 |
APBRmetrics
The statistical revolution will not be televised.
|
| View previous topic :: View next topic |
| Author |
Message |
Ben F.
Joined: 07 Mar 2005
Posts: 391
|
 Posted: Sun Nov 27, 2005 2:00 pm Post subject: Three Pointers - Foul and Block %s |
 |
|
It seems to me that shooting fouls and blocks occur on three pointers far less than on other field goals. Does anyone know the exact percentages? On what percent of three pointers is the shooter fouled? And on what percent of attempts are they blocked?
If anyone has this data, or any raw data that just needs some analysis in Excel, I'd be interested. |
|
| Back to top |
|
 |
Ed Küpfer
Joined: 30 Dec 2004
Posts: 786
Location: Toronto
|
| Posted: Sun Nov 27, 2005 6:58 pm Post subject: Re: Three Pointers - Foul and Block %s |
|
|
| FFSBasketball wrote: | | It seems to me that shooting fouls and blocks occur on three pointers far less than on other field goals. Does anyone know the exact percentages? On what percent of three pointers is the shooter fouled? And on what percent of attempts are they blocked? |
Let's do blocks. I'll look at two different ways, one easy and one more involved.
First, the easy way. The questiong you ask — what percentage of two- and three-point attempts are blocked? — can be answered by linear regression. The response variable will be total blocks, the independant variables two- and three-point attempts (team-level stats, obviously). But instead of attempts, lets look at missed shots, since attempts include made shots, a number that's irrelevant to blocks, which are a subste of missed shots only. Arrange the stats like this:
| Code: | TEAM p2misss p3miss BLK
ATL 3061 669 484
BOS 2651 815 402
CHB 3337 561 446
NO 2936 901 452
.... and so on.
|
Using Excel, DATA ANALYSIS has a REGRESSION function that can tell us what we need to know. Use the BLK columnas your input Y range, and p2misss and p3miss as the input X range (in Excel, if you use more than one column as X like we are here, they must be in adjacent columns). Hit enter, and look at the results.
When I did this using the 2005 data, the results weren't very significant. The coefficient for 2-point misses was 0.12, with a standard error of 0.06 and p-value of 0.05. The coefficient for 3-point misses was 0.06, with a standard error of 0.06 and p-value of 0.36. Ambiguous, since the standard error for 3-pts overlaps zero — this says that we cannot be confident that the percentage of 3-pt misses blocked is different from zero.
To dig a little deeper, I downloaded the play-by-play data for the current season, up to Nov 23. From these I counted directly the blocks, 2-point misses, and 3-point misses.
| Code: | SHOT MISSED BLK BLK%
2-pt 9986 1601 16.0%
3-pt 2897 31 1.1%
|
This is consitent with the regression results, since those were so ambiguous.
Looking further, we can see that blocks are a function of both type of shot and the distance.
The ESPN pbb logs often have a description of a type of shot. I looked at all the shot attempts, and looked to see whether the following categories had any effect on blocks: layups, jumpers, and hooks. I also split the distances of the shots into five groups. The following logistic regression results show how decreasing distance to the hoop increases the probability of a block. The results are relative to three point shots — the positive coefficients indicate an increasing probability of a blocked attempt.
| Code: | Predictor Coef SECoef Z P
Constant -5.24 0.22 -24.08 0.00
upTo5 3.70 0.21 17.71 0.00
UpTo10 3.06 0.20 15.46 0.00
UpTo15 2.42 0.20 12.02 0.00
UpTo23 1.08 0.20 5.43 0.00
layup 1.58 0.12 12.96 0.00
jumper 0.72 0.12 5.89 0.00
hook -0.25 0.25 -1.00 0.32
|
Interpreting logistic regression results is more complicated than the linear regression I used at the beginning of this post, but what you need to look at is the coefficients: if they are positive, they increase the probablity of a block. You can see that the closer, the better the chances of being blocked. The type of shot seems to matter very little (the coefficients are small).
So, to answer your question, yes, 2-pointers are by far more likely to be blocked, but more specifically, shots close to the hoop are by far the most likely to be blocked.
_________________
ed |
|
| Back to top |
|
|
mtamada
Joined: 28 Jan 2005
Posts: 377
|
| Posted: Mon Nov 28, 2005 4:59 am Post subject: Re: Three Pointers - Foul and Block %s |
|
|
| Ed Küpfer wrote: | | So, to answer your question, yes, 2-pointers are by far more likely to be blocked, but more specifically, shots close to the hoop are by far the most likely to be blocked. |
Nice analysis; presumably fouls would be similar but it would be great to see the numerical results. Basically: what is the TS% for 2-pointers compared to 3-pointers? I suspect that 82games.com has this info, but it doesn't appear to be publicly available -- all they do is list some leaders at shooting from various spots on the floor: http://www.82games.com/shotzones.htm.
