April 12, 2017 by J.H. Yeh
Deflections are a defensive metric that the NBA started to keep track of in the beginning of the 2013-14 season. This metric not only tells the whole story of a player’s defensive ability but it also conveys key information that is not shown on the traditional box score.
I compiled a “mini scouting report” that shows the relationship between steals and deflections and whether a defender is a “stealer” or a “deflector.” Below is a list of players’ steals and deflections per game. Only the top 100 players who have played at least 41 games and averaged more than 15 minutes are qualified to be on the list. The list is sorted by steals per game. The deflection per game stat is courtesy of NBA’s player tracking and is as of April 11, 2017.
The list is mostly dominated by wing players and guards. There are some power forwards and centers that also made the list. The mean is 1.26 steals per game and 2.45 deflections per game. Robert Covington leads the pack in deflections and is the only player to average more than 4 deflections per game. Draymond Green, Kyle Lowry and John Wall are just right below 4 deflections per game.
Now in order to yield better results on a player’s effectiveness in producing steals and deflections, I divided those numbers by their minutes to get steals per minute and deflections per minute. The per minute stats are scaled to five decimal places for a very precise look. The list below is ranked by steals per minute. Philadelphia 76ers’ TJ McConnell leads everybody in this category.
I also sorted the ranking for deflections per game and deflections per minute. The top two players in deflections per minute are 76ers’ Covington and McConnell (and former 76ers Nerlens Noel is at fourth place). Minnesota Timberwolves’ Kris Dunn is at third place with 0.13018 deflections per minute. Established veterans such as Tony Allen and Draymond Green are also in the top ten.
Now that I have the per minute stats, I plotted the numbers on the scatter chart to have a visualization of each player’s defensive styles and tendencies. With a high R-squared at 0.8114, the regression line validates that both steals per minute and deflections per minute are highly correlated. TJ McConell is at the top as he has very high steals and deflections per minute. Covington, Jimmy Butler, and Noel are also at the top. Surprisingly, Manu Ginobili, at the age of 39, is also near the top of the pack — this shows he can still be highly productive in limited minutes.
The linear regression analysis can explicitly tell us the players’ defensive tendencies. It is obvious that we prefer players who are at the top because they steal a lot and also deflect a lot. However, as we move down the trend line, we can observe different players with different defensive tendencies. Players who lie below the line steal more than they deflect — these are the type of players in which steals alone over-estimate their deflections. On the other hand, Danny Green, Kyle Lowry are the opposite as they deflect more than they steal.
Personally, I don’t think there is a preference for one over the other — or favoring “the stealer” over “the deflector.” This is only a matter of a player’s playing style; it is equivalent to deciding between an offensive player or a defensive player. Everybody prefers a two-way player but these players are hard to come across. Therefore, often times we fill up the roster with players who are better at one side of the game over the other and can complement with the rest of the team.
To sum up, the visualization gives a very interesting look at the different types of defenders (when it comes to steals and deflections). We can easily tell a defender’s defensive style (a stealer or a deflector) by understanding how they approach their defensive assignments. Furthermore, this research is also similar to a research I did back in February, in which I determine if a player can effectively defend the rim by blocking a lot of shots or by blocking very few shots (or by blocking a lot of shots but INEFFECTIVE at defending the rim). The full articles are here (part 1) and (part 2).
— J.H. Yeh
(all tables and graphs are created by me, and all sources of stats are courtesy of NBA.com as of April 11, 2017)