance and not these teams simply and suddenly (and simultaneously) learning how to dominate the 3 pt arc.  Ignoring the Thunder (for obvious reasons) here is where these teams would rank defensively last season if (all else equal) their opponents were averaging 33% (a very strong defensive number, typically #3-#5 over a season) or their rate from last year (using this years 3pt attempts).

Rockets- Current Defensive rating 98.0, opponents 3pt attempts per game 20.9
Adjusted for 33% shooting- defensive rating-  101.5- defensive ranking for that rating last season #3
Adjusted for 35.5% shooting- 103 - ranking #5

Trailblazers- 101.8- 17.3
Adjusted for 33% shooting- 103.9- #6/7
Adjusted for 35.5% - 105.2 #10

Spurs- 98.1 - 20.2
33% - 100.1 - #2
35.3%- 101.5 #3

Warriors- 97.7 - 21.5
33% - 99.5 #2
34.4%- 100.0 #2

This is a superficial analysis but as we can see only one of these teams actually looks like a fraud.  Rather than a top 5 defense Portland looks like a fringe top 10 defense when this variance is removed.  Obviously there are many other factors that could be distorting this picture that are subject to variance but I would tentatively say that the Portland hype train, much like last year, is going to slowly run out of steam as the season goes on. 

No comments:

Post a Comment