This is an interesting idea - I may try and poke around with this. Essentially a +/- stat that tries to isolate when the opposition takes its shot.Brickowski said:
Avery Bradley re-signs for 4 years/$32 million
- Thread starter RedOctober3829
- Start date
Even that will understate the impact---because the number of sets you can run when you get into the offensive zone with (say) 18 seconds left vs 12-14 is much more robust (so, the number of seconds left is only part of it---it also matters how many things you can try before you get to the end of the clock). I think if you're going to look at it you'd also have to consider things like the value of additional passes to generating points, since the benefit of slowing the ball down is you reduce the number of passes the offense has time to make.
There just is a lot more to evaluating defense than shooting percentage against. That noted, it's another way Bradley is 1) relatively unique and 2) more valuable against ballhandlers than not. Metrics are evolving in the NBA, but there's more to it than any one or two of them, too.
There just is a lot more to evaluating defense than shooting percentage against. That noted, it's another way Bradley is 1) relatively unique and 2) more valuable against ballhandlers than not. Metrics are evolving in the NBA, but there's more to it than any one or two of them, too.
It's hard to get either excited or outraged by this deal. It is an overpay for a player who may not fit the other players on the roster. But he has a unique and valuable skill set, if everything breaks the right way the deal has a chance to be a bargain, and even if Bradley ends up not working with Smart it won't be a backbreaking contract. Using Bradley as a sixth man would be fine.
I see this much like I saw the Jeff Green contract. It likely isn't the optimal move, but there's a chance it could work well for the team and it's unlikely the contract will become a millstone.
I see this much like I saw the Jeff Green contract. It likely isn't the optimal move, but there's a chance it could work well for the team and it's unlikely the contract will become a millstone.
Brickowski said:
PedroKsBambino said:
I will ask you the same thing I asked Sprowl, why wouldnt this show up in his On/Off Opponent Points or On/Off DRTG? Now, last year Bradley caused the Celts DRTG to go from 108.8 to 106.7 when he stepped on the court, and I cant tell you if that defensive ball pressure was responsible for 1 point of that or 0.2 points of that. But the effect of that is in there unless there is no effect and then whats the point?
The only effect I can think of that would result from this ball pressure that is missing from the numbers is the 'wear down' factor, where Bradley's ball pressure might wear down the opposing PG to the point that the opposing PG is less effective when Bradley is off the court but this is so minor I dont think its even worth considering.
The short answer is that there's two things to consider here.
First, the on/off data is a very noisy way to look at current impact (and it appears there is a positive one). So, with this group of players we see something there. Agree we can't tell how much and can't over-rely on it because it's such noisy data. But it tells a story that he is valuable today, which is what the data consistently says about Bradley.
Second, we need to think about how the skill would impact a more typical (read: non-awful) defense. Today, causing the offense to take an extra 5 seconds to get into their set is of limited value, since most competent offensive teams can score in less than 19 seconds against the Celts turnstile bigs. But if you imagine they build a credible interior defense, suddenly those 5 extra seconds are the difference between being able to run one action and a ball reversal for an open look and not, and that will matter.
NBA analytics are not like MLB---there's a high degree of interdependency between players and actions. We simply can't expect to isolate individual actions as cleanly and easily (especially on defense) as we can in baseball. So the desire to quickly draw conclusions about individual players in the NBA, divorced from the team context, is ill-advised. In Bradley's case, this does not necessarily make him an impact guy (not sure that he will ever be that) but it does mean we should consider the basketball impact, as well as the impact on the very limited, very noisy metrics we have to look at right now.
First, the on/off data is a very noisy way to look at current impact (and it appears there is a positive one). So, with this group of players we see something there. Agree we can't tell how much and can't over-rely on it because it's such noisy data. But it tells a story that he is valuable today, which is what the data consistently says about Bradley.
Second, we need to think about how the skill would impact a more typical (read: non-awful) defense. Today, causing the offense to take an extra 5 seconds to get into their set is of limited value, since most competent offensive teams can score in less than 19 seconds against the Celts turnstile bigs. But if you imagine they build a credible interior defense, suddenly those 5 extra seconds are the difference between being able to run one action and a ball reversal for an open look and not, and that will matter.
NBA analytics are not like MLB---there's a high degree of interdependency between players and actions. We simply can't expect to isolate individual actions as cleanly and easily (especially on defense) as we can in baseball. So the desire to quickly draw conclusions about individual players in the NBA, divorced from the team context, is ill-advised. In Bradley's case, this does not necessarily make him an impact guy (not sure that he will ever be that) but it does mean we should consider the basketball impact, as well as the impact on the very limited, very noisy metrics we have to look at right now.
I agree and disagree on some stuff:
On the bolded, there are 2 ways to take this. 1 - the numbers are so noisy that they are useless, which I dont think either of us really think is the case. 2 - There are some general trends we should expect to see in the numbers. So take last year where Bradley plays with a regular, non-spectactulor defense and he is a -2.1 On/Off in DRTG and a -1.22 in Adjusted Plus Minus and for his career his On/Off DRTG is virtually 0, it kind of indicates to me that defensively even if he is great the fact of the matter is he cant really impact a game. At first this doesnt make a lot of sense to me, but (and I'd have to do some digging to confirm) I dont think even an elite defensive SG can actually truly impact a defense on a consistent basis. Now, that doesnt mean you take it to an extreme and should be content with revolving door guards, but it would make me think long and hard about paying for defense from a guard. Bigs on the other hand, I do think can play plus defense that is worth paying for because if they can protect the rim and clog the lane then they can wipe out other players mistakes. But guards simply cant do that, its almost impossible for them to provide defensive impact beyond the defense on their man. They can get steals but its at most like 2 a game and its probably easier to just take care of the ball than trying to take chances and steal it. Thats a real long winded way of saying, if guards can have a defensive impact, I think we have enough defensive measurements today that there should be at least one we can look at to truly measure and see their impact and if we cant then we are searching for something that is so small that we cant see it.
