I think we are on the same page and I'm just doing a horrible job, as usual, of explaining myself. My point is that the plight of the Heat would suggest that clutch ability is not overrated. The probability that the Heat's losses in close games are attributable to pure chance is a little less than 2%. It's overwhelmingly more likely that clutch play is a team skill that can be measured and incorporated into better algorithm to anticipate a team's overall strength in the playoffs.
I get it now. I still think Hollinger would say it's a fluke. If they win 6 out of the next 10 games which are decided by 2 points or less, will it be because they gotten better or was it a fluke?
http://www.amazon.com/Pro-Basketball-Prospectus-John-Hollinger/dp/157488655X/ref=pd_sim_b_1 You can pick up a used copy of Hollinger's first book for less than $5 shipped to your door. Anyone that has interest in this thread would enjoy picking this up.
Yeah perhaps. I'd bet if you got Hollinger drunk he'd admit that it's a little bit of laziness -- that by May most teams have worked out their end of game play to the extent that you doesn't add much to the conversation . . . which may be true with the Heat this season. Who knows what's going to happen? As of now, though, if the Heat were to win six of the next ten games I believe it will be because they have gotten better, not because they currently fall in the 2% of situations where a team goes 1-8 in five point games by chance.
I saw that and was like, Johnstone's writing a reply right now. The funny thing is that the Heat's problem is that they are unable to figure what to do with both Wade and James on the court at the end of games. They were probably better at the end of the Piston's game tonight because Wade was out with an injury . . . which is really sad when you think about it.
All sportstwo articles are mixed in with the new stories on sportstwo.com's home page (nav button to the far left at the top of every sportstwo.com page) where they appear in order of when they're "promoted" to the home page. The last article I wrote (on the Bears' Jay Cutler) is now about 5 or 6 down on the home page's second page).
OK, here's a draft...please comment. Keep in mind that it's targeted to casual fans, not really you guys. High-Tech Hoops Stats for the Casual Fan – VOL 1 – True Shooting Percentage Before some people who know me start throwing things, let me begin by saying that I’m not any sort of stats expert and often chide those who I feel put too much stock in the numbers and not enough in what their eyes tell them when watching games. In fact, it’s because I’ve only recently become acquainted with the these advanced basketball statistics and view them with what I feel is a dose of healthy skepticism, that I decided to do this series. If you’re new to these stats, maybe this article will help you cozy up to them a bit. If you’re already a card-carrying stats maven, you may want to read my simplistic take to further enhance your feelings of superiority. Once upon a time, when basketball fans wanted to know how their favorite teams and players were doing, they looked to a few simple numbers like win-loss percentage, simple shooting percentage, points scored, rebounds snatched and assists dished. The truth is that this is still about all you’ll find in your local newspaper’s box scores. Then long about the 1970s, some new statistics surfaced including steals, blocks, turnovers and breaking rebounds into their offensive and defensive components. Of course today, these are considered to be as basic as the old standards. Following on the heels of baseball’s “saber-metrics” movement of the 80s, the 90s saw the genesis of a new generation of basketball measurements designed to bring us added insight into team and player performance. We’ll deal with something called True Shooting Percentage (TS%) today. We’re all familiar with simple shooting percentage…you take 6 shots, make 3 and you’re shooting 50%. Then along came the three-point shot. Because of the increased reward for converting from the arc, “Effective Field Goal Percentage (eFG%) was created. The formula is fairly straightforward: (Field Goals made + (0.5 * 3-point Field Goals Made)) / Field Goal attempts Put into words, the eFG% formula simply gives an extra 50% credit for a made three-pointer (the difference between 3 and 2). So with eFG%, if you take 6 shots and make 3, but one of them is a 3-pointer, you’re credited with 3.5 shots made in 6 tries and have a 58.3% eFG%. The next question asked was "What about free throws?" They count as points and help your team. Shouldn’t they be taken into account as well? The answer was “Yes” and the stats gurus came up with the True Shooting Percentage statistic. When you read or hear media types writing or talking about a basketball player being