20 Second Timeout is the place to find the best analysis and commentary about the NBA.

Tuesday, November 06, 2007

Watching Games Versus Crunching Numbers

In many sports, including basketball, there is an ongoing debate between those who believe that players and teams are best evaluated by trained individuals watching the games versus those who believe that players and teams are best evaluated purely by crunching numbers. There are extremists on both sides--people who claim that the observation process is worthless because it is too subjective are pitted against those who have no interest in looking at statistics. The most reasonable approach is to combine observations--either one's own or those provided by qualified people (such as scouts)--with pertinent statistics. That still leaves two questions: what are the most important things to observe when watching a game and which are the most important statistics to consider? Many of my articles have dealt with the subject of how to watch/evaluate players, including my two part "Scout's Eye View of the Game" series (click here for Part I and here for Part II), so this post will focus on how to most effectively incorporate statistics into the overall evaluation process.

During a recent telecast, ESPN analyst Jeff Van Gundy explained which statistics he looked at first after a game when he was coaching: 1) true shooting percentage (which takes into account a player's field goal, three point and free throw shooting); 2) rebound percentage; 3) turnovers; 4) free throws made. On the other hand, Van Gundy said that fast break points and points in the paint are two statistics that he considers to be overrated; he noted that he does not like the way that fast break points are compiled but did not give a reason for rejecting the value of the points in the paint statistic. It is important to keep in mind that Van Gundy's list has more to do with evaluating his team's performance as a whole than ranking players individually. That said, his approach is solid. Even most people who are not in the "numbers" camp should be able to understand that raw field goal percentage is not a very meaningful statistic because it does not take into account the extra point from each three pointer nor does it give a player credit for drawing fouls and making free throws. Rebound percentage is more precise than total rebounds, which can be affected by pace considerations and how many shots are missed in a given game. Most coaches try to limit their team's turnovers to less than 15 per game, while free throws made can be a good indicator of a team's aggressiveness.

By considering the four factors that he listed, Van Gundy can quickly determine how well both teams shot the ball, how effectively they rebounded missed shots, how well they protected the ball and how aggressively they played. This is a good quick and dirty method for examining his team's performances in several vital areas. I am a little surprised that Van Gundy did not mention defensive field goal percentage, although perhaps he includes that in true shooting percentage (by looking at the true shooting percentage of the opposing team). Gregg Popovich and coaches on his "coaching tree" (like Cleveland's Mike Brown, who has ex-Popovich assistant Hank Egan on his staff) are among the many coaches who place a lot of emphasis on defensive field goal percentage.

While Van Gundy's system is a good way to look at a team's performance, I am not sure how useful it is to evaluate the performances of individual players for two reasons.

First, a team's overall true shooting percentage is important but individual true shooting percentages are not created equally because players have different roles; some players are called upon to create shots for themselves and their teammates, other players are spot up shooters and other players hardly shoot at all. If a player hardly shoots at all because he is primarily a rebounder, defender and/or screener, then his shooting percentage is not of primary importance (which is not to say that a coach does not want him to shoot well). Spot up shooters should have excellent true shooting percentages because they are spoon-fed the ball when they are wide open; on the other hand, just because a spot up shooter like Jason Kapono shoots better than a creator like Kobe Bryant we should not conclude that Kapono is more "efficient" than Bryant. Kapono is not asked to score 30 ppg and it is not likely that he could do so for an extended period of time; if all Bryant did was spot up and shoot wide open shots then his shooting percentage would go up but his value as a player would decrease because he would not be as productive as his skills enable him to be. Therefore, players should be compared to players who have similar roles (not necessarily even guys who are listed as playing the same position)

Second, while it is important that a team commit less than 15 turnovers per game, not all individual turnovers are equally bad--yes, they all are part of the team's total, but just like different players have different shooting responsibilities it is also true that different players have different ballhandling responsibilities. Most teams have one or two players who do the lion's share of the ballhandling and it is inevitable that they will commit some turnovers just because of how often the ball is in their hands. That is why some of the greatest players of all-time--like Magic Johnson--appear prominently in the record books in the turnover category. Of course it would be preferable if a player committed no turnovers or at least had very few such miscues but that is not realistic, so a great player's turnovers have to be considered in the context of his overall production. That does not mean that I am a big fan of looking at assist/turnover ratio, which I consider to be an artificial statistic because not all good passes become assists and not all turnovers are the result of bad passes. If a player is a very productive scorer and/or playmaker then he most likely will commit three to four turnovers a game; sure, if he could be productive with fewer turnovers that would be great but I would not cut Magic Johnson because he turned the ball over a lot. What should raise a red flag is if a player who does not handle the ball that frequently commits a lot of turnovers. If a team is fortunate enough to have a Magic Johnson, then most of its other players should not be committing many turnovers because he will do most of the ballhandling while the other players will benefit from receiving the ball in their best scoring areas. In other words, if a player like Magic Johnson commits three, four or even five turnovers that is not a big deal but if players who do not handle the ball that frequently are also committing that many turnovers then the team's total turnovers will be well above 15 and that will be a problem.

