View Single Post
Old 02-20-2013, 01:05 PM   #4784
KC_Connection KC_Connection is offline
MVP
 
KC_Connection's Avatar
 

Join Date: Jan 2006
Casino cash: $10216731
http://kenpom.com/blog/index.php/web..._on_3p_defense

Quote:

I aggregated the numbers over a five season period from 2008 through 2012. (Why five seasons? It’s long enough to filter out season-to-season noise, but short enough to maintain some continuity in coaches and players on each team.) So, for instance, Syracuse’s opponents attempted 4,164 three-pointers over that time. Based on their opponents’ 3P% from each season, their opponents would have been expected to make 1,444 of those attempts. The Orange’s opponents actually made 1,316, a difference of 128, or 3.06% of their total attempts. Using this method, here are the top ten 3P% defenses over the past five seasons…

Team Pct SD
Syracuse -3.06 -4.3
Arizona -3.24 -3.9
BYU -3.13 -3.8
Memphis -3.18 -3.7
Michigan St -2.92 -3.7
Louisville -2.98 -3.7
Kansas -2.82 -3.6
Florida St -2.88 -3.6
Georgetown -3.10 -3.6
Virginia Tech -2.77 -3.4

I’ve ranked the teams by standardizing their percentage differences. One reason to do this is so we can compare 3P% to 2P% defense on the same scale, which I’ll do momentarily. But it’s also marginally more scientific to rank teams by this value. Syracuse’s 3.06% difference occurred on over 1,200 more three point attempts than Georgetown’s 3.10% difference. Thus, it’s more meaningful, 4.3 standard deviations better than their opponents’ average 3P%.

While favorable luck is probably contributing to the statistical effectiveness of these defense, there’s some signal here, too. The impact of each of these defenses is real, and in the context of 3P%, it’s spectacular. If 3P% defense were truly random, we might expect one Division-I team to have an opponents’ 3P% worse than 3 SD’s from average. (If we had 347 people flip a coin a few thousand times, some would rank better at flipping heads than others. This doesn’t imply skill in flipping heads.) In this sample, there were 19 such teams. However, even the best teams are only producing about a 3% influence on opponents 3P%.

By comparison, let’s observe the best 2P% defenses over this five-season span.

Team Pct SD
Kansas -8.34 -14.2
UConn -7.58 -13.4
Kentucky -7.37 -12.8
Texas -6.03 -10.2
Florida St -6.10 -9.6
Louisville -5.73 -9.5
Memphis -5.53 -9.4
Miss St -5.38 -9.1
Gonzaga -5.11 -8.1
USC -5.18 -8.1
KU's been the best by far at 2PT defense over the last five years. Pomeroy also shows that they've been 7th best at 3PT defense over that time as well (although that's mostly a product of their 2 PT defense). His old conclusion about defenses having much more impact on 2PT% than 3PT% doesn't change in this analysis, though.

Basically, if you're a team that wants to upset KU, you'd be best served to just jack up a bunch of threes and hope for the best (or just hope they're sleepwalking like at TCU). But that shouldn't be much of a surprise to anybody here.

Last edited by KC_Connection; 02-20-2013 at 01:14 PM..
Posts: 21,318
KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.KC_Connection Forgot to Remove His Claytex and Got Toxic Shock Syndrome.
  Reply With Quote