heypartner made it crystal clear. I have only one thing to add. the possession count from 2pointer FTs is (2pointer FTA - 2 pointer and1s)/2, and the possession count from 3pointer FTs is (3pointer FTA - 3 pointer and1s)/3. what the TS% really trying to do is to get the percentage of a player's points vs the ideal points if he makes all shots as 2 pointers for all of his possessions. 0.44 is the estimated ratio to get the possession count from FT. Every player has different number on that, actually every game should have different number even for the same player, it really depends how and where the player is fouled on the court. As I said, TS% is not a well-defined measurement in its formula, it is totally different from FG%, they are not directly related.
caneks, Thx for kind words. Don't get me wrong, I'm not against TS%, I was just pointing out how it's gives Hack-a-Shaq players too much credit for fouls when they don't have the ball. I just did that for Dwight and DJ to demonstrate a case (at the request of aelliott) for players where the stats multiplier is misapplied. I'm fine with TS%, but prefer eFG% because it doesn't come with statistical noise. eFG% is a pure stat, everything in the formula is in the boxscores. And TS% could become as perfect by just tracking And1s, 3pt FTAs and Technical Foul FTAs. With those, it is perfect. Here's the True Shooting Percentage formula with the necessary stats to eliminate the .44 multiplier. caneks, this first formula is identical to yours, just written a different way, which makes more sense to me. They should produce identical results. Base Formula TS% = Points / (2 * TSA) TS% Formula Expanded to include all necessary stats TS% = Points / ((2 * FGA) + (FTA - 3ptrFTA - And1s) + (.667 * 3ptrFTA)) TS% Formula Expanded with Technical Fouls removed my favorite TS% = (Points - TechFoulPoints) / ((2 * FGA) + (FTA - 3ptrFTA - And1s - TechFTA) + (.667 * 3ptrFTA)) Explanation Spoiler And1s TS% treats And1s as a pure bonus. They can never hurt you. Like a bonus question on a test, if you scored a 100% on the test, getting the bonus question wrong does not drop you down. So, the formula just subtracts all And1s from the total FTAs in the denominator, but leaves the score in the numerator. Treatment of FTAs from Two Pointers Take Total FTAs and subtract FTAs from 3ptrs, while also subtracting all the And1s. This leaves you with FTAs due to getting fouled on a 2ptr. (We'll discuss Technical Fouls below). These have a multiplier of .5, but that gets canceled by the formulas 2X multiplier, which specifies max potential pts of FGAs. So, you no longer "see" a multiplier, but math just wiped it out since (2 * .5 = 1). Treatment of FTAs from Three Pointers These have a lower multiplier than 2ptr FTAs, because TS% reward 3 points per shot greater than 2 points. So, if you make 2 of 3, you should get the same credit in the formula as you do for making both in a 2 out of 2 situation. Thus you see a .66667 multiplier, which is also 2 * .3333 = .6667. Technical Fouls Technical Fouls that can be shot by Coach's choice are not meant to be included in TS%. So the formula just zeros them out of the denominator and the numerator, as if they were never part of the game. (Is there any statement from the creators of the formula that they are meant as bonuses? Even for coach's arguing?? Taunting? etc). Flagrant Fouls These are, any foul were the coach cannot name the player who shoots, but rather are personal fouls whereby the player who it was committed on has to shoot the FTs. Most flagrants are largely just excessive personal fouls, so those should count as much as any FTA. Some are off-balls stuff, but again they should count certainly as much as Hack-a-Shaq fouls. bottomline: TS% can become a perfect stat, based purely on game events and not relying on league-wide average multipliers -- bleh. We have the data for it. Just NBA or someone needs to start charting And1s and Fouls on 3pt shooters. Also, I don't know if the creators of the stat wanted to treat all Technical Fouls as bonus points. I think they should be removed from the stat. If so, the last formula above does that. Side Note: Anyone actually believe Rubio should get credit for being #1 option for technical fouls, only because Kevin Martin left. If he were playing for GSW, he'd never shoot technicals with Curry and Durant around. He's an example of why I think Technical Fouls should be discarded from the formula. Let me know if anyone finds mistakes. I wish this could be a perfect stat.
i went through www.nbawowy.com data on free throws, which actually has and1's, 2pt ft's, 3 pt ft's, and tech ft's (just select a team and select "Free Throws" on the results). basically, most of the big deviations from the 0.44 number in the good direction are from guys who take lots of technicals, at least as a percentage of their overall free throws (TS% Adj w/ Techs column). if you take out the points and free throws attempts for technicals but keep accurate free throw possessions (TS% Adj w/o Techs), this is what a few guys look like: Code: Player Name TS% TS% Adj Diff TS% Adj Diff w/ Techs w/o Techs Stephen Curry 66.9 67.6 0.7 66.7 -0.2 Kevin Durant 63.4 64.0 0.6 63.1 -0.3 J.J. Redick 63.2 63.5 0.3 63.1 -0.1 Whiteside 62.9 63.0 0.1 63.0 0.1 Deandre Jordan 62.8 61.6 -1.2 61.6 -1.2 Kawhi Leonard 61.6 62.4 0.8 61.2 -0.4 Dwight Howard 60.4 60.2 -0.2 60.2 -0.2 James Harden 59.8 60.5 0.7 59.7 -0.1 Klay Thompson 59.7 59.8 0.1 59.6 -0.1 KA Towns 59.0 59.3 0.3 59.3 0.3 LeBron James 58.8 59.4 0.6 59.1 0.3 Draymond 58.7 58.7 0.0 58.7 0.0 Kyle Lowry 57.8 57.9 0.1 57.3 -0.5 Chris Paul 57.5 57.7 0.2 56.4 -1.1 Chris Bosh 57.1 57.1 0.0 57.0 -0.1 Al Horford 56.5 56.6 0.1 56.6 0.1 Eric Gordon 56.5 56.7 0.2 56.4 -0.1 Aldridge 56.5 56.3 -0.2 56.3 -0.2 Drummond 49.9 49.4 -0.5 49.4 -0.5 as you can see, a lot of guys would do better with accurate free throw possessions if you keep technicals, but once you factor out the technicals, very few players benefit. of course, outside of chris paul and deandre jordan, almost all of the changes are very small, which somehow seems to indicate that most of the guys whose numbers go up from and1's are taken back down from taking out technicals. possibly the best players get the most and1's and are also the guys chosen for technicals. all 3 big "hack-a" players did have at least some decline, though not much for dwight. i didn't notice if flagrants were tracked but that can't be a big factor. edit: just looking through guys with more than 500 fga and taking out technicals, about 3/4 of players are with 0.42 to 0.46. even including technicals 2/3 of players are within 0.42 to 0.46. given that most players don't take a ton of free throws as a percentage of overall scoring possessions, the adjustments to 0.44 shouldn't be big except in a few select cases.
That is cool. But if you have all that data, plug it into my formula(s) in my last post, and get rid of the multiplier altogether. What you show is just one of the variables (techs). I'm not really seeing the purpose of using the multiplier, while getting rid of only Techs. Techs are supposed to be built into the multiplier, so I'm not sure what it proves. And1s and 3pt FTAs are going to play a role, too. btw: with Techs, And1s and 3pt FTAs, you don't need 2pt FTAs. If you want to see perfect data TS%, you have everything needed, and the formulas above. I should just give it a shot, too.
maybe my post wasn't clear. "Adjusted" refers to using the actual 2 point foul and 3 point foul data, not the 0.44 multiplier (although i suppose using the 0.44 multiplier is the actual adjustment, but that formula came first so it gets dibs on being unadjusted). "TS% Adj w/ Techs" is the same as your second formula. "TS% Adj w/o Techs" is the same as your last formula, which i also would prefer as the best formula.
Gotcha. I think anything in the .5 difference and above is significant enough to warrant scrapping the Listed TS% formula and replacing it with the actual TS% formula with all the data. Since the data is now being collected (thx for turning me on to that site!!!), and NBA Stats could surely start getting it, the only reason not to throw out the current formula is you lose the ability to compare to historical players. But not using the actual numbers means you lose the ability to have a true stat with actual winners and real Top 10s, etc. Plus you greatly lose the ability to measure role players like Harrell accurately. I just did the Rockets. Again we see a Hack-a-Shaq player like Capella benefits by the .44 multiplier. And although Harrell doesn't have enough stats to qualify, his number proves what lack of And1s, Tech Fouls and 3ptr FTAs creates an aberration where Listed TS% is significantly too high, because your true multiplier is much higher than .44. DMo's stats also show the aberration when you have very low And1s, FTA3 and Techs. Here's the Rockets' Actual TS% and Listed TS% (Actual uses the correct formulas; Listed uses the .44 multiplier) <img src="https://scontent.fsnc1-1.fna.fbcdn.net/t31.0-8/14102918_10209025827576433_6225954301011540267_o.jpg" width=900>
Another note: The Actual TS% for the Rockets above is identical to the TS% that nbawowy.com lists with their formula. So, that convinces me Technical Fouls are not supposed to count.
I wasn't talking about describing scenarios where TS% could be inaccurate. I meant cases where a particular players TS% was obviously off. Is there some player where people feel that his shooting isn't accurately being reflected by TS%. My contention is that there are indeed contrived scenarios that would cause TS% to be inaccurate but they are so unlikely to happen that over the course of a season they become insignificant. If we can't even identify a case where it's obviously wrong then it's silly to be arguing about far fetched cases that aren't likely to happen very often where it "could" be off.
I disagree on your stated purpose of TS%. It's simply trying to adjust fg% to account for FT and the extra points on 3 pointers. It just as easily could have been called "adjusted FG%". There's lots of cases where FG% is misleading if you don't include 3's and FT. See the Dwight/Steph example that I cited earlier in the thread. TS% is trying to include all of those to give a more complete picture. Simply put, TS% reflects how efficiently a player scores. It's a much more accurate indicator of efficiency than FG%. It isn't 100% accurate but we're talking about less than 1% difference over the course of the season. As I've asked earlier in the thread, "are there any cases where anyone feels a particular player's TS% doensn't accurately reflect their efficiency?". So far I don't think that anyone has a complaint with any particular player's TS% so what does that tell you?
I said that Hack-a-Shaq players are misrepresented. DJ is way off. Capela is off enough. And also confirmed that role players are affected, too, like Harrell. It's not a big deal. It is good enough. But it is enough of a deal that you can't talk about rankings of players, like we do with other stats, because it isn't reliable to that accuracy. And since the perfect data is now available, it is an antiquated stat, which sticks around for the sole purpose (it seems) to make historical comparisons to players we don't have perfect data for. I'm on Team True TS% vs Team Estimated TS%. :grin:
cool. what's weird is that www.nbawowy.com has slightly different stats for players vs basketball-reference. for example, harden points are 2372 at nbawowy and 2376 at b-r. fga 1614 at wowy and 1617 at b-r. and chuck hayes doesn't exist at nbawowy.
The boxscore discrepancy is probably due to official scorers don't submit boxscores immediately at the end of the game, and even correct their first postings -- like who got the block, or who got the tip-in. Due to fantasy leagues, there is probably a race to scrape the data from some "official" source, which might later change. b-r.com doesn't post games right away. So, they probably wait for the official scorers. just a guess. but it doesn't explain why Hayes isn't mentioned.
http://basketball.realgm.com/wiretap/243121/George-Durant-Butler-Cousins-Lead-Team-USA-In-PER Team USA PER in 2016 Olympics 1. Paul George: 25.8 2. Kevin Durant: 23.5 3. Jimmy Butler: 23.3 4. DeMarcus Cousins: 23.1 5. DeAndre Jordan: 20.9 6. Kyrie Irving: 20.8 7. Carmelo Anthony: 19.6 8. Kyle Lowry: 19.4 9. DeMar DeRozan: 16.2 10. Draymond Green: 11.0 11. Klay Thompson: 8.3 Unlike in this year's Olympics, every player on Team USA in 2012 had a PER of at least 23.0 with four players posting marks above 35.0. Team USA PER in 2012 Olympics 1. Carmelo Anthony: 41.6 2. Kevin Love: 40.0 3. Kevin Durant: 39.4 4. LeBron James: 35.9 5. Kobe Bryant: 27.4 6. Russell Westbrook: 26.4 7. Andre Iguodala: 25.9 8. Tyson Chandler: 25.5 9. Deron Williams: 24.3 10. James Harden: 23.9 11. Chris Paul: 23.0 *Anthony Davis: 28.1 (fewer than 10 minutes per game) Team USA PER in 2008 Olympics 1. Dwyane Wade: 37.5 2. LeBron James: 28.9 3. Dwight Howard: 28.4 4. Chris Bosh: 28.4 5. Chris Paul: 22.3 6. Carmelo Anthony: 20.4 7. Kobe Bryant: 19.1 8. Tayshaun Prince: 17.1 9. Deron Williams: 13.8 10. Jason Kidd: 9.3 11. Michael Redd: 5.6 * Carlos Boozer: 15.6 (fewer than 10 minutes per game) When Team USA sent a perceived B team to the World Cup in 2014, they still had four players with PERs above 30.0. Team USA PER in 2014 FIBA World Cup 1. DeMarcus Cousins: 36.7 2. Anthony Davis: 31.6 3. Kenneth Faried: 31.0 4. James Harden: 30.2 5. Andre Drummond: 24.6 6. Kyrie Irving: 23.8 7. Klay Thompson: 22.1 8. Stephen Curry: 21.4 9. Rudy Gay: 18.7 10. DeMar DeRozan: 16.4 11. Derrick Rose: 7.3 * Mason Plumlee: 15.5 (fewer than 10 minutes per game) On an individual PER level, Team USA in 2016 most closely resembles the 2004 team that had to settle for Bronze. Team USA PER in 2004 Olympics 1. Shawn Marion: 26.1 2. Tim Duncan: 26.0 3. Lamar Odom: 23.5 4. LeBron James: 19.9 5. Allen Iverson: 17.1 6. Stephon Marbury: 14.2 7. Dwyane Wade: 13.7 8. Richard Jefferson: 8.1 * Carlos Boozer: 23.7 (fewer than 10 minutes per game) * Amar'e Stoudemire: 19.0 (fewer than 10 minutes per game)
From Jason Terry: [Thompson's] not leading s***. If he doesn't make shots, how effective is he? Go watch Klay Thompson vs. James Harden from last year and see what [LeBron] James did to him. Every time James got by him, you know who was there? Draymond Green or Andrew Bogut. ... I'm not a Klay Thompson hater. It's just when you put him up against James Harden, there's no comparison http://bleacherreport.com/articles/2665146-jason-terry-comments-on-klay-thompson-james-harden-more#
Let me give you guys a hot take. Klay's defense is overrated and he can't hold Harden's jock strap offensively.
He's a solid enough defender, but it is true that it has become over rated. He's not a stopper. He has a good frame and good hands. Solid defender, but he's not shutting down players. Guys like Khris Middleton are just as good defensively, but Khris is never called one of the best two way players in the league. Media hype.
If only J A lot of Klay Thompson's reputation stems from being apart of the Warriors at the right time and providing the skills that they need. That doesn't mean he's a role player, because I think he's a legit Top 25-ish player, but the amount of bending over backwards to make him out to be the best SG in the league remains silly.