efficient or inefficient, it’s a good bet that they’re basing their opinion on this statistic. The formula is a little hairier than eFG%, but let's stay with it: Points / (2 * (Field Goal attempts + (0.44 * Free Throw Attempts)) As you can see, since the formula starts with total points scored, 3-point shots are given full credit and made free throws are included. Perfection is defined in several ways. If you took only 2-point shots and made all of them, you’d have a perfect score of 1.000. Converting on 2/3 of your 3-point shots is also deemed perfection by the formula. For free throws, an 88% success rate is considered perfect (the reason for the 0.44 in the formula). Overall, I like the true shooting percentage stat. It does a good job of measuring the efficiency of a scorer and does a good job of exposing the hated “chuckers” who never met a shot they didn’t like. Of course, it has its flaws and they go beyond the fact that the 88% “perfect” free throw percentage is completely arbitrary. To some extent, I see it as a “rich get richer” statistic, since in the NBA, unfortunately, “superstar calls” remain the norm. The great players already convert a high percentage of their shots (that’s a big part of what makes them great), but it also seems that darn near every time they miss they go to the line. For them, TS% is clearly a win-win stat. Other nit-picks include the fact that, based on the formula, making 2 free throws is better than converting on an “and one” and I don’t see why making technical free throws count since the shooter did nothing to deserve the opportunity (other than being good at free throws, of course). As with any statistic, the problem is often not with the stat itself, but with what folks try to do with it. Yes, TS% measures scoring efficiency nicely and efficiency is good, but a great TS% does not make you a great player, or even a great shooter for that matter. Players who “play within themselves,” seldom risking a shot they can’t easily make are likely to be TS%-studs, but average or below-average scorers. Again, efficiency’s nice, but in the end you’ve got to score more points than your opponent to win. A quick look at the current TS% league leaders illustrates the point: 1. Tyson Chandler-DAL .736 2. Nene Hilario-DEN .694 3. Arron Afflalo-DEN .644 4. Steve Nash-PHO .642 5. Amir Johnson-TOR .634 Don’t get me wrong, these are all nice players, but other than Nash, none are close to being considered great. Chandler and Johnson each average fewer than 10 points per game. Nene and Afflalo are probably the third and fourth best players on their own team. When the Nuggets really need a basket, do they go these efficiency experts? Nope. They go to the pretty darned inefficient Carmelo Anthony (.523 TS%...the NBA median TS% is .532). Why? Because Anthony creates his own shot opportunities rather than waiting for them, and when the shot clock is running out and the game is on the line, you’ve got to get a quality (even if not ideal) shot up…NOW! Anthony, like many other superstars, is regularly asked to take these tough shots under pressure and as a result, sacrifice efficiency for effectiveness. As you probably know, the NBA announced the starters for its upcoming All Star game last week. Of these great players (which includes Anthony), only Nash is in the top 20 in terms of True Shooting Percentage. Michael Jordan, the greatest player ever, had a decidedly unspectacular career TS% of .569. Reigning MVP, Lebron James has a career TS% of .563. Kobe Bryant’s is .557. Again, True Shooting Percentage is a good stat as long as you know what it attempts to measure…and what it doesn’t. It measures scoring efficiency…nothing more and nothing less. And in basketball, while it’s good to be efficient, it’s better to be, well, just plain good. Next up, Player Efficiency Rating or PER. It’s a hummer. Tom Nossem Sources: basketball-reference.com, espn.com
Good article, Transplant! I think it would be more readable if you added the definition to the two formulas. For example: Effective Field Goal Percentage (eFG%) = (Field Goals made + (0.5 * 3-point Field Goals Made)) / Field Goal attempts
One thing... I don't think your interpretation of the .44 is accurate. It doesn't have to do with the "right" free throw percentage of 88%. I believe it has to do with how many possession that a FT uses as compared to a FG attempt. A Field Goal attempt will basically use one possession (although there is a offensive rebound potential so sometimes this is adjusted when compared to a Turnover which always losses a possession). However, occasionally, a player will go to the line throw line for 3 attempts, not two. I think this changes the formula from (0.5 * Free Throw Attempts) to (0.44 * Free Throw Attempts). I tried to Google to confirm this but didn't have any luck quickly.
