Matt Bonner is efficient, but should he take more shots?
I can sense that many of you are thinking, “Oh, no! Not more glowing in-depth Matt Bonner analysis.” But before anyone gets worked up, this post addresses a broader issue. Bonner is simply the Spurs’ poster boy for players who add significant value on the shots they do take, but probably won’t add much value by taking additional shots.
It should not come as a surprise that teams prefer that the player on the court with the best free throw percentage to take technical foul shots. Similarly, it’s typically in the team’s best interests to choose the most efficient scoring option for the bulk of the team’s possessions. However, a look at the leaders in single season Offensive Rating seems to yield more role players than Hall of Famers. In fact, many of these players don’t take many shots at all. Is there a good reason for this?
Visual observation may lead us to conclude that these players who have very high offensive ratings on few shots are very efficient on the shots they do take, but are unable to create additional shots. For example, if the defense collapses on Tim Duncan, Matt Bonner’s open three pointer is a welcome sight; but it doesn’t seem like a good idea to go into a game saying, “let’s get 30 out of Bonner.”
Diminishing Shooting Efficiency of Scorers
Dean Oliver coined a term called “Skill Curves” in his book “Basketball on Paper.” The idea behind Skill Curves is that players typically become less efficient as they take more shots, but certain players are able to maintain their level of efficiency better than others. The problem with trying to find individual player skill curves is that many players do not vary their attempts enough to find any meaningful information. For example, Matt Bonner only attempted at least 15 shots once in 2010. 25 players averaged at least 15 shots per game.
Additionally, there could be bias in selecting these games in which low usage players take a lot of shots. When players make a couple shots, they’re more likely to try to shoot more. This will limit the effectiveness of determining their skill curves. Additionally, they might take more shots against more favorable match ups.
Eli Witus determined the typical slope of a skill curve a couple years ago. In this study, he concluded that for every additional possession a player uses (per 100 team possessions), their offensive rating decreases by about 25%.
This is a step in the right direction. However, using the above conclusion, if we assume that if Steve Kerr had 33% usage instead of 13% in 1996, we would still expect his Offensive Rating to decrease from 141 to 136. In order to get to that rate, shots between 13% and 29% would have an Offensive Rating of 133. These efficiencies would still be superior to Michael Jordan. In mine and the opinion of many observers, this would appear very unlikely to happen in reality.
Estimating Skill Curves
Looking at the initial link, we find that players who take a lot of 3-pointers tend to be high on this list. Additionally, many of these players have reputations for not being able to create their own shots.
In order to test this, I employed the same strategy as Eli, but included six different usage types: assisted close shots, unassisted close shots, assisted 2-point jump shots, unassisted jumpers, assisted 3-point jump shots and assists. Since unassisted 3-point jump shots were rare events, I combined them with unassisted 2-point jump shots. I allocated missed shots, free throws and turnovers to each shot or pass type. I’ll spare you the details of my regression analysis, but I definitely encourage you to look check out my post on basketball-analysis to see how it works).
The results of my regression model most accurately predict the expected offensive rating for all players if they are asked to take on an average workload of a 20% usage rate. But even applying the same slope for other usage rates can yield some meaningful results. Using this assumption, I estimated the Skill Curves of each Spurs player. The following chart displays how player effectiveness expects to decrease for each additional shot a player is asked to take.
According to the chart above, Matt Bonner’s first choice shot is more effective than any other Spurs player’s shot, but his effectiveness decreases the most rapidly. Manu Ginobili and Tony Parker were best able to maintain their effectiveness with an increased offensive role.
The following table details the projected Offensive Ratings (points per 100 possessions) for each Spurs player according to the player’s shot number, sorted by easiest to most difficult.
Spurs Skill Curves
|Shot #||Tim Duncan||Manu Ginobili||Richard Jefferson||George Hill||Tony Parker||Matt Bonner||Others|
Keep in mind that these are just approximated tendencies. It would be similar to saying that DeMarcus Cousins, Derrick Favors and Cole Aldrich project as good shot blockers because of their height, reach, athleticism and block rates in college. It doesn’t always pan out and just as the college block rate for one player might be more meaningful than another, one player might have a smaller slope than projected above. The above Skill Curves just serve as a guideline for a player who might be likely to effectively handle a larger workload or should accept a smaller offensive role.
The table below summarizes some key efficiency estimates for the Spurs players:
Offensive rating is the expected cumulative average of efficiency for each shot. Based on the above results, George Hill probably should have taken more shots last year, and Parker and Richard Jefferson should have taken fewer. Of course, it is reasonable that these players took such a high percentage of the team’s shots after considering how efficient they were in 2009. Although RJ had the same Offensive Rating in 2009 as in 2010, his offensive rating in 2010 would have expected to be five points lower if he had been utilized as often as he was in 2009. Tony Parker’s Offensive Rating was effectively nine points worse than last year (after adjusting for usage) and George Hill’s rating was 14 points higher. Matt Bonner himself appeared to be utilized fairly effectively.
Applying these estimated skill curves can help to bridge the gap between perception and statistics. There may be a good reason that the player taking the most shots isn’t necessarily the team’s most efficient player. While efficiency may be one of the most important factors on offense, the ability to create a high level of efficient shots is also an essential component.