Another thing that is important to understand is that some kinds of turnovers are more damaging than others. For instance, if a player loses the ball because he dribbled into traffic or if he throws a bad pass that is stolen then the opposing team will have an excellent opportunity to score a fast break basket. However, an offensive foul or a ball that is dribbled or thrown out of bounds is like a made basket (without the two or three points of course) in the sense that the other team has to inbound the ball and try to score against a team's set defense (the offensive foul could be a problem of a different kind if it results in someone fouling out but right now we are looking specifically at the turnover issue). Last season, the players who committed the most turnovers were Dwight Howard, Eddy Curry, Steve Nash, Allen Iverson, Andre Iguodala, Kobe Bryant, Lebron James, Ben Gordon, Deron Williams and Gilbert Arenas. Iverson is often criticized for his turnovers but he scores a lot and distributes a lot of assists; the same is true of everyone else on that list except for three players: Howard and Curry commit far too many turnovers for players who have few ballhandling responsibilities, while Gordon is a jump shooter who does not have playmaking responsibilities and is not as productive overall as the other perimeter players on the list. Again, the issue is not assist/turnover ratio but rather the ratio of overall production to turnovers. Let's look at two of the NBA's best players, Kobe Bryant and Steve Nash. In 2006-07, Bryant committed 255 turnovers and Nash committed 287 turnovers but they committed their turnovers in very different ways. Bryant is a scorer who frequently slashes to the hoop, so--according to 82Games.com--he had 26 offensive fouls, 109 bad passes, 119 ballhandling turnovers and one miscellaneous turnover, while Nash had 11 offensive fouls, 223 bad passes, 53 ballhandling turnovers and one miscellaneous turnover. Both players committed turnovers at an acceptable rate considering their overall production and both players committed the kinds of turnovers that one would expect them to commit based on their roles.

This discussion of the limitations inherent in trying to evaluate individual players based on their shooting percentages and turnover rates illustrates the two main drawbacks with evaluating players primarily by statistics: numbers can only tell you some of the "what" and none of the "how"; numbers are useful for finding out who shoots the ball with the most accuracy or who had the most turnovers but only by observation can one provide the necessary context to understand what those numbers really mean. Another limitation of statistics is that they cannot capture in a meaningful, consistent way intangibles such as Bruce Bowen's man to man defense. Bowen does not get many steals or blocked shots yet he is universally recognized as the best perimeter defender in the NBA and, despite pedestrian statistics in most categories, is a starter for the NBA Champion Spurs. Yes, plus/minus data can give you a glimpse into Bowen's worth in a general way but no number or set of numbers really gives a clear description of exactly why he is so valuable.

I don't consider one statistic or one statistical system to be a "Holy Grail" in terms of individual player evaluation. I firmly believe that you have to look at a player's total profile: how he actually looks in games (in person if possible, otherwise on video) is one part of the picture, to be supplemented by his total statistical production. Some players are very dominant in one category but not so exceptional in most others (super rebounder Dennis Rodman is the classic example of this), while other players are productive across the board (Kobe Bryant, LeBron James are two examples of this). Watching a player perform is the only way to completely know what he does and how he does it; then you can look at his numbers compared to the numbers put up by other players who play the same position.

None of this means that I am against using statistics or even that I am against people trying to create a "Holy Grail" composite number that truly provides an accurate way to rate players--but until such a method is developed the best way to evaluate players will remain the tried and true approach of combining first hand observation with the judicious use of all available statistics. One excellent thing that has happened in recent years is the development of new ways of looking at players' statistical contributions, such as plus/minus (how well a player's team did while he was on the court) and adjusted plus/minus (the adjustment involves factoring in the contributions of the other nine players on the court to determine if the player in question really had a major impact or if he was basically an innocent bystander while others did the lion's share of the work).

posted by David Friedman @ 3:05 AM



At Tuesday, November 06, 2007 7:08:00 PM, Blogger element313 said...

more about Kobe:

read this, http://insider.espn.go.com/nba/insider/columns/story?columnist=ford_chad&page=KobeWorth-071106&action=upsell&appRedirect=http%3a%2f%2finsider.espn.go.com%2fnba%2finsider%2fcolumns%2fstory%3fcolumnist%3dford_chad%26page%3dKobeWorth-071106

in other thread you ask if my position is that other players cant develop around him & you cite Bynum & mihm...
my position is that other perimeter players are vastly impeded by his taking too many shots. but his shooting too much has far less of an inhibitng effect on big men, b/c they get more offensive rebounds (leading to shots) and get fed in teh post on different types of plays... so he is more harmful to the development of other perimeter players (Caron Butler, esp.) than big men

to follow up on prior post on KG: I should have added another point -- I think Boston has a decent chance of making the conference finals & even the NBA FInals, notwithstanding that KG is not as good as Duncan or ShaqinPrime

At Tuesday, November 06, 2007 7:35:00 PM, Blogger David Friedman said...

I already read Ford's article and posted a rebuttal here.

As I explained before, Butler has improved each year of his career other than the one season when he was impeded by injury. There is no evidence to suggest that Kobe held him back in any way. Moreover, not that I really care about such things, but Bryant and Butler are actually good friends off of the court and Bryant has called Butler one of his favorite teammates.

The other Lakers perimeter players are spot up shooters or defensive specialists. If Kobe were not on the team they would not be shooting much more frequently or effectively. It should always be easier to play with a great player--be it a perimeter player like Kobe or an inside player like Duncan--because great players distort the defense and create open opportunities for their teammates. The plus/minus data in the first couple Lakers wins hint at this, because even when Kobe is not scoring 40 or 50 points the team still has a great plus/minus when he is on the court; his presence-- and the defensive coverage that he attracts--makes others better (or at least gives them the opportunity to be better if they can make open shots).


Post a Comment

<< Home