Ironically, 0.44 is covered on page 1 of the Hollinger book I linked to. It is due to possessions. Other instances where a FTA does not result in a change of possession is technicals. The "And 1" also does not result in an extra lost possession. This is interesting to me because shooting a mere 50% from the line would improve the TS% of someone that shot 50% from the field. This doesn't seem quite right because 50% from the line is relatively poor and 50% from the field is good.
Fantastic article, Transplant. I'm not sure if you would want to put this in the article, but I was left with one curiousity after reading your piece. We all know TS% is pretty handy, but is it more telling than points-per-shot? On the surface, I don't see why it would be. It seems to me that PPS has the benefit of favoring players who get to the line a lot, but don't necessarily shoot it all that well. And the fact that it doesn't need any co-efficients built into it is a plus, IMO. It's probably not a 'high-tech' stat, but I think it would be interesting to see how players differ in the two stats, to see which one is more telling. I would guess that if you compared the top 50 players in each, PPS would have more of the volume scorers like Carmelo and DRose that TS% tends to underestimate.
There's a stat I don't think I see on any of the stats sites. 2pt FG%. DRose is 8.9 for 20.2 FGA, and 1.6 for 4.2 3PA. His FG% is .442. However, his 2pt FG% is (8.9 - 1.6) / (20.2 - 4.2)
To me the "advanced stats" are a natural attempt to fill in the gaps created by the incomplete nature of the basketball boxscore. Unlike baseball, where you can recreate and tease out the nuances of almost any situation with stats, in basketball there are large gaps in what we can decipher in areas as fundamental as what a players defensive impact on a game might be. It might be interesting to highlight the fact that as basketball stats get more "advanced" they are more likely to rely on approximations. TS% is a perfect example of this with .44 * FT. Pace and APM are other examples. Does that mean the stats should be given less credence? I'd say no and I'd agree with Transplants original assertion that it's important to understand what goes into a stats calculation before using it as the basis of any argument. I'd also argue that the imperfect nature of stats highlights the fact that there really is no substitute for actually watching the games to really know where and when to properly use any stat. That's why a guy like Hollinger occasionally comes off sounding like such a tool, like when obviously looked at Rose's early season usage and low shooting %'s last year and concluded that Rose's teammates were giving him dirty looks for shooting so much. In fact, the opposite was the case and they wanted Rose to shoot more, but you wouldn't get that from just looking at the stats.
Thanks to all. The final article is now on the site's home page. It's also a post on the site's NBA general message board. And Jay, at the end of the article I list my sources...you should check it out.
Nice article transplant. For your next one I'll try to dig up and send you the spreadsheet I once used to try to manually calculate out PER. Probaby better to tell you the punchline: I couldn't get it right. And I once made a living calculating out statistical models of things. That being said, as time goes on, I find that PER is a pretty robust statistic, meaning that it seems to predict wins and losses pretty well. Not perfect, but it probably gets the job done as well as most single measure stats. ----- Something I'd really like to see is a matrix of runs by NBA teams. If you look at http://popcornmachine.net/cgi-bin/gameflow.cgi?date=20110129&game=INDCHI (there's another one transplant!) they break out all the runs that happen in a game by parsing the game logs. I've aways heard, and generally believed, "basketball is a game of runs". This works on point differential too. A good team should rarely give up, but frequently go on positive runs. I'd guess a positive run is better to go on when the game is close. And a big run is better than a small one. So we ought to be able to make something like a table or graph that has Y | X - Size of Run + + Point Differential at beginning of run - For example, using the Bulls/Pacers game last night, the Bulls would look something like Diff Run -2 8 6 -5 4 -12 -7 5 -2 7 1 8 9 -7 4 5 9 -6 7 -7 0 5 5 6 11 -5 6 13 16 6 ---- (The pacers would be just the opposite) Anyway, if you tabulated this out, and started looking at it for lots of teams, you'd get a good idea of which teams keep it close in garbage time, and which teams pull ahead when the game is close and then coast to victories.
It actually means that one team is better than their record indicates. An analogy would be Green Bay Packers vs Atlanta Falcons. One team got lucky in close games (1 point wins come down to luck), the other was injured and without aaron rodgers in two losses. It tends to show the true talent level of teams.
