Analytics and Anecdotes is a four part series whereby I attempt to examine the growing influence of statistics and analytics in some sports -popularised by the Brad Pitt film, Moneyball- while it’s apparent inability to permeate existing structures of assessing players in others. In the first part, we discussed the basic tenets of analytics and in the second, we looked at how baseball has been transformed and is now light-years ahead of any other sport in it’s use of analytics.
In this part, we take a look at basketball and how analytics is slowly impacting the game but it has not progressed to the extent that baseball has. So where does basketball figure in the statistics game? And in today’s game where all kinds of numbers are being thrown, how do we decide which ones to use and which to disregard?
Basketball is one of the most popular sports in the world. And the NBA is it’s most competitive league. We live in an era where even the “superteams” like the Miami Heat of the early 2010s and the Golden State Warriors with Curry, Durant, Green and Thompson struggle to win three or four straight championships. And while having other contemporary strong teams like the San Antonio Spurs or the Cleveland Cavaliers has played a part, victories between closely matched opponents can often be put down to minor tactical adjustments in which statistics plays a huge role.
Just like in baseball, basketball too has basic and advanced statistics. The basic statistics are those that provide an easy measure of judgment and are most commonly used by fans and even pundits, when they need a quick reference. These include the points, assists, rebounds, steals and blocks per game. For anyone acquainted with the game of basketball, the terms are fairly self-explanatory. In addition to this, there are a few others such as field goal percentage, 3-point field goal percentage and the free throw percentage. These are basic stats for individual players while basic team stats are the same things extended to a whole team.
Advanced stats developed as coaches felt a need to assess individual players, the system that the team is playing and the effectiveness of both the individual and the team within that system. The advanced stats ranged from calculating frequencies with which a team ran a particular offense (Phil Jackson, actually used this to slowly embed the triangle offense into his Bulls and Lakers teams. You could often see him on the sideline making the symbol for a triangle when he wanted to run it) to calculating the offensive and defensive efficiency of a particular line-up using the plus/minus rating and the player efficiency rating.
Basketball is a team game but for a large part, still remains a matchup of individuals- with very few but notable exceptions. In other words, as we spoke about in the first part of this series, there is a sense of individuality about the game. Often decided due to physical limitations like size, length, athleticism and speed, matchups involve players who are more or less similar in character and coaches look to try and bring about situations in a game where these matchups are disrupted, forcing players to guard someone whom they normally wouldn’t.
Stats help in deciding which matchups to generate. For example, Karl Anthony-Towns, a 6’11” centre with the Minnesota Timberwolves, is one of the few centres who is mobile, can put the ball on the floor and drive and has a range of shooting that extends to beyond the three-point line. In contrast, apart from Sacramento’s Demarcus Cousins and New York’s Kristaps Porzingis, there is no center who is skilled enough to guard Towns on the perimeter or mobile enough on his feet to adjust to Towns’ drives. Coach Tom Thibodeau of the Timberwolves has looked to use this to his advantage this season. Towns is averaging 23.7 points per game with 34.2% from 3-point range- previously unheard of for a center. He shot a similar percentage last season.
The importance of Towns’ 3 point numbers is that it makes him a legitimate threat outside the perimeter. For Minnesota, it allows them to let Towns go one-on-one against a person like DeAndre Jordan (LA Clippers), the reason being that DeAndre, being far less mobile, cannot react as quickly to Towns’ ball handling as a smaller player would. And more often than not, Towns is going to exploit this to try and get his numbers.
The aim of creating such matchups also decides which statistics are relevant for every team. In every team, matchups are created based on the system a team runs and the system it’s opponent runs. For example, in the 2011 Eastern Conference finals between the Chicago Bulls and the Miami Heat, on defence, LeBron James played the point guard position to guard Derrick Rose. Rose is not a credible threat from 3-point range. However, his penetration and his mid-range game were beyond question (especially in season where he won the MVP award). By putting the 6’8″ James on a 6’3″ Rose, Miami ensured that they had an equally mobile player guarding the shifty Rose. In addition, they had additional size inside when Rose decided to penetrate. Under normal circumstances, the Heat would’ve had Mario Chalmers or Norris Cole guard Rose. However, the Bulls’ lack of a credible threat from the other perimeter positions combined with other favourable matchups on the court allowed for James to deal with Rose.
The system which a team employs also determines which statistics to take and which to ignore. The best example for this is the skewed relationship between the Player Efficiency Rating (PER) and defensive specialists. Player Efficiency rating is a statistic that measures a player’s per minute performance but eliminates the pace at which the game is played- as that differs from system to system. In short, it takes every accomplishment including defensive ones like steals and blocks and calculates this. However, the problem with PER is that it takes only three stats from the defensive side- blocks, defensive rebounds and steals- and a host of them on offence. This disparity has led to many defensive specialists like Bruce Bowen, a former small forward for the San Antonio Spurs, and Trenton Hassell, a former shooting guard with the Minnesota Timberwolves and Dallas Mavericks among others, to have low PERs. What exaggerates this is that these players are regarded as some of the best defensive players even though they didn’t have a lot of steals or blocks. The immediate question that follows is how. One expert says that if a player effectively sticks to his man throughout a game, there are not a lot of good shots that are going to go off from his player- because the latter isn’t going to get the ball that often and even if he does, it’s going to be a tough shot. This is not a statistic that is recorded but preventing a player from getting the ball is an important attribute that goes unnoticed.
To solve this quandary, and in growing evidence of how important stats are becoming in sport, ESPN came up with the plus/minus ratings which determine how many points a player adds to a team when he is on the floor. The obvious problem was that this depended heavily on how his teammates played. And in response to this came the Real Plus Minus Ratings which eliminated the teammates and their factors, allowing us to see exactly what an individual player brings to the table.
While ESPN may have succeeded in solving one problem, with nearly every advanced statistic, there is some problem or the other that remains unanswered. So how does one decide which stats to take? In basketball, the more a stat can be controlled, the more relevant it is. For example, let’s take field goal percentage. Statistics have evolved today to the point where they can show you which area of the court you shoot best from. This allows coaches to design plays and coach systems in which they are able to exploit these particular areas. This is why the Golden State Warriors have a run-and-gun offence exploiting both the speed of players like Kevin Durant, Klay Thompson and Stephen Curry (he may not be the fastest point guard but he has enough speed to make it a legitimate threat) and the deadly three point shooting of Curry, Thompson and Durant. To cite another example, their advantage in the paint with players like Tristan Thompson, is what allowed Cleveland to use more plays that exploited a weaker defense there against the Warriors in last year’s NBA Finals.
This season, with the latest addition of Kyle Korver, who is one of the league’s all-time best three point shooters, Cleveland is hoping to close the three point shooting gap that opened up between them and the Warriors with the latter’s addition of Durant. Another use of a simple statistic like three point percentage and an advanced stat like true shooting percentage to decide team building.
A team’s rate of turnovers is another basic stat that coaches use to assess a team’s offence. And the opponent who forces a turnover uses his team’s average of transition points to judge whether they are a bigger threat from transition or from half-court sets. An example would be the 2013 Portland Trailblazers and the 2013 Indiana Pacers. The Pacers were better in transition while the Trailblazers had more success with half-court sets run by Damian Lillard and LaMarcus Aldridge.
These are just examples on how both simple statistics can help decide how to run a team. I haven’t touched on advanced statistics because I hope you get the drift of the argument.
Video analysis plays a huge part in the NBA. Players and coaches often break down videos to analyse plays run by different teams, how they’re defended, what they can do that other teams cannot and so on. Video analysis of individual players also helps in breaking down certain tendencies and certain moves and improve their game in accordance with what is going right or wrong, both offensively and defensively. Video analysis in the NBA is important in that this provides an easy method of assessment.
Basketball, however, is becoming a little more complex to explain by simple stats alone. One key element that we saw in baseball was the denomination of a single pitcher-batter matchup. In basketball, such a uni-dimensional approach and such an individuality may not work. And there are multiple key reasons for that.
While most of basketball still follows man-to-man coverage and one-on-one matchups, the growing influence of European and international play has led to the development of the zone defence. In contrast to man defence, zone defence is based on ball movement and a reading of the game by the players on court. Simple one-on-one matchups are harder to find here and defensive weaknesses are more effectively hidden. Planning for zone defense with simple one on one matchup gameplans is ineffective because an opponent who plays zone defence relies on non-measurable attributes like basketball IQ, communication and vision- things that don’t show up on a stat sheet but are crucial to the success of the system. The stark example for this is the 2011 NBA Champion Dallas Mavericks, who upset the much stronger LA Lakers, OKC Thunder and the Miami Heat to win their 1st NBA title by using the zone defense almost entirely through the playoffs.
Double-teaming is also an important part of basketball these days. This is especially true for players in the post. In the second round of the 2004 NBA Playoffs, Dirk Nowitzki and the Dallas Mavericks faced off against Chris Webber and the Sacramento Kings. To deal with Nowitzki’s ability to beat Webber off the dribble after the first two games were split 1-1, the Kings put the smaller but more agile Hedo Turkoglu on him. And if Dirk decided to post up, Webber would help Turkoglu double team Nowitzki, forcing the German into bad plays. In Game 3 in Dallas, the Mavericks lost, 125–119; Nowitzki scored only 19 points and said: “I simply could not pass Türkoğlu, and if I did, I ran into a double team and committed too many turnovers.”
Double-teaming is a high-risk, high reward situation. It can lead either into an open shot for the offence or it can be used to force a turnover or at least a bad shot. And while turnovers can be accounted for on a stat sheet, the latter may or may not show up.
And this is true for a lot of defensive stats used in the NBA these days. Statistics have the capability to quantify things and defence in basketball is more than just a numbers game. It is a product of reading an offence, analysing player tendencies, rotating well on defence, communication and many other factors that do not show on a stat sheet and some of which actually require scouting to be effective.
Different studies have also failed to conclusively establish a link between the statistics earned in a college game with that of the NBA game because unlike in baseball, basketball stats have not evolved to the point of predicting future development of a player. This has left scouting as the only reliable technique to enable teams to assess and draft youngsters from both the international and the college arenas. The most famous example is that of Isaiah Thomas, a 5’9″ point guard for the Boston Celtics and a two-time All-Star for the Eastern Conference. Another recent example is that of the New York Knicks’ Kristaps Porzingis.
Moreover, as we spoke about in baseball, the mental aspects of a player- like his determination and his willingness to handle the pressures of being a professional basketball player- can only be assessed by scouting. The methods to do this can range from personal interviews to watching the player interact with others and so on. I’d like to give what is a very relevant example in today’s world. In 2008, the Seattle Supersonics (later the OKC Thunder) drafted Kevin Durant for his scoring ability and the next year, they drafted UCLA’s Russell Westbrook for his leadership qualities and his ability to handle the pressures of being a leader both on and off the court on a young team.
Basketball is a sport that finds itself in transition, both on and off the court. On the court, the European style of play comprising of an increased use of the three point shot (read as Warriors), the pick and roll and zone defense is clashing with the traditional ethos of grit and grind (read as the 1999-2001 Lakers) and man to man coverage. Off the court, the game is being flooded on a yearly basis with statistics- some of which are yet to prove their efficiency in the long run. While stats in basketball are here to stay, it lacks the same sense of purpose and direction that baseball statistics have. What makes the transition harder is that coaches and general managers too are still finding their feet in this new era of computer-based planning. As of now, nobody knows how much more importance statistics will gain through the coming years. But there is no doubt that it will.