Clustering Teams by PPP and Pace

Home for all your discussion of basketball statistical analysis.
Post Reply
BigLeagueInsights
Posts: 63
Joined: Thu Oct 23, 2014 9:59 pm
Location: Las Vegas
Contact:

Clustering Teams by PPP and Pace

Post by BigLeagueInsights »

I ran a k-means cluster (k=5) of all NBA teams using points per possession differential (ppp minus opponent ppp) and pace.

While the results are somewhat interesting, especially with Golden State getting their own cluster, I feel that pace dominates too much.

Anyone have any suggestions for other variables to use for clustering?

Here's what came out (apologies for the formatting, my bbcode skills are lacking):

Code: Select all

╔═══════════════╦═══════════╦═══════════╦════════════╗
║     Team      ║  Cluster  ║ PPP Diff. ║ Avg. Poss. ║
╠═══════════════╬═══════════╬═══════════╬════════════╣
║ Indiana       ║ cluster_0 ║ -0.009    ║ 92.1       ║
║ Miami         ║ cluster_0 ║ 0.018     ║ 91.7       ║
║ New York      ║ cluster_0 ║ -0.060    ║ 91.5       ║
║ Oklahoma City ║ cluster_0 ║ -0.058    ║ 92.7       ║
║ Utah          ║ cluster_0 ║ -0.037    ║ 91.6       ║
║ Golden State  ║ cluster_1 ║ 0.099     ║ 101.7      ║
║ Boston        ║ cluster_2 ║ -0.010    ║ 99.7       ║
║ Denver        ║ cluster_2 ║ -0.037    ║ 97.6       ║
║ Minnesota     ║ cluster_2 ║ -0.075    ║ 98.2       ║
║ Philadelphia  ║ cluster_2 ║ -0.160    ║ 98.9       ║
║ Phoenix       ║ cluster_2 ║ -0.005    ║ 99.8       ║
║ Sacramento    ║ cluster_2 ║ 0.004     ║ 98.5       ║
║ Dallas        ║ cluster_3 ║ 0.111     ║ 95.0       ║
║ Detroit       ║ cluster_3 ║ -0.048    ║ 93.2       ║
║ Memphis       ║ cluster_3 ║ 0.062     ║ 93.5       ║
║ San Antonio   ║ cluster_3 ║ 0.029     ║ 94.8       ║
║ Toronto       ║ cluster_3 ║ 0.100     ║ 95.0       ║
║ Atlanta       ║ cluster_4 ║ -0.008    ║ 95.8       ║
║ Brooklyn      ║ cluster_4 ║ 0.005     ║ 96.2       ║
║ Charlotte     ║ cluster_4 ║ -0.062    ║ 95.6       ║
║ Chicago       ║ cluster_4 ║ 0.048     ║ 96.2       ║
║ Cleveland     ║ cluster_4 ║ 0.022     ║ 95.7       ║
║ Houston       ║ cluster_4 ║ 0.051     ║ 95.3       ║
║ L.A. Clippers ║ cluster_4 ║ 0.019     ║ 96.6       ║
║ L.A. Lakers   ║ cluster_4 ║ -0.079    ║ 96.8       ║
║ Milwaukee     ║ cluster_4 ║ 0.000     ║ 97.1       ║
║ New Orleans   ║ cluster_4 ║ 0.056     ║ 95.6       ║
║ Orlando       ║ cluster_4 ║ -0.034    ║ 96.2       ║
║ Portland      ║ cluster_4 ║ 0.084     ║ 95.3       ║
║ Washington    ║ cluster_4 ║ 0.017     ║ 96.1       ║
╚═══════════════╩═══════════╩═══════════╩════════════╝
Crow
Posts: 10624
Joined: Thu Apr 14, 2011 11:10 pm

Re: Clustering Teams by PPP and Pace

Post by Crow »

