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

[PatsSoxKnicks]

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.

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

[DenButsu]

FWIW, hoopdata has them, but their stats only go back to 2006-07.

[Swashcuff]

Thanks so much for this post patsSOXknics....

[PatsSoxKnicks]

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.

[VCaintdead17]

patsSOXknicks....have I mentioned how much I love you?

[DenButsu]

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.

[RZZZA]

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....

[DenButsu]

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?

[PatsSoxKnicks]

patsSOXknicks....have I mentioned how much I love you?

Uh, thanks lol.

[PatsSoxKnicks]

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.

[PatsSoxKnicks]

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.

[DenButsu]

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.

[Chronz]

Can I post these on my website? Can I see your methodology or the link to that BBR page.

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.

[PatsSoxKnicks]

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.

[DenButsu]

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.