- Agree that its noisy, take all the non-KG Celts starters during the KG years. I would completely throw away their DRTG (including On/Off) numbers because I think 80% of the impact there is playing with KG and Pierce played a very minimal role in the Celts DRTG
- Having said that, outside of extreme situations (like KG), its probably has an error rate of +/- like 2 points, ORTG & DRTG are really only going to fluctuate between the range of 100 and 110, so a 2 point fluctuation due to error represents like a 20% error rate, which seems like an acceptable error rate to still find it significant enough to warrant discussion
- The Celts defense last year was not exceptionally bad, they were 18th in DRTG for the year, it was ORTG (27th) where they were putrid
- To test your theory lets look at the 11/12 and 12/13 versions of the Celts. They were 1st and 7th in DRTG respectively. In 11/12 when Bradley was a reserve his DRTG +/- was -3.6 while he was a reserve. But in 12/13 as a starter playing most of his minutes with KG (who was a -9.0 DRTG that year) which should have been the ideal situation to utilize this ball pressure skill, Bradley's on/off DRTG was +1.3 We would expect to see Bradley at least be a positive contributor to defense in this ideal situation, but it didnt happen. And if we dont see it in an ideal situation, where would we expect to really see it?
PedroKsBambino said:
On the bolded, there are 2 ways to take this. 1 - the numbers are so noisy that they are useless, which I dont think either of us really think is the case. 2 - There are some general trends we should expect to see in the numbers. So take last year where Bradley plays with a regular, non-spectactulor defense and he is a -2.1 On/Off in DRTG and a -1.22 in Adjusted Plus Minus and for his career his On/Off DRTG is virtually 0, it kind of indicates to me that defensively even if he is great the fact of the matter is he cant really impact a game. At first this doesnt make a lot of sense to me, but (and I'd have to do some digging to confirm) I dont think even an elite defensive SG can actually truly impact a defense on a consistent basis. Now, that doesnt mean you take it to an extreme and should be content with revolving door guards, but it would make me think long and hard about paying for defense from a guard. Bigs on the other hand, I do think can play plus defense that is worth paying for because if they can protect the rim and clog the lane then they can wipe out other players mistakes. But guards simply cant do that, its almost impossible for them to provide defensive impact beyond the defense on their man. They can get steals but its at most like 2 a game and its probably easier to just take care of the ball than trying to take chances and steal it. Thats a real long winded way of saying, if guards can have a defensive impact, I think we have enough defensive measurements today that there should be at least one we can look at to truly measure and see their impact and if we cant then we are searching for something that is so small that we cant see it.
I definitely agree the impact of a big is greater defensively than that of a guard. I do not think it follows, though, that this means guards are kind of per se 'zero' either.
You are ruling out the exact impact several of us have noted---good perimeter defense reduces the chances the offense gets to run the sets and actions it wants, which I think most who come at basketball from a scouting/coaching perspective would agree is an impact from a good defensive guard. Steals, which are easy to count, definitely do not capture the full potential value. I do think PG defense is a lot more valuable than SG defense (because stopping the ball is a primary value of guard defense, and the PG has the ball far more), and in Bradley's case one of the reasons he may have significant defensive value is that he can play PGs at least part of a game at a plus or plus-plus level.
If you wanted to test value of perimeter D I think you'd need a lot more than two years of Bradley (especially given injury questions), you'd want to do something like look at on/off from 10 or so clearly 'plus' defensive guards where their primary backup was materially weaker. I don't have the data (or in this case the desire) to do the study, sadly. There's more aspects to consider, too...these things are really complicated, and it just takes a lot of data crunching to reach the kind of conclusions I read in these threads here...
You are ruling out the exact impact several of us have noted---good perimeter defense reduces the chances the offense gets to run the sets and actions it wants, which I think most who come at basketball from a scouting/coaching perspective would agree is an impact from a good defensive guard. Steals, which are easy to count, definitely do not capture the full potential value. I do think PG defense is a lot more valuable than SG defense (because stopping the ball is a primary value of guard defense, and the PG has the ball far more), and in Bradley's case one of the reasons he may have significant defensive value is that he can play PGs at least part of a game at a plus or plus-plus level.
If you wanted to test value of perimeter D I think you'd need a lot more than two years of Bradley (especially given injury questions), you'd want to do something like look at on/off from 10 or so clearly 'plus' defensive guards where their primary backup was materially weaker. I don't have the data (or in this case the desire) to do the study, sadly. There's more aspects to consider, too...these things are really complicated, and it just takes a lot of data crunching to reach the kind of conclusions I read in these threads here...
If we throw out the data for extreme situations like 'playing with KG' or 'horrific defense under Mike D'Antoni', then we are left with a sample we both feel can be analyzed even though we might disagree on the level of crunching/critical thinking that might have to go into it. But before we go further, one question that I just cant find a reasonable simple answer for is why, if we throw out extremes and use some basic intuitive analysis, can we determine who the good defensive bigs are? And we both agree that guard defense is left impactful, but there is indeed an impact there, but why cant we see that as easily as we can see the bigs in the data?