Perhaps own and opp. PPP (as you had considered earlier) with or without pace. With pace might face some of same issues but maybe not or not as much as a three data point set?
DSMok1
Posts: 1119
Joined: Thu Apr 14, 2011 11:18 pm
Location: Maine
Contact:

Re: Clustering Teams by PPP and Pace

Post by DSMok1 »

First get the right number of decimals shown, then convert form Excel to unicode art here: http://www.sensefulsolutions.com/2010/1 ... table.html


I'd be interested in seeing clustering based on shot distance distribution (and/or shot type distribution), pace, and offensive rebounding percentage, effectively measuring offensive style.
Developer of Box Plus/Minus
APBRmetrics Forum Administrator
Twitter.com/DSMok1
BigLeagueInsights
Posts: 63
Joined: Thu Oct 23, 2014 9:59 pm
Location: Las Vegas
Contact:

Re: Clustering Teams by PPP and Pace

Post by BigLeagueInsights »

As requested. With pace included, Golden State gets its own cluster. Without pace, the Sixers are in a world of pain and get their own cluster.

Without pace, Cluster 4 looks like "contenders," while cluster 2 stands out as teams with offensive firepower. Houston and San Antonio are strangely grouped with weak teams in cluster 1.

With pace

Code: Select all

╔═══════════════╦═══════════╦══════╦══════╦════════╗
║     Team      ║  Cluster  ║ ppp  ║ oppp ║  poss  ║
╠═══════════════╬═══════════╬══════╬══════╬════════╣
║ Indiana       ║ cluster_0 ║ 0.99 ║ 1.00 ║ 92.08  ║
║ Miami         ║ cluster_0 ║ 1.06 ║ 1.04 ║ 91.73  ║
║ New York      ║ cluster_0 ║ 1.05 ║ 1.11 ║ 91.54  ║
║ Oklahoma City ║ cluster_0 ║ 0.97 ║ 1.03 ║ 92.69  ║
║ Utah          ║ cluster_0 ║ 1.07 ║ 1.10 ║ 91.58  ║
║ Golden State  ║ cluster_1 ║ 1.07 ║ 0.97 ║ 101.70 ║
║ Boston        ║ cluster_2 ║ 1.07 ║ 1.08 ║ 99.70  ║
║ Denver        ║ cluster_2 ║ 1.04 ║ 1.08 ║ 97.64  ║
║ Minnesota     ║ cluster_2 ║ 1.04 ║ 1.11 ║ 98.20  ║
║ Philadelphia  ║ cluster_2 ║ 0.90 ║ 1.06 ║ 98.91  ║
║ Phoenix       ║ cluster_2 ║ 1.04 ║ 1.05 ║ 99.75  ║
║ Sacramento    ║ cluster_2 ║ 1.04 ║ 1.04 ║ 98.45  ║
║ Dallas        ║ cluster_3 ║ 1.15 ║ 1.04 ║ 95.00  ║
║ Detroit       ║ cluster_3 ║ 0.99 ║ 1.04 ║ 93.17  ║
║ Memphis       ║ cluster_3 ║ 1.05 ║ 0.99 ║ 93.50  ║
║ San Antonio   ║ cluster_3 ║ 1.00 ║ 0.97 ║ 94.82  ║
║ Toronto       ║ cluster_3 ║ 1.11 ║ 1.01 ║ 95.00  ║
║ Atlanta       ║ cluster_4 ║ 1.07 ║ 1.08 ║ 95.80  ║
║ Brooklyn      ║ cluster_4 ║ 1.06 ║ 1.05 ║ 96.18  ║
║ Charlotte     ║ cluster_4 ║ 0.98 ║ 1.04 ║ 95.58  ║
║ Chicago       ║ cluster_4 ║ 1.07 ║ 1.02 ║ 96.18  ║
║ Cleveland     ║ cluster_4 ║ 1.10 ║ 1.08 ║ 95.70  ║
║ Houston       ║ cluster_4 ║ 1.02 ║ 0.97 ║ 95.33  ║
║ L.A. Clippers ║ cluster_4 ║ 1.06 ║ 1.05 ║ 96.60  ║
║ L.A. Lakers   ║ cluster_4 ║ 1.06 ║ 1.14 ║ 96.75  ║
║ Milwaukee     ║ cluster_4 ║ 0.99 ║ 0.99 ║ 97.08  ║
║ New Orleans   ║ cluster_4 ║ 1.09 ║ 1.04 ║ 95.60  ║
║ Orlando       ║ cluster_4 ║ 1.00 ║ 1.04 ║ 96.23  ║
║ Portland      ║ cluster_4 ║ 1.10 ║ 1.01 ║ 95.27  ║
║ Washington    ║ cluster_4 ║ 1.03 ║ 1.01 ║ 96.10  ║
╚═══════════════╩═══════════╩══════╩══════╩════════╝
Without pace:

Code: Select all

╔═══════════════╦═══════════╦══════╦══════╗
║     Team      ║  Cluster  ║ ppp  ║ oppp ║
╠═══════════════╬═══════════╬══════╬══════╣
║ Philadelphia  ║ cluster_0 ║ 0.90 ║ 1.06 ║
║ Charlotte     ║ cluster_1 ║ 0.98 ║ 1.04 ║
║ Detroit       ║ cluster_1 ║ 0.99 ║ 1.04 ║
║ Houston       ║ cluster_1 ║ 1.02 ║ 0.97 ║
║ Indiana       ║ cluster_1 ║ 0.99 ║ 1.00 ║
║ Milwaukee     ║ cluster_1 ║ 0.99 ║ 0.99 ║
║ Oklahoma City ║ cluster_1 ║ 0.97 ║ 1.03 ║
║ Orlando       ║ cluster_1 ║ 1.00 ║ 1.04 ║
║ San Antonio   ║ cluster_1 ║ 1.00 ║ 0.97 ║
║ Cleveland     ║ cluster_2 ║ 1.10 ║ 1.08 ║
║ Dallas        ║ cluster_2 ║ 1.15 ║ 1.04 ║
║ New Orleans   ║ cluster_2 ║ 1.09 ║ 1.04 ║
║ Portland      ║ cluster_2 ║ 1.10 ║ 1.01 ║
║ Toronto       ║ cluster_2 ║ 1.11 ║ 1.01 ║
║ Atlanta       ║ cluster_3 ║ 1.07 ║ 1.08 ║
║ Boston        ║ cluster_3 ║ 1.07 ║ 1.08 ║
║ Denver        ║ cluster_3 ║ 1.04 ║ 1.08 ║
║ L.A. Lakers   ║ cluster_3 ║ 1.06 ║ 1.14 ║
║ Minnesota     ║ cluster_3 ║ 1.04 ║ 1.11 ║
║ New York      ║ cluster_3 ║ 1.05 ║ 1.11 ║
║ Utah          ║ cluster_3 ║ 1.07 ║ 1.10 ║
║ Brooklyn      ║ cluster_4 ║ 1.06 ║ 1.05 ║
║ Chicago       ║ cluster_4 ║ 1.07 ║ 1.02 ║
║ Golden State  ║ cluster_4 ║ 1.07 ║ 0.97 ║
║ L.A. Clippers ║ cluster_4 ║ 1.06 ║ 1.05 ║
║ Memphis       ║ cluster_4 ║ 1.05 ║ 0.99 ║
║ Miami         ║ cluster_4 ║ 1.06 ║ 1.04 ║
║ Phoenix       ║ cluster_4 ║ 1.04 ║ 1.05 ║
║ Sacramento    ║ cluster_4 ║ 1.04 ║ 1.04 ║
║ Washington    ║ cluster_4 ║ 1.03 ║ 1.01 ║
╚═══════════════╩═══════════╩══════╩══════╝
Post Reply