Like us on Facebook


Follow us on Twitter





Page 1 of 2 12 LastLast
Results 1 to 15 of 23
  1. #1
    Join Date
    Aug 2009
    Location
    RI
    Posts
    14,915
    vCash
    1500

    league averages of TS%, eFG% and assisted vs. unassisted shots

    For whatever reason, bball-ref doesn't have league averages for advanced stats (I actually asked them about that and they said they'd consider doing it) so I took it upon the liberty to calculate it myself. Now, I used their league average page to estimate the totals based on games played during the different seasons so the figures aren't exact but I think they're at most off by .1% points, which is no big deal.

    Anyways, thought this would be of interest to some people.

    Code:
    Season	 TS%	eFG%	Ast/FG	Unassisted
    2010-11	 0.542	0.498	0.578	0.422
    2009-10	 0.543	0.501	0.562	0.438
    2008-09	 0.545	0.499	0.566	0.434
    2007-08	 0.540	0.498	0.584	0.416
    2006-07	 0.541	0.496	0.584	0.416
    2005-06	 0.535	0.489	0.575	0.425
    2004-05	 0.530	0.482	0.593	0.407
    2003-04	 0.516	0.471	0.609	0.391
    2002-03	 0.519	0.473	0.602	0.398
    2001-02	 0.520	0.477	0.605	0.395
    2000-01	 0.518	0.473	0.611	0.389
    1999-00	 0.523	0.477	0.606	0.394
    1998-99	 0.511	0.466	0.605	0.395
    1997-98	 0.524	0.478	0.613	0.387
    1996-97	 0.536	0.493	0.609	0.391
    1995-96	 0.542	0.498	0.614	0.386
    1994-95	 0.543	0.500	0.616	0.384
    1993-94	 0.528	0.485	0.621	0.379
    1992-93	 0.536	0.491	0.607	0.393
    1991-92	 0.532	0.487	0.593	0.407
    1990-91	 0.534	0.488	0.597	0.403
    1989-90	 0.536	0.489	0.600	0.400
    1988-89	 0.537	0.489	0.602	0.398
    1987-88	 0.538	0.489	0.613	0.387
    1986-87	 0.538	0.488	0.610	0.390
    1985-86	 0.541	0.493	0.602	0.398
    1984-85	 0.543	0.497	0.600	0.400
    1983-84	 0.543	0.495	0.602	0.398
    1982-83	 0.531	0.488	0.595	0.405
    1981-82	 0.539	0.494	0.582	0.418
    1980-81	 0.535	0.489	0.593	0.407
    1979-80	 0.531	0.486	0.592	0.408
    1978-79	 0.530	0.485	0.580	0.420
    1977-78	 0.515	0.469	0.573	0.427
    1976-77	 0.511	0.465	0.558	0.442
    1975-76	 0.504	0.459	0.546	0.454
    1974-75	 0.502	0.458	0.571	0.429
    1973-74	 0.503	0.459	0.571	0.429
    1972-73	 0.498	0.456	0.570	0.430
    1971-72	 0.504	0.454	0.555	0.445
    1970-71	 0.500	0.449	0.552	0.448
    1969-70	 0.511	0.460	0.540	0.460
    1968-69	 0.491	0.441	0.529	0.471
    1967-68	 0.498	0.446	0.507	0.493
    1966-67	 0.493	0.441	0.492	0.508
    1965-66	 0.487	0.433	0.517	0.483
    1964-65	 0.479	0.426	0.494	0.506
    1963-64	 0.485	0.433	0.499	0.501
    1962-63	 0.493	0.441	0.509	0.491
    1961-62	 0.479	0.426	0.521	0.479
    1960-61	 0.469	0.415	0.533	0.467
    1959-60	 0.463	0.409	0.508	0.492
    1958-59	 0.457	0.395	0.485	0.515
    1957-58	 0.449	0.383	0.503	0.497
    1956-57	 0.449	0.379	0.526	0.474
    1955-56	 0.458	0.386	0.688	0.312
    1954-55	 0.455	0.385	0.709	0.291
    1953-54	 0.442	0.373	0.722	0.278
    1952-53	 0.445	0.370	0.737	0.263
    1951-52	 0.439	0.366	0.740	0.260
    1950-51	 0.428	0.356	0.705	0.295
    1949-50	 0.410	0.339	0.695	0.305
    1948-49	 0.390	0.327	0.641	0.359
    1947-48	 0.337	0.283	0.257	0.743
    1946-47	 0.327	0.279	0.266	0.734
    Patriots Forum HOF Class of 2011

  2. #2
    Join Date
    Oct 2005
    Location
    Shakedown Street, Japan
    Posts
    30,290
    vCash
    1500
    FWIW, hoopdata has them, but their stats only go back to 2006-07.
    I blog basketball at Roundball Mining Company///Twitter: @denbutsu

    Atheists Of PSD

  3. #3
    Join Date
    Feb 2010
    Location
    Trinidad and Tobago
    Posts
    14,914
    vCash
    1500
    Thanks so much for this post patsSOXknics....

