Pages Navigation Menu

Expected Points: Team Edition

This is a follow-up to my Shot Selection and Efficient Scoring post from a few weeks ago. In that post I used the Expected Points Per Shot data calculated by Albert Lyu of ThinkBlueCrew to look at the scoring outputs of several of last years top scorers, from various locations on the floor.

After looking at these numbers for individuals, I wanted to go back to last season and take a look at the numbers on a team level. I know everyone is ready for the new season to start, moving ahead with previews and forecasts and the like. Call me late to the dance, but I still have some analysis of last season that I haven’t worked out of my system yet.

 Here is the explanation of my technique:

Another way to think of Expected Points Per Shot, is the average number of points scored on a shot attempt. For example, over the last 4 NBA seasons, factoring in makes and misses, a field goal attempt at the rim was worth an average of 1.208 points.

Using numbers from Hoopdata’s Shot Location Database, and the Expected Points Per Shot from Lyu’s post I was able to calculate what I am calling Expected Points per 40 Minutes (XPts/40). I began by calculating each players XPts/40 from each area of the floor. To do that I took each player’s per 40 minute field goal attempts from each area of the floor and multiplied it by the expected Points Per Shot for that location. Adding these categories together results in XPts/40. Another way to think about this is, given a player’s per 40 minute shot selection, how many points would he score, shooting the average percentage from each location. The numbers I borrowed from Lyu are below:

  • At Rim – 1.208
  • <10ft. – 0.856
  • 10-15ft. – 0.783
  • 16-23ft. – 0.801
  • 3PT – 1.081
  • I needed to include Free Throws, so I used 0.759 the league average for last season.

In this post we are going to use the same technique and apply it to each team’s per game shot selection from last season. This whole exercise is just another way of looking the relationship between a team’s FG% and the league average from different areas of the floor. I like using the statistic of Expected Points as a more explicit example of what teams are losing or gaining with their shooting percentages. Instead of expressing the difference in FG% points, we can express it in actual points, making the results a little more tangible. Below is the Expected Points Per Game, Actual Points Per Game, and Points Per Game Differential for each team last season.

These numbers are not pace adjusted, so there is not tremendous value in comparing the XPPG and Act. PPG numbers. I do think that the Point Differential is a great illustration of a team with an efficient offense. The next graph shows the same numbers but broken down by floor location.

A few thoughts:

  • Only 10 of the 3o teams scored at a rate below expected on shots at the rim. However 6 of those 10 teams, (Charlotte, Chicago, Detroit, Houston, Milwaukee, New Jersey) were significantly below expected, a point per game or more. The different between the best team in the league, Cleveland, and the worst team in the league, Houston, was a total 5.149 points per game.
  • Oklahoma City was scored at a rate very close to expected on shots close to the basket. They were noticeably below average on long jumpers but almost completely offset this loss of points with their terrific free throw shooting, picking up an extra 1.227 points per game more than expected at the line.
  • Dallas’ efficiency on long range jumpers stands out remarkably in this analysis. They scored 1.418 more points per game than expected on jump shots from 16-23ft. This number is simply incredible. Only 9 of the other 29 teams scored more than expected on shots from this range last year. Dallas’ number is almost a full point better than the next two closest teams Toronto and Golden State. This amazing efficiency can be largely attributed to Dirk Nowitzki, Jason Terry and Caron Butler, who all shot better than 45% from this area on more than 4 shots per game.
  • Phoenix’s excellent 3PT shooting was worth an average of 3.350 extra points a game versus the expected rate. The Knicks were among the league leaders in 3PTA per game, but their poor shooting cost them an average of 1.022 points versus the expected rate.
  • Teams with up-tempo offenses, Phoenix and Golden State, were among the leaders in Point Differential. They were not alone, as teams with slower more deliberate offenses like Boston, Orlando and Utah also scored significantly more than expected.

Having now looked at this statistic from an individual and team perspective, I was curious what the relationship might be between the two. It would seem that certain players are responsible for large portions of their team’s performance in this metric. For example:

  • Last season Dirk Nowitzki scored 3.255 more points than expected per 40 minutes. Playing 37.5 minutes per game, he’s almost completely responsible for the 3.566 points per game more than expected that Dallas scored last season.
  • In the least surprising analysis of this post, Kevin Durant was very important to the Oklahoma City offense last season. Durant scored 2.712 more points than expected per 40 minutes. Without Durant’s scoring output, in his 39.5 minutes per game, the Thunder would have scored 2.3654 points less than expected per game. Both the quantity and quality of his free throw shooting contribute heavily to this scoring advantage.
  • Tyreke Evans had the lowest differential in my individual player analysis, scoring 1.9247 fewer points per 40 than expected. Playing 37.2 minutes per game last season, Evans bears a large share of responsibility for Sacramento scoring 2.070 fewer points per 40 than expected. I am not implying that Sacramento would be better off with someone else on the roster taking shots from Evans, but his poor shooting from several areas on the floor cost his team almost 2 points per game.
  • The Phoenix Suns were the team with the biggest positive difference between their expected points per game and their actual points per game, 7.264. Steve Nash’s shooting is responsible for nearly half of this total, as he scored 3.331 more points per 40 than expected. It goes without saying that his ability to create quality shots for his teammates is likely responsible for the rest of the Suns’ scoring more than expected.

As we head into next season I’ll be trying to keep these numbers posted, current and updated on a fairly regular basis. It will be interesting to watch the performance of both individuals and teams in this area. I am guessing that some of the player movement from this summer could have a huge effect.

%d bloggers like this: