NBA Four Quarters (02/10/17)

By Sean Pyritz


No game represents the beauty of the NBA better than this game. You never know when the best game of the year might happen. On a seemingly random Monday night in February, a super-skilled fireworks show erupted between the Eastern Conference's two top heaviest teams. Overtime was needed to settle the official victor but we the fans were winners long before the final buzzer sounded. Both teams scored nearly 1.30 points every possession (the league leading Warriors score 1.16 points per possession for reference). Combined, both teams poured 31 of 67 three-pointers (46.3 %) and dished out 60 assists on 99 made baskets. As evenly matched games often do, the final came down to which stars made more plays. On this night Kyrie Irving scored nine of his 23 points in the last six possessions of overtime, while Bradley Beal missed two three-pointers in the last minute. I would love to see a similar sequence of events play out in a playoff series between these two.

Player of the Game - LeBron James

Before fouling out in overtime, James put forth a masterful performance scoring 32 points on 12-18 shooting, including 6-8 from long distance. He also chipped in seven rebounds, two blocks, two steals, and a career-best 17 assists. James was in full wizard mode, stupefying Washington defenders with a steep array of skips, dump-offs, and pocket passes. His relentless attack of the paint turned Marcin Gortat into the Polish Nail. And when his team needed him most as the Wizards were crushing at the end of regulation, James buried four three pointers in the final four minutes including what may have been the most amazing play I've ever seen live - his fadeaway bank-in three off the full court pass to tie the game with 0.3 seconds left in the fourth quarter. See below.

Play of the Game



Magic on any given night in the NBA. Need I say more?


Skeptical Statistics - Possessions

In practice there are only four ways for a possession to end: a made field goal, a made free throw, a defensive rebound, or a turnover. Using the end results of possessions hidden in the box score, we can estimate the number of possessions in a given game. That is exactly what the major NBA statistics sites like Basketball-Reference.com (B-R) and Stats.NBA.com do. Let's break down the four components and some of the nuances of the estimation process.

(1) Field goal attempts - this is universal, all formulas include field goal attempts as the starting point for calculating possessions.

(2) Free throw attempts - free throws are tricky because the box score does not distinguish between a pair of free throws for a routine shooting foul or an and-one free throw after a made basket. Ideally we are looking trips to the free throw line. But, without simple access to that information, an estimate is made using a fraction of free throw attempts. B-R uses Dean Oliver's formula: 0.4*FTA. NBA.com uses 0.44*FTA.

(3) Rebounds - there are actually three types of rebounds, two of which can end a possession (offensive rebounds are considered an extension of an existing possession). The first is obviously a defensive rebound. The second is the lesser known team rebound. An example of a team rebound would be a shot that clangs off the rim out of bounds, changing possession to the other team - no one gets credit for the rebound so it counts as a team rebound. Including team rebounds would require additional work digging into the play-by-play data, so NBA.com ignores them and simply subtracts offensive rebounds in its possession formula. B-R again defers to Dean Oliver's method of estimating total rebounds into the formula. It is this distinction that makes the B-R calculation slightly more accurate.

(4) Turnovers - turnovers are added as the final component to the possession formula. A slight difference occurs here between the two sites. B-R only includes turnovers attributed to individuals in the box score while NBA.com includes team turnovers like 24 second violations in its definition of a turnover. This difference is small, usually one or two per game.

In one last difference between the two methods, B-R takes an average of both teams' estimated possessions and uses that as the number of possessions in the game for all possession based statistics. On the other hand, NBA.com keeps the team specific possession estimates separate.

As I have alluded to over the past two weeks, I have been tracking games manually and have two measures of possessions. The first is the raw possessions (RAW) that should correspond to the estimates on these websites. The second is my personal adjusted possessions, accounting for and removing what I defined last week as non-possessions, which I will call SPos (Sean's Possessions).

In the table below I compare the possession estimates to my own from the eight games I tracked by hand. I have included both the averaged B-R possession estimate for the game and the possession estimate for each team (B-Rtm) for more accurate comparison. We can clearly see that the NBA.com estimates are alwasy higher than those from B-R, by as much as three possessions in some instances. Also, the B-R estimates are closer to my raw tracking numbers, confirming the superior accuracy of B-R over NBA.com. Comparing my adjusted possessions to the NBA estimates there are some serious differences, as much as 9.4 in a single game.





Next week we'll examine some of the implications of these estimates on possession based statistics.




In my preseason preview, I picked Dario Saric as my Player to Watch in the Atlantic Division. If it weren't for his massive teammate, Joel Embiid, he could easily be in the Rookie of the Year conversation. Amongst all qualified rookies, Saric is top ten in points, rebounds, assists per 36 minutes. He has strengthened his case over the past week in Embiid's absence including Thursday's win over Orlando. While the Sixers went 1-3 in that stretch, Saric averaged 15.3 points, 4.8 rebounds, 2.0 assists, and 1.3 steals per game off the bench. The most promising trend? The 44.4 % he shot from three-point range in those four games. Improving his jump shot will be an important part of his coexistence with the Philadelphia core of Embiid and Ben Simmons. Monitoring how/if the Sixers use Saric with Simmons at the end of this season will be fascinating to watch for those with the long-term in mind.


Getting Buckets

Currently in the NBA, there are 35 players averaging at least 20 points per game. At this point last season, only 23 players were scoring at such a rate. By the end of the year last year, only 20 players who qualified for the points per game leaderboard eclipsed the 20 point marker. In fact, looking back over the past 20 seasons, the most 20 point scorers in any one season came in 2007-08 when 27 players achieved that milestone. We are witnessing a historic scoring season due to a historic collection of young, dynamic talent in the league all at once. The NBA is in a good place and very well-positioned for the future when it comes to star power. Heck they might have to put the NBA logo on the new $20 bill if this continues.