I've long been a proponent of more 3-point shooting by most teams, but our analysis can only go so far without some key stats: fouls drawn and FTs shot by 2-point and 3-point shooters, and (this last one is a hard one) turnovers.
That is, the analysis would progress something like this, using the Sonics of 2004-05 as an example:
1. The Sonics last year shot 47.4% on 2-point attempts, and 36.5% on 3-pointers. Low FG% on 3-pointers, right?
2. Even the newest hoopstatistician recognizes the fallacy above. Those 3-pointers are worth 3 points each, so the effective FG% for the Sonics on 3-pointers was a whopping 54.8%!
3. But it gets better for the Sonics. They almost certainly retained possession on a greater percentage of 3pt attempts than on 2pt attempts. Because every missed shot gave them about a 1/3 chance to grab an offensive rebound and retain possession -- but every made shot automatically gave possession to their opponents (with the exceptions of shots at the end of the quarter and flagrant fouls, which are very rare).
The higher percentage of successful 2pt attempts meant more possessions going to the opposition. But not more points -- we've already seen the Sonics had a higher EFG% on 3-pointers.
4. BUT. The analysis above ignores FTs, and TS%. (MikeG and I had a discussion about this on APBR a few months ago.) We know from simple observation that fouls on 3pt attempts are rare. But how rare, and how common are fouls on 2pt attempts? And what was the FT% on those FTAs. Those are the pieces of information we need to calculate the TS% of 2pters vs. 3pters.
If we assume that EVERY single FTA that the Sonics shot was due to a 2pt FGA, then the Sonics TS% on 2pters was 53.3%, still lower than their 3pt EFG%. However the Sonics were an unusually good-shooting 3-point team. Most teams probably have a lower TS% on 3pters if we do this analysis for them.
On the other hand, not all FTAs can be credited to some 2pt shooter; many come on loose ball fouls and ordinary fouls on the floor commited in bonus time.
5. But it gets harder still. To really compare the effectiveness of a 3pt strategy vs a 2pt strategy, we also need to know the turnover statistics, because that's a vital part of offense also -- avoiding turnovers. I suspect that teams which are willing to settle for 3pt attempts commit fewer turnovers than teams which try to force the ball inside, but what the TO% numbers are, I do not know.
6. The final analysis requires the principles of game theory and takes us back to basketball 101. Even if we find that the Sonics had an average TS% on 3pters that was better than their 2pt TS%, we don't know that that means they should have shot moore 3-pters. A team which does nothing but shoot 3pters, like any 1-dimensional team, becomes easier to defend. Just as a football team needs at least a bit of a threat of a running game to complement its passing attack, a good NBA offense should have at least a little bit of an inside threat to complement its outside threat (and vice versa, Shaq didn't win any championships until he had the likes of Kobe, Horry, and Derek Fisher to create an outside threat to complement his inside game).
What's the optimal mix of 3pt and 2pt FGAs? A full analysis requires that we go through all 6 steps. But standard data today only permit us to go to Step 3. Play-by-play data would enable us to fully complete Step 4 and perhaps to start making headway into Step 5.
I do suspect that the answer is that most teams should be shooting more 3-pointers (once they acquire the sharpshooting personnel of course). Last year's Suns and Sonics may be the wave of the future, if and when teams start to realize this. If indeed it is correct. |
|
| Back to top |
|
|
Mike G
Joined: 14 Jan 2005
Posts: 3604
Location: Hendersonville, NC
|
| Posted: Mon Nov 28, 2005 8:42 am Post subject: |
|
|
Mike, you may very well have covered the advantages an Offense may achieve by shooting lots of 3's. But you know, those same 5 players have to play Defense. They have to box out, stop drives, grab defensive rebounds.
There are some guys who shoot 3's effectively and are also huge in-the-paint defenders. But not many. OK, I can't think of any. But if you are putting 3-4 3-point threats on the floor, and bombing away, I can counter with 3-4 rebounders. You may kill me with 3's, or I may kill you with putbacks, and limiting you to one shot.
If you have exactly 1 3-pt ace on the floor, it's pretty easy to assign one ace defender to limit his shots. He may also be a bit of a hole on defense. If you have 2 such players, etc., the mismatch escalates.
It may be that the Suns' offense worked because they had a guy (Marion) who was a big rebounder that shot 3's, the league's quickest-to-the-rim big man (Amare), and those other guys. Not everyone has those weapons; in fact, no one has, for some years. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
