What you're saying here is very clear. What's not clear is why you consider that a flaw in TS%. I mean, you keep insisting its a flaw, but I've yet to see why. If you shoot more free throws, that means you're taking more shots to get those free throws. Efficiency = points / shots attempted. Denominator is supposed to go up. If a player is attempting more 2s and 3s compared to free throws, then yeah his efficiency will drop. Why? Because 2s and 3s are typically less efficient shots. If I decide to take a higher volume of shots that are inefficient relative to shots that are more efficient, than my overall efficiency should go down. This is perfectly reasonable.
not if there is a neglible difference in results at normal sample size. you are too focused on proven yourself via small samples that you don't even see you are not comparing samples correctly...albeit them being too small and too far out on the edges of normality to matter. I mean, hell, there is a reason why the league doesn't show stats for the regular season until about a month in.
Scouting report on Novak: Run at him full speed and attempt the shot block. Do not close out because there is zero chance he will put the ball on the floor and go around you.
Allow me to quote myself in this thread: "FTA * 0.44 = approximation for number of trips to the foul line (historicaly evidence shows that to be very accurate over a significant stretch of games)" "The 0.44 factor is an approximation, but it works fairly well over a significant stretch of games. I'm not sure, over the course of a season, if your refinement using "and-1s" would make much of a difference." TS% is an approximation. I don't know how many times I've said it in this thread, but there I'll say it again. And, yes, it's understood that the less games (or shot attempts) we're dealing with, the less precise it will be. But even for a single game, even with that worse precision, TS% will still usually give you a better indication of how efficient a player was with his shot attempts than PPS. A typical stat line might be 3/6 from the field, 1/3 from three, and 3/4 from the line. Suppose the player took two trips to the line for those 4 FTA. If we define efficiency as points score per shot attempt, we can precisely calculate it here. 10 points scored on 8 shot attempts (6 FGA plus the two trips to the line). That's 1.2 points per shot. TS% is a little off and gives us 1.3 points per shot (or .65 TS%). But PPS is much further off -- it comes to 1.7 points per shot.
heypartner, I think I misunderstood your point. You wrote there are negligible differences at "normal" sample sizes. I'll assume a full season's worth of data is normal sample size. Do you mean to say that there's negligible difference between PPS and TS% over the course of a season? I'm not understanding your meaning. There is a considerable difference in results between the two over the course of a season. I can drop the 1-game examples, if you like. Edit: More on PPS vs TS%. I looked at leaders in both last season for players playing at least 1000 minutes. Let's compare Dwight Howard and Andrew Bynum, two similar players, using these two metrics (I'll drop the 0.5 factor from TS% and call it tsPPS). Dwight Howard had a 1.74 PPS (ranked first), and a 1.23 tsPPS (ranked 11th). Andrew Bynum had a 1.55 PPS (ranked 7th) and a 1.31 tsPPS (ranked first). Which is more accurate? Let's suppose that the .44 coefficient should actually be somewhere between 0.4 and 0.5 (* see below) for both these players. I can calculate a lower and upper bound for their real efficiency, which I'll call realPPS: For Howard, realPPS = [1.19, 1.27]. For Bynum, realPPS = [1.29, 1.34]. As you can see, the realPPS for Bynum should be greater than Howard's as long as the coefficient for both is somewhere between .40 and .50. Even if their coefficient were on opposite ends (unlikely, since they're so similar), Bynum would still be scoring more points per shot attempt. So there's a clear-cut case, for a large sample size, where PPS gets it wrong. And you can run through the all the cases where there's a difference in ranking in PPS and TS%, and TS% will always be more likely to be right. Simply because the formula for PPS is fatally flawed -- it completely ignores FTA attempts and therefore doesn't properly penalize players for missed freebies. And, in fact, that's why it messed up with Howard and Bynum. Howard was a much worse free throw shooter, but PPS didn't penalize him for it. * A 0.4 coefficient would suggest that about 20% of a player's FTAs were And-1 free throws. This would be abnormally high, since for leaders in And-1s as a percentage of shot attempts in 05/06, Shaq O'neal had the highest percentage of FTAs that were Ad-1 free throws, and it was only 15.4%. A 0.5 coefficient would suggest that none of the players FTAs were And-1 free throws. So, for almost all cases we should expect the coefficient to fall between these two extremes.
I agree. TS% is the best we have to measure "points per shot attempts" (although I still would like to have the And-1 data factored in), simply because it is pretty much the definition of TS%. So you are defining something by its definition. *shrug* Nothing to argue about, but not very enlightening either. Efficiency in basketball is a very complicated thing and I am not sure if we can come up with something simple enough to discuss here. All I am saying is that neither TS% nor PPS gives us the whole picture because they make certain kinds of stats more important than others. I think you said it very early on in this thread that while Novak's TS% went off the roof, it doesn't tell us what he did in between his shots. That's exactly the problem of defining efficiency just by TS%. It does not account for the volume of shots. If a player is not very involved in the offense and just take some rare open 3s, then he is not an efficient player because he is wasting one of the five spots on the floor by doing nothing much.
TS% is meant to be a single number that's an improvement on FG% and PPS for reflecting scoring efficiency. And I think it is. I don't think it makes sense for a pure efficiency stat to also incorporate volume of shots. It's possible that one player could be more "efficient" than another, and yet the other makes his team overall more efficient. That's because team efficiency requires good offensive players. But a good offensive player needs to be efficient (individually) AND capable of producing in volume. It's that individual efficiency that TS% is trying to describe. What you're suggesting is an adjusted TS% that gives extra credit for players that take a lot of shots, and removes credit from players that take less shots (to take into account how the player's role impacts his teammates). It's not a bad idea, if we're looking to describe something beyond individual efficiency and more about overall offensive value. I've seen something like that for ORtg, which is a more comprehensive efficiency statistic that also incorporates turnovers, offensive rebounding, and assists.
durvasa...first off, you can't compare Bynum to Amare. Bynum got injured just when he was showing his numbers for like only one month. The league will adjust to him. Amare has proven his numbers much longer. sorry, i wasn't able today to run what JayZ suggesting...a statistical report on the stats...haha...i got too excited walking my dogs and watching DaDakota offer beers for everyone after making a stink in the GARM. just want to continue pointing out that the FTA part or the denominator creates a drag on the formula such that players who shoot alot of FTs get less value for 2s and 3s. That's just a math fact. No, how that plays out under PPS, is not known. I'm just saying that TS% does indeed drag down 2s and 3s for high volume FT shooters...based on pure math...so to say PPS favors FT shooters...I'm not so sure. It's probably a negligible difference considering PPS rewards 2s and 3s much more than TS% does.
That's irrelevant to the point I was making. The argument still holds even if I was talking about hypothetical players (and it was Dwight Howard, not Amare). I'm not saying that Bynum is as good or better than Howard. We could just as easily call them player X and player Y, with those same stats. I'm saying that PPS doesn't properly account for the impact of FTA on a player's efficiency. I just wanted to make that point concrete with an example. I could use two other players if you like. I can take any two players, and TS% will approximate scoring efficiency (points scored per scoring attempt) better than PPS. I'm not disputing the "drag" you speak of. I just don't understand the relevance of it. It's equally true that FGA in the denominator creates a "drag" on a player's FG%. Or FTA creates a "drag" on a player's FT%. Does that mean it doesn't belong there? As I said before, unless we agree on what efficiency means we're probably not going to see eye to eye on this. By the definition I'm assuming, FTA has to be in the denominator because the more FTA a player has, the more scoring attempts he's made. And scoring efficiency is inversely proportional to scoring attempts, just a FG% is inversely proportional to FGA and FT% is inversely proportional to FTA. I guess you think of scoring efficiency for an individual player in a different way. And maybe that's why you see the FTA in the denominator as a flaw. Honestly, I don't see it.
ehh... i think basketball is more complex than the simple arithmetic used to compute most of these stats. I mean, its good to get an objective view of a player's production but i just think there's more to it. Guys like Steve Novak only make me care less about stats. I watch games.
I agree with that. Like you said, it's good to have an objective view as part of the evaluation. And, for fans like us, it's nice to be able to glance at a couple numbers to get a quick idea of how well/poorly someone is playing. It doesn't substitute actually watching the game, but it's still useful.
Why is it a problem? Unless you don't want to factor in a player's FT shooting as part of his efficiency, FTA is going to impact the numbers. The more you shoot FT relative to your FGA, the less significant your FGA is as part of your offensive production. What's wrong with that? My only beef with TS% (other than the .44 coefficient) as I pointed out to durvasa, is that it does not account for shot volume. Maybe that's why you don't like it. For example, Player X takes 6 FGs per game, and 10 FTs per game. Player Y takes 3 FGs per game, and 4 FTs per game. According to TS%, Player X's FG% has relatively less impact to his offensive game (or conversely, his FT shooting has more impact) than Player Y. That is plain true. I don't see any problem with that. BUT, it does not tell you how the two players' TOTAL offensive games compare. Being able to get more than double of the other player's shots should count for something in a guy's efficiency, imo.
My rationale behind free throws (unrelated to heypartner's thoughts or not) is they're like field goal kickers in the NFL. Yeah they get a lot of points from kicks. But free throws in the NBA like field goals kicks in the NFL, cheap and BOOORING. Who wants to include in stats of the most boring aspect of the game? Though it all tallies up on the scoreboard the same I guess.. if Ladanian Tomlimson after scoring a TD lines up a kicker and gets the extra point kick, you'd have to create a statistical measure for it. Luckily most free throws are earned instead of handed out.
HP and dursava, Let me try again. Player X takes 10 FGs per game and 4 FTs per game. Player Y takes 10 FGs per game and 12 FTs per game. Let's say they have identical FG% .50 and FT% .75. And let's eliminate 3pt shots and make the coefficient .50 for simplicity. Player X's TS% is .542. Player Y's TS% is .594. In a way, the significance of Player Y's FG's shots are "dragged" down by his high volume FT. But the end result is still good for him because FT% is almost always higher than FG%. So I don't see HP's criticism about the "drag." It is almost always more efficient to score from the line than from the field. Now, I think maybe HP is trying to say FT shouldn't count as shooting efficiency, which is legitimate. Like dursava kept saying, what are we talking about? Is it (1)shooting efficiency or (2)scoring efficiency, or (3)offensive efficiency? If it is shooting, then FG% is just that, pure and simple. Just adjust it for 3pters. If it is scoring, then I think both TS% and PPS have their own merits. If it is offensive efficiency we are talking about, then neither is adequate. I think HP is talking about #1, dursava is talking about #2, and I am talking about #3. That's why we are still not seeing eye to eye. Does that make sense?
I hope Novak doesn't read cf.net. A 5 page thread is started for him and yet most of it is dedicated to stats. I know I'd be disappointed.
if he just learn more DE ,he may be the one of rockets in this year .so we do not have to miss this guy ,right.now what i miss ,just the tickets.i just no money~
At least this thread, right? I never intended for it to be an argument over different kinds of stats at the time I started it. Here's the link of a previous post of mine in this thread of questions open for debate concerning Steve Novak's future in the league since he's now on his final year of his contract. http://bbs.clutchfans.net/showpost.php?p=3968399&postcount=54