  4. #4
    Join Date
    Aug 2009
    Location
    RI
    Posts
    14,915
    vCash
    1500
    Quote Originally Posted by DenButsu View Post
    FWIW, hoopdata has them, but their stats only go back to 2006-07.
    I know, which is exactly why I thought this would be useful. Were my numbers close to hoopdata for that time period though? I think so, which means, as I said, good enough for the seasons pre-07.
    Patriots Forum HOF Class of 2011

  5. #5
    Join Date
    Jul 2008
    Posts
    10,256
    vCash
    1500
    patsSOXknicks....have I mentioned how much I love you?
    YOU TRIED HARD MIAMI

  6. #6
    Join Date
    Oct 2005
    Location
    Shakedown Street, Japan
    Posts
    30,290
    vCash
    1500
    Quote Originally Posted by patsSOXknicks View Post
    I know, which is exactly why I thought this would be useful. Were my numbers close to hoopdata for that time period though? I think so, which means, as I said, good enough for the seasons pre-07.
    It actally didn't occur to me to check, but that's a good idea. But yeah, thanks for posting these, it could definitely come in handy.
    I blog basketball at Roundball Mining Company///Twitter: @denbutsu

    Atheists Of PSD

  7. #7
    Join Date
    Nov 2010
    Posts
    14,853
    vCash
    1500
    but I don't understand the purpose of averaging the TS% and eFG% of every player of every position in the league. It skews the numbers when you're trying to compare a TS%/eFG% of a certain player to the league average because centers will naturally have higher efficiency and FG% numbers than guards would, etc....

  8. #8
    Join Date
    Oct 2005
    Location
    Shakedown Street, Japan
    Posts
    30,290
    vCash
    1500
    Quote Originally Posted by RZZZA View Post
    but I don't understand the purpose of averaging the TS% and eFG% of every player of every position in the league. It skews the numbers when you're trying to compare a TS%/eFG% of a certain player to the league average because centers will naturally have higher efficiency and FG% numbers than guards would, etc....
    That seems to be a good point to keep in mind, but I don't think it necessarily eliminates the potential usefulness of these stats. It does, however, present a solid caveat about being careful how to use them, or perhaps also advice about how to improve their usefulness (eg. position specific averages).

    That said, anyone who just threw out TS% to compare, say, Nash and Howard, would automatically get called out by anyone who understands what the stat measures. And there could be ways to use these numbers to drive a point home.

    Say, for example, the fact that Tim Duncan had a 53.7 TS% in 2010-11, which is under the league average, let alone the average for centers (which we can safely assume should be even higher), certainly says something about the extent to which his offensive game has dropped off.

    As a single season point of reference, the 2010-11 season average TS% was 54.2% by pSk's calculation. What hoopdata has, by position, is:

    Season 54.2 (nice work, pats!)
    C 55.8
    PF 54.6
    SF 54.4
    SG 54.0
    PG 53.3

    So the total deviation (sorry if my math terms aren't precise, but I think you understand what I mean) is 2.5%, 1.6% on the upper side of the average, and 0.9% on the lower. Which actually is less deviation than I'd have expected.

    Which in turn points to another potentially interesting statistical experiment: Which stats deviate the most greatly when broken down by position from league averages?
    I blog basketball at Roundball Mining Company///Twitter: @denbutsu

    Atheists Of PSD

  9. #9
    Join Date
    Aug 2009
    Location
    RI
    Posts
    14,915
    vCash
    1500
    Quote Originally Posted by VCaintdead17 View Post
    patsSOXknicks....have I mentioned how much I love you?
    Uh, thanks lol.
    Patriots Forum HOF Class of 2011

  10. #10
    Join Date
    Aug 2009
    Location
    RI
    Posts
    14,915
    vCash
    1500
    Quote Originally Posted by RZZZA View Post
    but I don't understand the purpose of averaging the TS% and eFG% of every player of every position in the league. It skews the numbers when you're trying to compare a TS%/eFG% of a certain player to the league average because centers will naturally have higher efficiency and FG% numbers than guards would, etc....
    For 07 onwards, it obviously makes sense to use positional averages and the data you find at hoopdata. Unfortunately, pre-07, there's no one that I know of that even bothers to calculate league averages for TS%/eFG%/other advanced stats, let alone for position.

    The only way I can think of to calculate positional numbers pre-07 would be to go through the basketball-ref play index for each individual year, look at the totals, sum it up and manually calculate it for EACH position. And even then, you'd only get guard/forward/center averages. And that is a LOT of work. You're adding up the totals and calculating it for 3 different positions for each year going back 60+ years. That's a lot isn't it?

    Sadly, we'll just have to make do with what I posted. However, perhaps there is a way to estimate positional numbers based on a regression. This would take some thought and time though.

    As long as you keep in mind the caveat you mentioned, then this can still be useful. For example, if you are trying to compare 2 Centers from different eras, you may be ok. I think it would be reasonable to assume the deviation in TS% for Centers hasn't changed much from league average (Centers will always have a higher TS% than average), other then the effects of the competition, which would be something you'd want to take into account anyways. The one area where it obviously gets really tricky is comparing guards before the 3 point line vs. guards after the 3 point line. Obviously guards who have the advantage of the 3 point line are naturally going to have higher TS%s. And since there's no positional data available for TS%/other advanced stats pre-07, that would be a tough comp.

    Edit: Just thought of a way to approach that- Calculate league average 2 point % for the recent year and then use average 3 point % for that year to finish the TS% calculation. This should negate some of the effect, although a player who shoots the 3 well will still have an advantage.

    So to explain using an example, say I'm comparing Bob Cousy in 57-58 to Chris Paul this year (he shot the 3 above average so that will help him in this comp).

    The league average in 2 point % could be calculated as such:
    (FGM-3PTM) / (FGA-3PTA)
    You could do this for Paul last year, although there are sites that have 2 point % already calculated. You don't need to do this for Cousy since obvioiusly TS% in 57-58 is just based on 2 point shots anyways.

    Add in FTs and you have TS% without 3 pointers. Do that for the league average too and you've essentially removed the 3 pointer from the equation. Then you get a more fair comparison (though still flawed) between league averages.

    Anyways, there's actually quite a few ways you could approach this.
    Last edited by PatsSoxKnicks; 08-25-2011 at 12:53 PM.
    Patriots Forum HOF Class of 2011

  11. #11
    Join Date
    Aug 2009
    Location
    RI
    Posts
    14,915
    vCash
    1500
    Personally, I find the Ast/FG stat more interesting (or assisted vs. unassisted). I could also post Ast/Shot attempts too, if anyone wants. There isn't much difference though. The same basic trends apply.
    Patriots Forum HOF Class of 2011

  12. #12
    Join Date
    Oct 2005
    Location
    Shakedown Street, Japan
    Posts
    30,290
    vCash
    1500
    Season 54.2
    C 55.8
    PF 54.6
    SF 54.4
    SG 54.0
    PG 53.3

    ^The deviations I posted above...

    If we think of them instead like this:

    C +1.6
    PF +0.4
    SF +0.2
    SG -0.2
    PG -0.9


    ...and then those deviations were done for all seasons we have the data for, and then we averaged each position's deviation over those 5 or so seasons, we'd probably end up with a rough standard for what should be a fairly reasonably assumed deviation from league average by position in seasons predating 06-07.
    I blog basketball at Roundball Mining Company///Twitter: @denbutsu

    Atheists Of PSD

  13. #13
    Join Date
    Jun 2007
    Location
    LA
    Posts
    33,095
    vCash
    1894
    Can I post these on my website? Can I see your methodology or the link to that BBR page.

    Quote Originally Posted by RZZZA View Post
    but I don't understand the purpose of averaging the TS% and eFG% of every player of every position in the league. It skews the numbers when you're trying to compare a TS%/eFG% of a certain player to the league average because centers will naturally have higher efficiency and FG% numbers than guards would, etc....
    I dont think thats true, eFG% incorporates 3's and Bigs rarely shoot those, TS% is obviously about getting to the line. I dont see how the #'s are skewed in any way.
    Last edited by Chronz; 10-20-2011 at 02:59 PM.

  14. #14
    Join Date
    Aug 2009
    Location
    RI
    Posts
    14,915
    vCash
    1500
    Quote Originally Posted by Chronz View Post
    Can I post these on my website? Can I see your methodology or the link to that BBR page.
    Sure. Here's the link to the BBR page: http://www.basketball-reference.com/leagues/stats.html

    If I remember correctly, I calculated them by multiplying out each season by games played and then using the totals to calculate each stat. But I have the spreadsheet somewhere, let me look for it and see how I did it.
    Patriots Forum HOF Class of 2011

  15. #15
    Join Date
    Oct 2005
    Location
    Shakedown Street, Japan
    Posts
    30,290
    vCash
    1500
    Does the idea in my post above about average deviation by position work out right, or is it flawed somehow? It makes sense to me, but the math with this stats stuff usually tends to go a few levels deeper than I'm capable of handling.
    I blog basketball at Roundball Mining Company///Twitter: @denbutsu

    Atheists Of PSD

Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •