Looked up the formula for TS% to remind myself: True Shooting Percentage; the formula is PTS / (2 * TSA). True shooting percentage is a measure of shooting efficiency that takes into account field goals, 3-point field goals, and free throws. True Shooting Attempts; the formula is FGA + 0.44 * FTA. Why the heck do they need the complexity? What does this tell you that Points Per Shot wouldn't tell you? Why would they invent this formula when a more basic one was staring them in the face? TS% also devalues And 1's compared to Points Per Shot, which I think is a flaw. EDIT: Thinking about it, I assume they are trying to factor in wasted possessions a Dwight Howard would cause with missed free throws. I don't like it though, because Steph Curry's made 3 pointer will have a better impact on TS% than Harden's drive where he gets the two + 1. Most of the time, I'd prefer Points Per Shot.
Points per shot attempt neglects possessions used on free throws while true shooting percentage accounts for those with very little error.
OP. PPS gives extra value to getting fouled, wrt player rankings, than TS% does. Both formulas have foul points counted in numerator, but TS% also puts FTAs in denominator. ESPN's formulas TS% = PTS / ((FGA + 0.44 × FTA) * 2) PPS = PTS / FGA It's up for debate whether you'd consider one a better efficiency rating than the other, depending upon your definition of efficiency. It does include credit for FTs
Points per shot is (Total Points)/(Field Goal Attempts), it doesn't include freethrows. http://www.espn.com/nba/statistics/player/_/stat/field-goals/sort/scoringEfficiency You can see Harden scores more points per shot than Durant, but in reality he is a lot less efficient(uses more possesions).
Total points does include free throws. That's why the FT leader is higher than Durant. Do the math on that list. They are not subtracting out points due to FTs So yeah, whether you use the .44xFTA part or not will hurt the players who get fouled a lot, in a way. Note, NBA.com uses exact data, so they know when FTs are due to a shot or not.
OK, you're right. They don't show pps anymore (or maybe never did). My point was just that they now know in their data what free throws are not due to shot attempts (ie non-shooting fouls and technicals), hence they don't really need that .44 multiplier like other sites do (versus .5). It still remains that both TS% and PPS include free throw points. PPS gives extra value to getting fouled than TS% does, wrt player rankings. I'm not advocating one over the other. They serve two different purposes.
I don't agree with your argument. Ideally, you want exact points per scoring attempt, where scoring attempts accounts for and1's and shooting threes and technicals, etc. It turns out that fga + 0.44*fta is a very good approximation of this for most players. Straight points per FGA is far too crude. It's not a fair way of comparing different player's actual efficiency.
Depends what your definition of efficiency is. In Morey terms, getting fouled is more efficient that shooting, right? Let's not get into a theoretical discussion of whether defenses are saving pts by fouling drives/layups/alley-oops/shots at the rim. In pure math terms, free throws score more points than any other shot, right? And since PPS gives more value to points via free throws, by not having FTAs in the denominator, then, under one definition of efficiency, it's better, no? I think it is more that they serve two different purposes, than one is better than the other.
Getting fouled is more efficient than shooting only because FT% is usually much higher than FG%. How much higher varies from player to player. For a guy like Harden, it is more efficient to shoot FTs. For a guy like Dwight, it is not, not much anyway. PPS does not account for FT efficiency. In other words, it does not punish poor FT shooting.
It does, because you don't score as much points. All other things remaining equal, the better FT shooter will have a better PPS. If I may, what you really mean to say is PPS gives FTs too much value, since they don't count in the denominator at all ... they are statistical freebies. Hence, why Durant and Curry rank higher than Harden in TS%, but Harden beats both on PPS ... despite Harden being the worst FT shooter of the three. ----------------------------------- Now, to take this difference further ORtg and PPP are the same as TS%, when you ignore rebounds and turnovers. In other words, without rebounds and TOs, the formula is points for each FGA and accounts for FTAs in the denominator the same as TS%. (ORtg is PPP times 100 possessions.) TS% = pts / ((fga + .44 x FTAs) * 2) PPP = pts / (fga + .44 x FTAs) (this is the team formula, for simplicity sake). so, PPP is TS% times two. And ORtg is TS% times two times 100 possessions. Which leads us to, is team TS% a better measure of system than team PPS ,,, when you ignore offensive boards and turnovers. One focuses more on percentages and the other on total points.
Yes, that's a fair comment. I've always maintained that team ORtg (and DRtg) is the best single statistical measurement of a team's performance. For individual performance, it's a lot more complicated.
Out of curiosity how did they test that 0.44 was a good approximation? I mean I get running a bunch of different tests, finding correlations, r-squared, etc. but what were they testing against what?
It seems everybody replied with the same conclusion stupid me should have had before I ever created the thread - Points Per Shot is dumb, because it totally ignores the fact that you often get FTA's with no FGA. Really dumb of me - I apologize for wasting your time with that. I really had tunnel vision on And 1's. And I still don't like TS% b/c of that. To me, a 2 + 1 should be valued just as much as a 3PM. What about just Points Per Possession? Is there anything like (Points + Points Created Off Assists) / Usage Or (Points + Points Created Off Assists - Points Off Turnovers) / Usage That last one would be cool to really see offensive value. Maybe ORPM does that - I don't know that formula. I like these kinds of stats when comparing Harden vs Westbrook vs any backcourt player who creates. Seems like it addresses questions like whether a player's stats are inflated b/c of their usage or how bad turnovers really are.
Yes, those seem to be the best, well the Net. But I'm not so sure team TS% is better than team PPS. Maybe someone can create a table of that for last few years, or decades. It's probabaly TS, but more about the role players getting open looks that fouled a lot. Hard for me to say if one makes Harden better than Curry or Durant or not,...at scoring.
I didn't say it was dumb. BUt I won't defend it, especially to this crowd, without running some analysis on Team PPS vs Team TS%, and then posing the question about "well, if Team PPS looks better than Team TS%, then why not ind PPS vs ind TS%. Anyhoot, I'm not going to defend it. Just saying whether you give FTs more value in a formula or not could serve different purposes....one not necessarily being better than the other. PPP is ORtg....as they must get Off Rebounds and TOVs into the calculation to fairly estimate possessions. The only difference between PPP and ORtg is ORrtg is per 100 possessions, so they multiple by 100, which is a better looking number for Teams, as Pace historically hovers in the 90s...and toiday, closer to 100. So, instead of Harden having a 1.13 right now, he has a 113. They do use Assists somehow in Ind ORtg. I think you might be looking for Ind ORtg. Although there are some points produced stuff out there....NBA.com has them....but don't seem to publish the formula. As for ORPM, anything with PM in it, I highly question....especially for role players. They are just trying to repackage +/- as sexy as possible, and call it a new stat. And the DRPM is laughable. Likewise, I'm also skeptical of individual ORtg and DRtg. But, fwiw, all of that stuff seems fine when comparing a player to himself, or players on the same team to one another.
Like HP said, team PPP is just ORtg. As a team stat, it pretty much includes everything. From the team's perspective, offensively you have only one goal: score as many points as you can in each possession. In that sense, PPP pretty much IS the definition of team offense. The only caveat is how you define a "possession." Basketball's possession is more fluid than something like football's or baseball's. Individual offense, however, is not that simple. There are too much interconnectedness between the actions of all 10 players on the floor. There is just no simple way to separate them into a neat and clean measurement on how much a player actually contributes to his team's offense.
They aren't running statistical tests/models. It's more simple and accurate than that. They are using PBP data to find the exact number of technicals and AND1s throughout the league. They can scrap the Basketball-Reference PBP data, or hell, they could just ask Elias Sports Bureau to find the exact numbers. There is a lot of data recorded per game that is not in the boxscore. It's in the official scorers' sheets.
Ok. Still doesn't quite answer how they concluded 0.44. When they are actually tracking the PBP data, what are they defining as a possession? Obviously a FGA = 1 possession, hence its obvious use in the formula. Are they saying an AND-1 FG attempt is also basically just one possession? As opposed to the 1 FGA plus some % for the FT? Is a technical then half a possession? And the logical extension of this is, with the data now available, why not just use that actual possession count, instead of the formula approximation with 0.44 as the multiplier? [I assume because the data is available, but not so readily available, and would make comparison purposes historically very difficult.... still they can start tracking a Points Per ACTUAL Possession stat from now forward]
TS% is not trying to measure possessions. It measure shots against their expected value of 2 pts. And it is trying to be a stat that rewards players who draw fouls, by giving FTs a bonus. To that end, it's trying to be different than eFG%...not any ppp type stat. In going through the thinking, I now see why you are asking "Why .44 vs another number?" I've always been told so missing bonuses like AND1s and technicals aren't penalized. Let me go through a line of thinking with how the formula developed, to get to the bottom, where I have a question, too. Excuse me if you know this. The Formula at its bare bones Primarily, the formula establishes 2 pts as the expected outcome of a FGA. The natural forumal is: TS% = PTS / ((FGA + .5 x FTA) x 2) If FGs were awarded 3 points, and shooting foul FTs were shot in threes, then the formula would be: TS% = PTS / ((FGA + .33 x FTA) x 3) All the formula is doing is establishing what a FGA normally scores. It scores 2 pts, while each FT is 1 point. The end result is a: a FGM scores 100% a FTM scores 100% a 3FM scores 150% (as it should) OK, but why give FTs an added bonus? Obviously, if you lower the .5 multiplier, then you are giving all FTs an added bonus. each FT scores 114% vs 100% So, the .44 clearly gives all free throws a bonus score...above and beyond what they normally do by themselves due to FTs being the easiest shot in the game. I always thought they generated the .44 by treating AND1 and Technicals as different, so they count all those using PBP data, and adjust accordingly. It's not about counting possessions. It's about rewarding FTs more than they normally are. Bottomline: Why .44 vs the more normal .5? You're right. I see why you are asking, "Why .44?" What efficiency measure does it add that FTs don't already provide. Why give FTs extra value over a make = 100%? And why .44? I'm assuming .44 comes from looking at all AND1s and Techs and adjusting value of FTs so that missing them doesn't hurt your score at all, since the stat. Is orimarily measuring shooting plus fouls in the act
