Player A -Scores 1.5 points per minute (the exact metric is irrelevant, we're just saying that Player A is more productive than Player B in a certain context) when the margin of the game is 10 points or fewer -Gives up 0.5 points per minute when the margin of the game is 10 points or fewer (again, the exact metric is irrelevant, we're just using it as a point of comparison) -Both scores and gives up 1 point per minute when the margin of the game is larger than 10 points Player B -Scores 1.5 points per minute when the margin of the game is larger than 10 points -Gives up 0.5 points per minute when the margin of the game is larger than 10 points -Scores and gives up 1 point per minute when the margin of the game is 10 points or fewer Everything else about the players is assumed to be equal. For the sake of this argument, let's assume that the score margin is 10 points or fewer in exactly half of a basketball game and more than 10 points in exactly half of a game. So as a result, both players will average the same number of points per minute and give up the same number of points per minute, and their statistical totals will be identical. Do both players contribute equally to wins? No! Player A's excellence when the game is close helps his team win more games. If his team is down a little, his play gets his team closer, and hopefully over the hump, and he contributes greatly to this result until his team goes up by 11. If Player A's team is winning by a little (or if the game is tied), then Player A contributes greatly to enlarging that lead until it hits 11. Once Player A's team goes up by 11, Player A starts playing worse. However, the other team cannot get any closer to coming back, because any time the game gets close enough, Player A "puts his foot on their throat" and puts it out of reach again, so that they cannot come back to take the lead. When Player A's team is winning by 10 points or less, it's relatively harder for the other team to come back than against Player B's team. When Player B is playing great, the result of the game is ALREADY ALMOST DETERMINED. His excellence does not affect the %s of each team's chances of victory as much as Player A's. Even if his great play helps his team get back within striking distance from a large deficit, his worse play once the score gets close keeps his team from significantly increasing its chances of victory. What would happen in a hypothetical matchup between Player A and Player B's teams? Player A's team would get out to an early lead in this theoretical matchup, and Player B's team would really never get that close again. Once Player A's team's lead gets bigger, Player B starts excelling and Player A starts to perform like a replacement player, so by the end of the game, if the game is 50% margin>10 and 50% margin < or =10, then both players will have accrued the same statistics. If these teams played 82 games against each other, then theoretically both players would have the same statistical totals, but Player A would win all of the games. The key to this idea is that the margin of victory is irrelevant. A 20 point win counts for just as much as a 1 point win, and a 20 point loss counts for just as much as a 1 point loss. Those 19 irrelevant points create a veil of statistical data that skews the perceptions of players' contributions to their team's wins when one doesn't watch the games and looks solely at data (and wonder why this 20ppg scorer never makes it to the postseason while that one keeps making deep postseason runs). To use a more real-world example, a certain anonymous player, let's just call him K. Martin- no no no, too obvious, let's call him Kevin M.- has a TS% of .600 but tends to play better in low leverage situations. Now take another player, let's call him Carmelo A.- no no no, let's go with C. Anthony- who has a TS% of .550 but plays better in high leverage situations. If C. Anthony's advantage in crunch time outweighs Kevin M.'s advantage in overall efficiency, then he is still the better player by TS%. I believe that this "clutchness" is a very real effect, and probably measurable if someone had the time and labor to keep track of basketball metrics in relation to score margins (this clutchness stat would be even more useful if it were parsed by how close to the end of that game each statistic was accrued). Psychologically, I believe that the effect is caused by a player's mental state- specifically, how high they are on Maslow's Hierarchy of Needs. If a player is higher on the pyramid, then they will be able to focus on playing creatively and chasing victory, and play uninhibited in crunch time. If a player is lower on the pyramid, and still needs their self-esteem, confidence, or security MORE than self-actualization, then their will will be divided among different goals and not be as competitive toward the objectives of the basketball game. As a caveat, people do change, so there is hope for the chokers out there. Also, actual levels of clutch changes from moment to moment, even though general patterns do exist. I wrote this piece to mostly once and for all (hopefully) end the idea that 30 ppg = 30 ppg, no matter HOW they do it. It DOES matter how they do it, in high leverage situations or in low leverage situations. Guys who do it better in high leverage situations, EVEN IF their overall efficiency is lower, are better basketball players because they help their teams win more.
uh...you can have said that in just 2 sentences.. anyways I don't think Kevin Martin has had enough opportunities esp in the playoffs to prove his "clutchness" He just hasn't had a chance....
He's only played 6 playoff games back in his sophomore year. He hasn't had a chance to prove his worth in the playoffs in big key moments. If Martin was on a winning team his whole career (say like the Pistons team Rip was on) then he would be regarded as a completely different player, and would receive alot more respect than he does now. Therefore your great analysis is irrelevant
That being said..I would still prefer melo over Kmart because i believe carmelo is simply just a better player.
Can't really comment on how this stuff applies to Martin vs Anthony and what conclusions we should draw. But I agree with your premise. Not all points are "equal" in basketball. Some moments are obviously more important than others, and you want to have players who can execute in those moments. I posted this before, but its worth posting again because it applies to the thread topic: http://www.nba.com/rockets/news/Q_and_A_with_Daryl_Morey-299705-34.html [rquoter] JCF: I heard you discussing something the other day that I found to be fascinating. You were talking about “high leverage moments” late in games. Can you go into the meaning of that term and how it relates to the Rockets? DM: I don’t want to go into details since it’s somewhat proprietary, but it basically looks into what people would consider key moments of the game, where the chance of one team winning or the other would shift dramatically. We just add a little more sophistication to it, that’s all. JCF: And the Rockets have been very poor in that area, correct? DM: Yeah, we’ve been last out of all the playoff teams and we’ve been even worse recently. So we have to address that. I think we might see Aaron a little more, because he’s had some success in those situations. JCF: But I imagine a big part of getting better depends on Tracy getting back to being Tracy? DM: Yeah, that’s the most important thing. I mean, we’re not counting on Aaron. We’re expecting Tracy to get back to that role. JCF: How have you been in recent years in that area? DM: We haven’t been the worst, but we’ve not been good. Below average. JCF: So if history indicates you’ve been below average and you’ve been even worse this year, do you feel like it’s realistic to expect that to change? DM: That’s why we’re looking at things like Aaron, and also coach may try to get Tracy into better situations. [/rquoter]
i keep forgetting to mention this since late last season...but Martin does seem to hit some timely buckets...i didnt pay as much attention to this throughout the preaseason, but i think i noticed it then too...he looks to shoot more when there's a chance to stop the other team's momentum...i dont think i've heard this addressed by morey or adelman though, so maybe it's not as frequent as i think it is
A team full of either type of player would never lose a game, since they never give up more points than they score. Of course, a team of Player As would always win by 10.5, and a team of Player Bs would always tie. In this hypothetical world, you'd want to start Player A until you get the lead past 10, then let Player B run away with the lead.
Huh? I don't get your logic at all, both in comparing Melo and Martin, and your theoretical players A and B. 1) Regarding A & B: I don't see what you're trying to compare in the two players, both of them are equally clutch and equally chokers. Player A starts choking when the opponents' lead becomes big, and player B starts choking when the lead becomes small. In the case of player A and player B meeting up, the challenge is will team B be able to get an 11 pt lead? If they're good enough to get an 11 pt lead even once and no matter when victory is assured. In your theoretical game, team B is actually the one who wins. Why? Well because you said 50% of the game the margin will be 11 pts or higher. This means team B will win, because once the margin reaches 11+ pts player B will come alive and player A will perform like a scrub. Considering team B was good enough to get 11 pts when player B is performing like a scrub and player A is perfoming like a star, what's the chance of Team A catching up when player B is perfoming like a star and player A perfoming like a scrub? So is player A better than player B? Yes...but only because you put the margin of clutchness at 10 pts. This makes the lead almost unattainable because a 10 pts swing is like a 20 pt difference between the two teams. If you put the margin of clutchness at 2 pts (which is more accurate because "clutch" players make their shots in win-or-lose games) which is better? Player B of course. Once you get that 2 pt spread you win the game. I mean, in the course of a game I'm sure the Memphis grizzlies can get 2 pts over the Lakers, and once that happens its a sure win. On the other hand, a clutch player who only performs on a 2 pt spread is a role player, someone like old Robert Horry who can take that last shot but is crap 99.99% of the game. So you see, in your scenario the "margin of clutchness" that you set is what determines which is the better player, and not the type of players. 2) Regarding Melo and Martin: *You can't compare the two on equal footing because Melo is making 2x more than Martin *Melo doesn't become as efficient as Martin "in the clutch", being clutch he just performs more or less at the same crappy rate. So really your argument should be: Player A: -Makes 20M a year -scores 1 pt per minute no matter what the circumstances or Player B: -Makes 10M a year -Scores 1.5 pts when the game isn't on the line -Scores 0.5 pt when the game is on the line If we assume "on the line" means 4rth quarter (or the money quarter), which player would you get? If you average it out: Martin: 1.5 +1.5+1.5+0.5=5 Melo: 1+1+1+1=4
I guess as my final opinion on this matter, isn't being clutch (or a choker) reflected in a player's efficiency? I mean, sure, not all points are equal, but shouldn't you take the whole body of work into account? If someone like Melo shoots 2-10 from the game but hits the game winner (making him 3-10), is he better than Dirk who shoots 7-10 from the field but misses the shot that would have tied it? IMHo it doesn't matter how clutch you are if you're bricking the entire game prior you basically just broke even by hitting that shot. Again if every one of your players shoot 7-10 when the game isn't on the line then it would be a blow out and you won't need clutch guys in the first place.
1) Your 2 point margin hypothetical only makes sense because the amount of time that the game maintains a margin of 2 or less points is WAY less than the amount of time a game has a margin of greater than 2, so Player B would have VASTLY superior productivity over the course of a game. I arbitrarily chose the 10 pt margin because it seems to split the game in half so that the total statistics would be equal and thus the difference would be more easily perceived, and thus be a more useful illustration for the clutchness phenomenon. 2) I agree that adding salaries into the analysis adds another element to the comparison. However, knowing whether we're better off with Melo and his higher salary depends on who else is on the team and who else is available- cap space is useless when there's no one out there who adds to the team. I was just trying to shed light on a hidden element that is rarely discussed in statistical analysis, how a player performs relative to a high/low leverage point in the game (which can be applied to ANY metric), when trying to evaluate players. And of course points per minute is only one out of a million metrics to evaluate players. But the whole point I made was that clutchness was obscured by "noise" in commonly used statistics, which don't differentiate between production in low and high leverage moments, and that it can be more precisely teased out by separating low and high leverage moments, whether by score margin or time remaining in the game, and representing the clutchness stat (at least my version) using a variable or coefficient (for example -1.32 or +1.15) that compares the particular metric (TS%, offensive rating, defensive rating, rebounding%, etc.) divided into certain moments of the game (0-12 minutes into the game, margin < 10; 0-12 minutes into the game, margin >= 10; 12-24 minutes into the game, margin < 10, etc.) and then taking into account the value of those stats in each different situation as it relates to the chances of winning the game (a 2 point last second shot when you're down one point can make a 100% difference in your chances of winning the game, whereas a 2 point shot in the 1st quarter might make a 5% difference, as an extreme example), which gives the observer a more precise tool to evaluate players with. Of course the whole body of work matters- for example Player A might be worse than Player B if he could only score 1.1 point per minute when the margin is 10 or less while Player B could score 1 point per minute, and Player B could score 2 points per minute when the margin is more than 10 and Player 1 could only score 0.5. Clutchness is not the only factor in evaluating players, it's just a hidden element that I'm trying show ways to make more visible.
I guess my question is how is a 10 pt margin for clutchiness splitting the game in half? In a battle between two completely even teams a win of 10+ points is already considered a blow out, or not close, right? So by making the margin of clutchiness 10 pts you're effectively comparing a player who only scores during blowouts and a player who performs in non-blowout situations, which as you know is already the majority of game time. My 2 pt margin of clutchiness is the other extreme of what you're using, whereas your 10 pt is biased towards player A then the 2 pts is biased towards player B. That's why I'm saying its not the type of players that matter in your example, but where you decide the margin of clutchiness is. Regarding Melo and Martin, other than the salary thing you can't really compare your player A and player B to Melo and Martin because Melo doesn't become as efficient as Martin at any point in the game. In your scenario Player A becomes efficient as player B when the game is close, and becomes inefficient when the the game is a blow out. However in Melo's case he doesn't post a TS% of 0.50+ when the game is close, his TS% just lowers less than Martin's, whose shot becomes 0.30's when its contested (AND NOT A 3 or a FT). I don't know if you were influenced by that Nuggets post that talked about contested stats, but if you are that has a huge asterisk as it only took into account contested shots that aren't threes or were called fts. Maybe its just me but contested shots that aren't threes or were called fts is a small portion of the game, like maybe that last shot that loses the game or wins the game. I don't know, is a guy who's better in that area but worse in every other area in b-ball worth twice the salary? I agree with you on this issue but unless you can get your hands on DM's raw data or can actually track the difficulty of every shot per possession its quite hard to put it into practice. If your intent was just to prove that not all shots are equal and some buckets are harder to score than others then I agree with you. However if your point is to prove that games in the last quarter (or last seconds or that last possession) is worth more than buckets made during the beginning of the game then I don't agree with you, simply because in the NBA all shots inside the 3 pt line count as 2 pts regardless of difficulty. If you only scored 2 buckets, one in the beginning and one in the end to win it, which is more important? If you missed the beginning but made the end would you win? If you missed the end but made the beginning would you win? No, you need both, as missing either would result in a tie.
Because it controls for the amount of time that the game is spent below and above the margin, making the effect of clutchness clearer and without the interference from unequal weight. We can use WHATEVER the margin is that splits the game into 2 halves of time. 10 was just a guesstimate, it doesn't matter what the number is. I don't know what Martin and Melo's clutchness stats look like, as I don't have the time and labor to keep track and crunch all the numbers as I would like, so I can't comment on that at all. Buckets in the last seconds ARE worth more than buckets made at the beginning of the game, because they affect the team's chances to win the game by different amounts. In response to your position, once a bucket is scored (or missed) early in the game, it's done and it's in the past. It no longer has the potential to be made or missed, like in the question that you raise, "If you only scored 2 buckets, one in the beginning and one in the end to win it, which is more important?" They're actually independent events. If we could go back in time and run simulations from each point in the game at any later point in the game, then they would be equally important. But since we can't, once it's made or missed, then it no longer affects the adjustment to each team's chances of winning the game made by the outcome of the second shot. At that point, only the current shot is taken into account in terms of importance, and the later the shot, and the closer the margin, the larger the swing of winning% created by each shot attempt.
Interesting premise, but I think the shatterpoint of the argument is the notion that a tie game at the beginning of the first quarter is equivalent to a tie game with one minute left, which is equivalent to a nine-point lead in the third quarter, and so on. I think most would agree that Martin and Martin-like players are going to see a dip in production in the last three minutes of a close game, but I honestly won't believe that the same is true of first-quarter tie games and seven-point affairs until someone proves it. If you only look at the current definitions of clutch stats, you'll see they're only applied across a fraction of a player's minutes, and even a massive difference in player value between two players with equivalent minutes (say, James and Wade in the above chart) played is a drop in the bucket for the season at large.
How can point margin control the amount of time in the court in the first place? Point margin is erratic and doesn't correlate to time played on the court at all. When people talk about "clutch" they usually refer to either the last shot, last quarter or last few minutes of the game, not a certain point spread throughout the entire game. In a close game where you exchange leads throughout the game your "clutch" guy will be clutch throughout the entire game. What kind of clutch player is clutch the entire game? He's not clutch, he's Michael Jordan. You don't even need clutchness stats to see that Melo isn't as efficient as Martin is normally, its common sense. A contested shot is more difficult than an open shot, yes? Martin scores more efficiently than Melo on open buckets. So how will it be possible that Melo reaches Martin's open shot% on Melo's contested shots? Like I said you assume that Player A reaches player B's efficiency when its harder to score, and that's just impossible. What kind of player plays poorly when baskets are open but then makes every bucket when hands are in his face on every possession? Its not logical at all, and if there are players like that I think we should exclude them as those will probably just be anomalies or outliers in the data. So let me get this straight, on one hand you say they're independent events, but on the other you say the first bucket is done and in the past? How can you treat them as independent events, but then view one event as inevitable? Yes I agree in terms of calculating the value of the last second shot, the value of the first shot is already known. However we aren't talking about each shot, we're talking about calculating all shots in the context of the entire game. In the context of the entire game, there is no "game-winning bucket" because if you only make that one bucket but miss the others you lose the game anyway. Also just because something is known doesn't mean it loses its value, rather its value becomes identifiable. In terms of making the last shot, you already KNOW you made the first shot. That means the first shot's value becomes known, because you know that shot gave your team two points. By this logic only the last shot deserves credit, all previous baskets are rendered useless simply because you now know you made the shot. Each total score is made up of a collection of pts you scored, each of which affects the value of each shot. You can't independently identify whether one shot is worth more than the other because taking out one shot alters the value of all shots in the game. If you had a 20 pt lead and you took out all of the blowout points then all the remaining shots would spike up in value. So how can one shot in the end be worth more than one shot in the beginning when if you missed the shot in the beginning the shot in the end loses its value?
Because when you miss the shot at the beginning, you can still win the game, but when you miss the shot at the end, you definitely lose the game. It's impossible to lose a game by going back in time and taking away an earlier bucket, but it is possible to win a game by making buckets after missing one. Not only the last shot deserves credit because each shot contributes some (I would only guess that the value of each shot would be determined by a model of the likelihood of a team winning after a certain score and a certain outcome of the shot, based on the sample size of every game ever played), but the more clutch the moment in the game, the more the shot contributes to winning.
I guess one player thrives on the situation and pressure while the other falters under it. A Jordan would lift his game to a whole new level when the pressure is on, whereas another player (not naming names) will not or cannot, or may even crack under the pressure. With Robert Horry, I tend to think he cruises through the game, but when the pressure is on, he concentrates better and lifts for the occassion. Playoffs are all about pressure and this is what distinguishes the legends from the also-rans.
Look, when you evaluate the shots that win or lose you the game more than likely the game is already done right? How can you go back and look at the last retroactively (as in determine that if it misses you lose the game and if you make it you win the game), but you can't apply this same "back to the future" approach when looking at the first shot of the game? If the every shot played out the same way, with the only difference you missing the first one instead of making it, would you have won? In a close game where you only won by 2 pts, that first bucket that was made is the difference between overtime and the win. Saying "the last shot won or lost you the game" is wrong because you are assuming these events are independent, which they most certainly are not. Like I said they are independent if you only made the final shot but you missed every shot prior that would you would have won? NO right? So these shots aren't independent actions, but are actually interconnected and affect each other.