On your approach in the last paragraph about how we would do the analysis the right way, I think thats a logical way to go about it, I just have a tough time reconciling why we really need to do it and the most plausible answer for me is that this impact that we think intuitively is there is just much smaller than we think. Going back to your catcher ERA example, that was somewhat unique because the impact of the catcher, because he is involved with the strategy and 'accuracy' of pitchers, to a small degree because they frame pitches, is different than the defensive impact of a first baseman. But with guards and bigs, I feel like we are comparing bigs who are sort of like a SS with a lot of opportunity for impact plays to guards who are sort of like 3rd basemen who have less opportunities but still indeed have them.
This week I am going to try to pull a bunch of data into a database so I can write some SQL to do some more complicated analysis on a larger sample very quickly. If I can get that done then I can probably do some work on your approach and we can see what we see and have the data lead our debate which would be cool.
On your approach in the last paragraph about how we would do the analysis the right way, I think thats a logical way to go about it, I just have a tough time reconciling why we really need to do it and the most plausible answer for me is that this impact that we think intuitively is there is just much smaller than we think. Going back to your catcher ERA example, that was somewhat unique because the impact of the catcher, because he is involved with the strategy and 'accuracy' of pitchers, to a small degree because they frame pitches, is different than the defensive impact of a first baseman. But with guards and bigs, I feel like we are comparing bigs who are sort of like a SS with a lot of opportunity for impact plays to guards who are sort of like 3rd basemen who have less opportunities but still indeed have them.
This week I am going to try to pull a bunch of data into a database so I can write some SQL to do some more complicated analysis on a larger sample very quickly. If I can get that done then I can probably do some work on your approach and we can see what we see and have the data lead our debate which would be cool.
The problem with throwing out data is that substitutions aren't random. A good defensive big might be out when the other teams best scorer is getting a rest.
luckiestman said:
You really have to balance making things overly complicated and accounting for factors that are incredibly small. In baseball you typically just look at OPS or OPS+ and there are some adjustments for parks because that can have a huge impact. But there isnt an adjustment for 'ERA+ against' or something like that to account for the fact that the opposing manager is trying to get his best pitchers to face David. I think you are worry about 10.4% being incorrect because 10.15% was the correct answer but knowing something is in the 10% range is perfectly fine to make an intelligent decision with.
If someone could make any case (even just illustratively) that the impact was just 0.25% I'd agree....since no one has, it feels like a major reach to just assume 'who the other guys on the court are' matters that little seems to me. It feels more like saying platoon advantage doesn't matter, to stay with baseball comparison.wutang112878 said:
But I could be wrong, too
Give me an example of what you might consider acceptable considering the variables involved. For example if we compared the DRTG of Bass and Sully unadjusted and then somehow adjust it to account for the defensive skill each would be playing alongside. Id hate to spend the time doing it to have discount it because Sully has been asked to play out of position at C while Bass didn't have this degree of difficulty
I've been making the case that Bradley kind of is what he is at this point from a shooting perspective, and I was able to accumulate some stats for all the active players in the league and wrote a query to crunch the data.
My logic was this: Scope was player who were in the league at 21 and had a decent sample (1,000 2pt attempts before and after and 500 3pt attempts) in the 4 seasons when they were 21 to 24, and after they were 24. I then calculated the FG% for the first 4 years and compared it to the age 24+ FG% and sorted by improvement. If you look at the distribution those who improve only improved by a few percentage points but ~60% of the 2pt population (29 out of 48) actually declined, and ~60% of the 3pt population (23 out of 38) actually declined.
So chances are there is a more than likely chance that Bradley actually regress as opposed to improving over the next 4 years.
If anyone has any other ideas on how to look at this or slice it, if its do-able I will take a shot.
Here is the data:
2 Point FG% Improvement
F4TwoPtFG = 2 Pt FGs made in First 4 years
F4TwoPTAtt = 2 Pt Atts in First 4 years
F4TwoPtFGPercent = 2 Pt FG% in First 4 years
RestCareerTwoPtFG = 2 Pt FGs made in Rest of Career
RestCareerTwoPTAtt = 2 Pt Atts in Rest of Career
RestCareerTwoPtFGPercent = 2 Pt FG% in Rest of Career
Improvement = 2 Pt FG% in Rest of Career - 2 Pt FG% in First 4 years
My logic was this: Scope was player who were in the league at 21 and had a decent sample (1,000 2pt attempts before and after and 500 3pt attempts) in the 4 seasons when they were 21 to 24, and after they were 24. I then calculated the FG% for the first 4 years and compared it to the age 24+ FG% and sorted by improvement. If you look at the distribution those who improve only improved by a few percentage points but ~60% of the 2pt population (29 out of 48) actually declined, and ~60% of the 3pt population (23 out of 38) actually declined.
So chances are there is a more than likely chance that Bradley actually regress as opposed to improving over the next 4 years.
If anyone has any other ideas on how to look at this or slice it, if its do-able I will take a shot.
Here is the data:
2 Point FG% Improvement
| Player | F4TwoPtFG | F4TwoPTAtt | F4TwoPtFGPercent | RestCareerTwoPtFG | RestCareerTwoPTAtt | RestCareerTwoPtFGPercent | Improvement |
| Raymond Felton | 1277 | 2997 | 42.6% | 1611 | 3518 | 45.8% | 3.2% |
| Ray Allen | 1401 | 3026 | 46.3% | 4590 | 9330 | 49.2% | 2.9% |
| Jason Richardson | 1706 | 3680 | 46.4% | 2832 | 5801 | 48.8% | 2.5% |
| Joe Johnson | 1584 | 3454 | 45.9% | 3192 | 6663 | 47.9% | 2.0% |
| Mike Conley | 1101 | 2392 | 46.0% | 666 | 1386 | 48.1% | 2.0% |
| Richard Hamilton | 1727 | 3840 | 45.0% | 3712 | 7918 | 46.9% | 1.9% |
| Jamal Crawford | 907 | 2077 | 43.7% | 2805 | 6168 | 45.5% | 1.8% |
| Chauncey Billups | 654 | 1561 | 41.9% | 2868 | 6565 | 43.7% | 1.8% |
| Delonte West | 828 | 1773 | 46.7% | 545 | 1126 | 48.4% | 1.7% |
| Ben Gordon | 1396 | 3181 | 43.9% | 1270 | 2792 | 45.5% | 1.6% |
| Mahmoud Abdul-Rauf | 1827 | 4016 | 45.5% | 1213 | 2588 | 46.9% | 1.4% |
| Rodney Stuckey | 1296 | 2968 | 43.7% | 820 | 1824 | 45.0% | 1.3% |
| Kenny Anderson | 1397 | 3278 | 42.6% | 2767 | 6309 | 43.9% | 1.2% |
| Mo Williams | 979 | 2161 | 45.3% | 1872 | 4048 | 46.2% | 0.9% |
| Terrell Brandon | 1051 | 2314 | 45.4% | 2663 | 5756 | 46.3% | 0.8% |
| Boris Diaw | 981 | 1878 | 52.2% | 2307 | 4352 | 53.0% | 0.8% |
| Jason Kidd | 1208 | 2851 | 42.4% | 3361 | 7850 | 42.8% | 0.4% |
| Kobe Bryant | 2608 | 5414 | 48.2% | 6031 | 12408 | 48.6% | 0.4% |
| Chris Paul | 1698 | 3359 | 50.6% | 1369 | 2704 | 50.6% | 0.1% |
| O.J. Mayo | 1231 | 2677 | 46.0% | 459 | 1000 | 45.9% | -0.1% |
| Andre Iguodala | 1386 | 2740 | 50.6% | 1815 | 3595 | 50.5% | -0.1% |
| Stephon Marbury | 2134 | 4563 | 46.8% | 2796 | 5995 | 46.6% | -0.1% |
| Baron Davis | 1313 | 2881 | 45.6% | 2467 | 5441 | 45.3% | -0.2% |
| Nate Robinson | 874 | 1862 | 46.9% | 929 | 1997 | 46.5% | -0.4% |
| Allen Iverson | 2066 | 4550 | 45.4% | 6308 | 14115 | 44.7% | -0.7% |
| Tony Parker | 2024 | 3867 | 52.3% | 3326 | 6459 | 51.5% | -0.8% |
| Deron Williams | 1536 | 3079 | 49.9% | 1870 | 3854 | 48.5% | -1.4% |
| Louis Williams | 1020 | 2236 | 45.6% | 468 | 1062 | 44.1% | -1.5% |
| Quentin Richardson | 895 | 2066 | 43.3% | 679 | 1632 | 41.6% | -1.7% |
| Tracy McGrady | 2476 | 5279 | 46.9% | 2461 | 5447 | 45.2% | -1.7% |
| Larry Hughes | 1796 | 4153 | 43.2% | 2167 | 5229 | 41.4% | -1.8% |
| Gilbert Arenas | 1631 | 3537 | 46.1% | 1049 | 2372 | 44.2% | -1.9% |
| Mike Bibby | 1467 | 3101 | 47.3% | 2431 | 5364 | 45.3% | -2.0% |
| Leandro Barbosa | 769 | 1519 | 50.6% | 1443 | 2973 | 48.5% | -2.1% |
| DeShawn Stevenson | 1330 | 2997 | 44.4% | 571 | 1352 | 42.2% | -2.1% |
| Jerry Stackhouse | 1690 | 3687 | 45.8% | 3085 | 7073 | 43.6% | -2.2% |
| Ramon Sessions | 998 | 2152 | 46.4% | 1231 | 2795 | 44.0% | -2.3% |
| Mickael Pietrus | 514 | 1020 | 50.4% | 490 | 1020 | 48.0% | -2.4% |
| Kyle Lowry | 911 | 1918 | 47.5% | 547 | 1226 | 44.6% | -2.9% |
| Ron Mercer | 2188 | 4933 | 44.4% | 980 | 2384 | 41.1% | -3.2% |
| Trevor Ariza | 799 | 1597 | 50.0% | 681 | 1460 | 46.6% | -3.4% |
| Tim Hardaway | 1247 | 2529 | 49.3% | 3399 | 7410 | 45.9% | -3.4% |
| Devin Harris | 1106 | 2214 | 50.0% | 1603 | 3447 | 46.5% | -3.5% |
| Monta Ellis | 1821 | 3624 | 50.2% | 2375 | 5087 | 46.7% | -3.6% |
| Rajon Rondo | 1464 | 2856 | 51.3% | 574 | 1205 | 47.6% | -3.6% |
| J.R. Smith | 803 | 1609 | 49.9% | 894 | 1966 | 45.5% | -4.4% |
| Kevin Martin | 924 | 1841 | 50.2% | 1702 | 3753 | 45.4% | -4.8% |
| Ronnie Brewer | 1246 | 2307 | 54.0% | 468 | 1034 | 45.3% | -8.7% |
F4TwoPtFG = 2 Pt FGs made in First 4 years
F4TwoPTAtt = 2 Pt Atts in First 4 years
F4TwoPtFGPercent = 2 Pt FG% in First 4 years
RestCareerTwoPtFG = 2 Pt FGs made in Rest of Career
RestCareerTwoPTAtt = 2 Pt Atts in Rest of Career
RestCareerTwoPtFGPercent = 2 Pt FG% in Rest of Career
Improvement = 2 Pt FG% in Rest of Career - 2 Pt FG% in First 4 years
3 Point FG% Improvement:
| Player | F4ThreePtFG | F4ThreePTAtt | F4ThreePtFGPercent | RestCareerThreePtFG | RestCareerThreePTAtt | RestCareerThreePtFGPercent | Improvement |
| Kyle Lowry | 252 | 815 | 30.9% | 370 | 990 | 37.4% | 6.5% |
| Chauncey Billups | 379 | 1120 | 33.8% | 1866 | 4690 | 39.8% | 5.9% |
| Mahmoud Abdul-Rauf | 167 | 524 | 31.9% | 307 | 815 | 37.7% | 5.8% |
| C.J. Miles | 292 | 885 | 33.0% | 209 | 539 | 38.8% | 5.8% |
| Jason Richardson | 407 | 1223 | 33.3% | 1488 | 3829 | 38.9% | 5.6% |
| Trevor Ariza | 207 | 641 | 32.3% | 367 | 1009 | 36.4% | 4.1% |
| Mo Williams | 194 | 557 | 34.8% | 824 | 2128 | 38.7% | 3.9% |
| Jerry Stackhouse | 324 | 1150 | 28.2% | 711 | 2247 | 31.6% | 3.5% |
| Jason Kidd | 398 | 1216 | 32.7% | 1770 | 4962 | 35.7% | 2.9% |
| Raymond Felton | 315 | 990 | 31.8% | 509 | 1470 | 34.6% | 2.8% |
| Louis Williams | 276 | 832 | 33.2% | 234 | 656 | 35.7% | 2.5% |
| Nate Robinson | 363 | 1036 | 35.0% | 569 | 1529 | 37.2% | 2.2% |
| Stephon Marbury | 379 | 1210 | 31.3% | 568 | 1717 | 33.1% | 1.8% |
| O.J. Mayo | 477 | 1272 | 37.5% | 226 | 576 | 39.2% | 1.7% |
| D.J. Augustin | 350 | 936 | 37.4% | 333 | 854 | 39.0% | 1.6% |
| Ray Allen | 497 | 1281 | 38.8% | 2677 | 6681 | 40.1% | 1.3% |
| Mike Bibby | 292 | 772 | 37.8% | 1456 | 3749 | 38.8% | 1.0% |
| Mickael Pietrus | 251 | 716 | 35.1% | 513 | 1430 | 35.9% | 0.8% |
| Jamal Crawford | 462 | 1333 | 34.7% | 1336 | 3775 | 35.4% | 0.7% |
| Kobe Bryant | 264 | 800 | 33.0% | 1223 | 3638 | 33.6% | 0.6% |
| Kevin Martin | 305 | 803 | 38.0% | 782 | 2049 | 38.2% | 0.2% |
| Quentin Richardson | 540 | 1519 | 35.5% | 577 | 1618 | 35.7% | 0.1% |
| Andre Iguodala | 252 | 762 | 33.1% | 464 | 1400 | 33.1% | 0.1% |
| Tim Hardaway | 250 | 694 | 36.0% | 1683 | 4742 | 35.5% | -0.5% |
| Deron Williams | 326 | 910 | 35.8% | 688 | 1950 | 35.3% | -0.5% |
| J.R. Smith | 644 | 1692 | 38.1% | 535 | 1427 | 37.5% | -0.6% |
| Chris Paul | 258 | 695 | 37.1% | 304 | 840 | 36.2% | -0.9% |
| Devin Harris | 175 | 536 | 32.6% | 425 | 1341 | 31.7% | -1.0% |
| Joe Johnson | 463 | 1207 | 38.4% | 1090 | 2941 | 37.1% | -1.3% |
| Baron Davis | 541 | 1617 | 33.5% | 965 | 3024 | 31.9% | -1.5% |
| Allen Iverson | 372 | 1150 | 32.3% | 786 | 2554 | 30.8% | -1.6% |
| Stephen Curry | 644 | 1443 | 44.6% | 261 | 615 | 42.4% | -2.2% |
| Sasha Vujacic | 318 | 826 | 38.5% | 220 | 607 | 36.2% | -2.3% |
| Mike Conley | 310 | 805 | 38.5% | 211 | 584 | 36.1% | -2.4% |
| Ben Gordon | 597 | 1434 | 41.6% | 539 | 1390 | 38.8% | -2.9% |
| Tracy McGrady | 509 | 1410 | 36.1% | 548 | 1711 | 32.0% | -4.1% |
| Gilbert Arenas | 638 | 1748 | 36.5% | 499 | 1543 | 32.3% | -4.2% |
| Leandro Barbosa | 411 | 983 | 41.8% | 525 | 1423 | 36.9% | -4.9% |
wutang112878 said:
Thanks for doing this. Interesting data.
Would it be possible to compare 21-24 seasons with 25-28 seasons as opposed to rest of the career? For some of these players, I would suspect their end of their careers might be so bad it would impact the numbers.
TIA.
Yes excellent idea and should be real simple to do once this annoying day job stops getting in my way
wade boggs chicken dinner said:
Here is the 2 point data comparing 21 to 24 to 25 to 28, slightly different distribution:
| Player | F4TwoPtFG | F4TwoPTAtt | F4TwoPtFGPercent | RestCareerTwoPtFG | RestCareerTwoPTAtt | RestCareerTwoPtFGPercent | Improvement |
| Raymond Felton | 1277 | 2997 | 42.6% | 1433 | 3105 | 46.2% | 3.5% |
| Ben Gordon | 1396 | 3181 | 43.9% | 1044 | 2245 | 46.5% | 2.6% |
| Richard Hamilton | 1727 | 3840 | 45.0% | 2083 | 4383 | 47.5% | 2.6% |
| Jason Richardson | 1706 | 3680 | 46.4% | 1731 | 3561 | 48.6% | 2.3% |
| Ray Allen | 1401 | 3026 | 46.3% | 1820 | 3761 | 48.4% | 2.1% |
| Mike Conley | 1101 | 2392 | 46.0% | 666 | 1386 | 48.1% | 2.0% |
| Joe Johnson | 1584 | 3454 | 45.9% | 1872 | 3922 | 47.7% | 1.9% |
| Delonte West | 828 | 1773 | 46.7% | 545 | 1126 | 48.4% | 1.7% |
| Mo Williams | 979 | 2161 | 45.3% | 1338 | 2850 | 46.9% | 1.6% |
| Jason Kidd | 1208 | 2851 | 42.4% | 1269 | 2903 | 43.7% | 1.3% |
| Kenny Anderson | 1397 | 3278 | 42.6% | 1444 | 3288 | 43.9% | 1.3% |
| Rodney Stuckey | 1296 | 2968 | 43.7% | 820 | 1824 | 45.0% | 1.3% |
| Chauncey Billups | 654 | 1561 | 41.9% | 942 | 2185 | 43.1% | 1.2% |
| Mahmoud Abdul-Rauf | 1827 | 4016 | 45.5% | 1097 | 2356 | 46.6% | 1.1% |
| Terrell Brandon | 1051 | 2314 | 45.4% | 1601 | 3461 | 46.3% | 0.8% |
| Leandro Barbosa | 769 | 1519 | 50.6% | 917 | 1787 | 51.3% | 0.7% |
| Jamal Crawford | 907 | 2077 | 43.7% | 1464 | 3303 | 44.3% | 0.7% |
| Boris Diaw | 981 | 1878 | 52.2% | 1651 | 3126 | 52.8% | 0.6% |
| Baron Davis | 1313 | 2881 | 45.6% | 1423 | 3084 | 46.1% | 0.6% |
| Stephon Marbury | 2134 | 4563 | 46.8% | 2409 | 5111 | 47.1% | 0.4% |
| Chris Paul | 1698 | 3359 | 50.6% | 1369 | 2704 | 50.6% | 0.1% |
| Larry Hughes | 1796 | 4153 | 43.2% | 1201 | 2778 | 43.2% | 0.0% |
| O.J. Mayo | 1231 | 2677 | 46.0% | 459 | 1000 | 45.9% | -0.1% |
| Kobe Bryant | 2608 | 5414 | 48.2% | 2361 | 4912 | 48.1% | -0.1% |
| Nate Robinson | 874 | 1862 | 46.9% | 823 | 1768 | 46.5% | -0.4% |
| Tim Hardaway | 1247 | 2529 | 49.3% | 1289 | 2638 | 48.9% | -0.4% |
| Tony Parker | 2024 | 3867 | 52.3% | 1987 | 3856 | 51.5% | -0.8% |
| Andre Iguodala | 1386 | 2740 | 50.6% | 1352 | 2720 | 49.7% | -0.9% |
| Gilbert Arenas | 1631 | 3537 | 46.1% | 692 | 1533 | 45.1% | -1.0% |
| Mike Bibby | 1467 | 3101 | 47.3% | 1511 | 3262 | 46.3% | -1.0% |
| Tracy McGrady | 2476 | 5279 | 46.9% | 1884 | 4113 | 45.8% | -1.1% |
| Louis Williams | 1020 | 2236 | 45.6% | 468 | 1062 | 44.1% | -1.5% |
| Deron Williams | 1536 | 3079 | 49.9% | 1646 | 3406 | 48.3% | -1.6% |
| DeShawn Stevenson | 1330 | 2997 | 44.4% | 507 | 1191 | 42.6% | -1.8% |
| Jerry Stackhouse | 1690 | 3687 | 45.8% | 2002 | 4588 | 43.6% | -2.2% |
| Ramon Sessions | 998 | 2152 | 46.4% | 1231 | 2795 | 44.0% | -2.3% |
| Allen Iverson | 2066 | 4550 | 45.4% | 2349 | 5471 | 42.9% | -2.5% |
| Quentin Richardson | 895 | 2066 | 43.3% | 502 | 1237 | 40.6% | -2.7% |
| Kyle Lowry | 911 | 1918 | 47.5% | 547 | 1226 | 44.6% | -2.9% |
| Ron Mercer | 2188 | 4933 | 44.4% | 980 | 2384 | 41.1% | -3.2% |
| Trevor Ariza | 799 | 1597 | 50.0% | 681 | 1460 | 46.6% | -3.4% |
| Monta Ellis | 1821 | 3624 | 50.2% | 2375 | 5087 | 46.7% | -3.6% |
| Rajon Rondo | 1464 | 2856 | 51.3% | 574 | 1205 | 47.6% | -3.6% |
| Devin Harris | 1106 | 2214 | 50.0% | 1421 | 3081 | 46.1% | -3.8% |
| J.R. Smith | 803 | 1609 | 49.9% | 894 | 1966 | 45.5% | -4.4% |
| Kevin Martin | 924 | 1841 | 50.2% | 1185 | 2620 | 45.2% | -5.0% |
| Ronnie Brewer | 1246 | 2307 | 54.0% | 468 | 1034 | 45.3% | -8.7% |
Here is the 3 point data comparing 21 to 24 to 25 to 28, again with a slightly different distribution
| Player | F4ThreePtFG | F4ThreePTAtt | F4ThreePtFGPercent | RestCareerThreePtFG | RestCareerThreePTAtt | RestCareerThreePtFGPercent | Improvement |
| Kyle Lowry | 252 | 815 | 30.9% | 370 | 990 | 37.4% | 6.5% |
| Chauncey Billups | 379 | 1120 | 33.8% | 568 | 1417 | 40.1% | 6.2% |
| Mahmoud Abdul-Rauf | 167 | 524 | 31.9% | 303 | 801 | 37.8% | 6.0% |
| Jason Richardson | 407 | 1223 | 33.3% | 794 | 2027 | 39.2% | 5.9% |
| C.J. Miles | 292 | 885 | 33.0% | 209 | 539 | 38.8% | 5.8% |
| Mo Williams | 194 | 557 | 34.8% | 589 | 1510 | 39.0% | 4.2% |
| Trevor Ariza | 207 | 641 | 32.3% | 367 | 1009 | 36.4% | 4.1% |
| Raymond Felton | 315 | 990 | 31.8% | 447 | 1275 | 35.1% | 3.2% |
| Jerry Stackhouse | 324 | 1150 | 28.2% | 406 | 1306 | 31.1% | 2.9% |
| Louis Williams | 276 | 832 | 33.2% | 234 | 656 | 35.7% | 2.5% |
| Nate Robinson | 363 | 1036 | 35.0% | 511 | 1375 | 37.2% | 2.1% |
| O.J. Mayo | 477 | 1272 | 37.5% | 226 | 576 | 39.2% | 1.7% |
| D.J. Augustin | 350 | 936 | 37.4% | 333 | 854 | 39.0% | 1.6% |
| Mickael Pietrus | 251 | 716 | 35.1% | 432 | 1184 | 36.5% | 1.4% |
| Ray Allen | 497 | 1281 | 38.8% | 981 | 2439 | 40.2% | 1.4% |
| Kobe Bryant | 264 | 800 | 33.0% | 519 | 1520 | 34.1% | 1.1% |
| Stephon Marbury | 379 | 1210 | 31.3% | 411 | 1273 | 32.3% | 1.0% |
| Jamal Crawford | 462 | 1333 | 34.7% | 664 | 1897 | 35.0% | 0.3% |
| Andre Iguodala | 252 | 762 | 33.1% | 311 | 938 | 33.2% | 0.1% |
| Quentin Richardson | 540 | 1519 | 35.5% | 350 | 989 | 35.4% | -0.2% |
| Jason Kidd | 398 | 1216 | 32.7% | 287 | 885 | 32.4% | -0.3% |
| Mike Bibby | 292 | 772 | 37.8% | 644 | 1720 | 37.4% | -0.4% |
| J.R. Smith | 644 | 1692 | 38.1% | 535 | 1427 | 37.5% | -0.6% |
| Deron Williams | 326 | 910 | 35.8% | 590 | 1682 | 35.1% | -0.7% |
| Tim Hardaway | 250 | 694 | 36.0% | 397 | 1129 | 35.2% | -0.9% |
| Chris Paul | 258 | 695 | 37.1% | 304 | 840 | 36.2% | -0.9% |
| Baron Davis | 541 | 1617 | 33.5% | 596 | 1837 | 32.4% | -1.0% |
| Joe Johnson | 463 | 1207 | 38.4% | 566 | 1520 | 37.2% | -1.1% |
| Kevin Martin | 305 | 803 | 38.0% | 509 | 1381 | 36.9% | -1.1% |
| Devin Harris | 175 | 536 | 32.6% | 317 | 1010 | 31.4% | -1.3% |
| Gilbert Arenas | 638 | 1748 | 36.5% | 292 | 850 | 34.4% | -2.1% |
| Stephen Curry | 644 | 1443 | 44.6% | 261 | 615 | 42.4% | -2.2% |
| Sasha Vujacic | 318 | 826 | 38.5% | 219 | 605 | 36.2% | -2.3% |
| Mike Conley | 310 | 805 | 38.5% | 211 | 584 | 36.1% | -2.4% |
| Ben Gordon | 597 | 1434 | 41.6% | 418 | 1069 | 39.1% | -2.5% |
| Allen Iverson | 372 | 1150 | 32.3% | 317 | 1076 | 29.5% | -2.9% |
| Tracy McGrady | 509 | 1410 | 36.1% | 427 | 1345 | 31.7% | -4.4% |
| Leandro Barbosa | 411 | 983 | 41.8% | 370 | 1011 | 36.6% | -5.2% |
Another thing that will clearly impact those shooting percentage changes are usage rates and where the shots are occurring. If someone is initially shooting lots of end of the shot clock threes because they are a focal point of the offense and later in their careers develop into a pure catch and shoot corner 3 specialist (basically what happened to Ray Allen) you would expect to see their percentages increase, even assuming some platonic "shooting skill" remaining constant.
Does 2pt % include layup attempts? Because that obviously is going to affect things. For example, Rajon Rondo has clearly improved his pure shooting abilities over his career, and this is reflected when you look at his percentages on actual jump shots (>8 feet). But his 2 percentage apparently has gown down, clearly because he is going to the hoop less and when he does he isn't as effective at finishign as he used to be.
Does 2pt % include layup attempts? Because that obviously is going to affect things. For example, Rajon Rondo has clearly improved his pure shooting abilities over his career, and this is reflected when you look at his percentages on actual jump shots (>8 feet). But his 2 percentage apparently has gown down, clearly because he is going to the hoop less and when he does he isn't as effective at finishign as he used to be.
slamminsammya said:
The 2pt data is for all 2 point FGs, so yes layup attempts are included. Unfortunately I cant download the 'shooting' data from Bberf unless I pulled it player by player to differentiate.
Here is a table with the 2pt improvement, the Avg Usage the first 4 years and Avg Usage the 2nd 4 years. This is simply the 'average' of the 4 usage values for each of the years its not weighted by minutes or shots or games. So if you were a 20, 20, 20, 24 guy your you played 70% of your games in minutes that last year, your average is 21 I could write in some adjustment for that but it would take a little time.
Usage is difficult though because its a mixed bag. Anyone that gets worse should logically have a decrease in usage. There is clearly some relationship between usage and FG% improvement but they are intertwined and I'm not a smart enough at most things but most certainly as a statistician to figure out how to sort through that.
| Player | FG Improvement | F4AvgUsage | RCAvgUsage | Increase/Decrease |
| Jason Richardson | 3.6% | 24.4 | 23.2 | 1.1 |
| Raymond Felton | 3.5% | 21.5 | 20.9 | 0.6 |
| Delonte West | 3.1% | 18.1 | 18.0 | 0.1 |
| Ben Gordon | 2.6% | 28.0 | 23.6 | 4.4 |
| Richard Hamilton | 2.6% | 28.1 | 26.0 | 2.0 |
| Mike Conley | 2.0% | 18.5 | 22.8 | (4.3) |
| Joe Johnson | 1.9% | 21.1 | 26.7 | (5.6) |
| Ray Allen | 1.8% | 24.0 | 27.4 | (3.4) |
| Chauncey Billups | 1.8% | 20.2 | 22.2 | (2.0) |
| Jason Kidd | 1.5% | 18.0 | 21.7 | (3.7) |
| Rodney Stuckey | 1.3% | 24.3 | 23.0 | 1.3 |
| Mahmoud Abdul-Rauf | 1.1% | 26.1 | 24.8 | 1.3 |
| Kenny Anderson | 0.7% | 23.5 | 21.6 | 1.9 |
| Boris Diaw | 0.7% | 15.1 | 18.1 | (3.1) |
| Leandro Barbosa | 0.7% | 20.3 | 24.4 | (4.1) |
| Larry Hughes | 0.6% | 26.5 | 25.0 | 1.5 |
| Jamal Crawford | 0.6% | 23.1 | 23.2 | (0.1) |
| Stephon Marbury | 0.3% | 27.3 | 25.5 | 1.8 |
| Mo Williams | 0.3% | 22.7 | 24.0 | (1.3) |
| Nate Robinson | 0.2% | 23.3 | 24.1 | (0.8) |
| Chris Paul | 0.1% | 24.6 | 22.9 | 1.7 |
| O.J. Mayo | -0.1% | 22.7 | 22.0 | 0.8 |
| Kobe Bryant | -0.1% | 30.5 | 33.3 | (2.8) |
| Terrell Brandon | -0.1% | 22.1 | 24.1 | (2.0) |
| D.J. Augustin | -0.2% | 20.3 | 20.9 | (0.6) |
| Baron Davis | -0.3% | 24.3 | 27.8 | (3.5) |
| Tim Hardaway | -0.4% | 21.6 | 24.4 | (2.8) |
| Tony Parker | -0.8% | 25.5 | 27.8 | (2.3) |
| Andre Iguodala | -0.9% | 18.5 | 20.3 | (1.8) |
| Gilbert Arenas | -1.0% | 27.5 | 29.2 | (1.7) |
| Mike Bibby | -1.0% | 20.7 | 24.2 | (3.4) |
| Tracy McGrady | -1.1% | 32.7 | 32.9 | (0.2) |
| Deron Williams | -1.2% | 22.6 | 25.6 | (3.0) |
| Louis Williams | -1.5% | 25.2 | 24.0 | 1.3 |
| Ramon Sessions | -2.1% | 20.9 | 22.2 | (1.3) |
| Allen Iverson | -2.5% | 30.7 | 35.5 | (4.8) |
| Jerryd Bayless | -2.6% | 22.8 | 20.1 | 2.7 |
| Quentin Richardson | -2.7% | 22.2 | 17.6 | 4.6 |
| Ron Mercer | -3.0% | 24.3 | 22.1 | 2.2 |
| Jerry Stackhouse | -3.3% | 25.5 | 31.2 | (5.7) |
| Mickael Pietrus | -3.4% | 19.9 | 17.2 | 2.6 |
| DeShawn Stevenson | -3.4% | 20.1 | 12.4 | 7.7 |
| Shaun Livingston | -3.4% | 14.3 | 14.7 | (0.5) |
| Kyle Lowry | -3.4% | 18.2 | 21.7 | (3.6) |
| Rajon Rondo | -3.6% | 19.2 | 21.2 | (2.1) |
| Monta Ellis | -3.7% | 24.9 | 27.8 | (3.0) |
| Dion Glover | -4.3% | 21.4 | 21.9 | (0.5) |
| Trevor Ariza | -4.3% | 16.7 | 18.0 | (1.2) |
| J.R. Smith | -4.4% | 24.9 | 23.1 | 1.9 |
| Devin Harris | -4.5% | 21.9 | 24.7 | (2.7) |
| Sebastian Telfair | -4.8% | 20.1 | 19.2 | 0.9 |
| Kevin Martin | -4.9% | 20.7 | 25.8 | (5.1) |
| Ronnie Brewer | -9.7% | 13.8 | 8.2 | 5.6 |
I haven't heard anything about the ever popular opt-out which is very good news. Once the new TV deal kicks in Bradley could be a steal over those final two seasons assuming he remains healthy and committed to competing.
HomeRunBaker said:
It was reported when the contract was announced that there wouldn't be either team or player options. Just a straight 4-year deal.
Hope he stays healthy.