The Games Played Conundrum
The second in a three-part series on how we think about the NBA Awards
If you made it through Part I, “On Criteria,” congratulations. You’ve only completed one of three articles, but you are roughly halfway through this three-part series.
In that 3,000 word dissertation, I examined all seven major NBA Awards, and how we ought to think about them if we want accurate, deserving and well-aging winners. Today, I will not go in-depth with each award, but rather focus on a major area of contention that affects all of them: What value do we place on attendance?
Humans pushing their bodies to the level that NBA players do will get hurt with regularity. As a result, our annual NBA Award debates always include players who have missed significant time. We know we must factor games played into our thinking, but we don’t always know how—even when we think we do.
How We Tend To Think About Games Played
Everyone has their own idea of how to weigh games played vs. performance, but how much weight an individual places on either side of the scale is based on where they land on the following issues:
- Games Missed Demerit: The Slope vs. The Cliff. The slope is exemplified as follows: Player X giving his team 82 games of 90th-percentile play is equal to Player Y giving his team 72 games of 95th-percentile play, which is equal to Player Z giving his team 62 games of 100th-percentile play, with equal intervals in between. Individuals will vary in how steep the games missed slope is, but this is the general structure. Others draw a line where they believe insignificant time missed ends and significant time missed begins. Player Y giving his team 72 games of 95th-percentile play is still better than Player X giving his team 82 games of 90th-percentile play, but choosing Player Z and his 62 games — even at his 100th-percentile play level — would be a bridge too far. This is the “cliff” approach.
- Best Player vs. Most Valuable. This is most frequently used as the distinction between the NBA Awards and the All-NBA Teams. The awards are seen as having more to do with attendance (which in theory increases value), while the All-NBA Teams are considered to more about who was best in the games they played. But this dichotomy is applicable to every award, and every voter comes to their own conclusion on it.
- Minutes Played vs. Games Played. This is a sneakily complex distinction. Those who look at minutes played are essentially contesting the notion that simply stepping on the court for one minute is worth anything more than just that: one minute. If a guy plays roughly 2,100 minutes, who cares whether those we distributed across 78 games (Jusuf Nurkic) or 58 games (Jimmy Butler)? Meanwhile, those who look at games played would argue that there is inherent value to appearing in a game beyond minutes played. Stephen Curry has famously sat out myriad fourth quarters during the Steve Kerr era, but only because his present attendance put those games away early.
How We Should Think About Games Played
I’m going to let you in on a little secret about The Slope vs. The Cliff demerit: Both sides are equally wrong. The wrongness has nothing to do with the slope or the cliff concept, but with the idea of either being a demerit.
Now I’m going to let you in on another secret: By proving that time-missed-as-demerit is a false notion, I can also prove that Best Player vs. Most Valuable and Minutes Played vs. Games Played are false dichotomies.
Games or minutes played should not be viewed as a stat akin to points, rebounds, assists, true-shooting percentage, usage rate and DRPM. It should not be viewed as the dark side of a player’s award resume, the counter argument to the player’s otherwise-strong case.
Rather, time on court should be the framework in which everything a player does is viewed.
Remember back to Part I, where we established that the primary goal of NBA Awards is to reward those individuals who have had the greatest influence on team success. Therefore, every candidate’s season should be measured foremost by their total impact on their team’s record.
For example, Chris Paul has played for 1,819 minutes across 57 games this year. Rather than viewing that as a demerit, let’s try and view it for exactly what it is.
The Rockets have gone 49–8 with Paul on the court. That means that, in 57 games, he contributed to 49 wins. Russell Westbrook has played for 2,877 minutes across 79 games this year. In those contests, the Thunder are 46–33.
Don’t worry for now about what each player’s contribution was to those wins. That is a separate exercise. It is entirely possible that Westbrook’s contributions to his team’s 45 wins are far greater than Paul’s to his team’s 49. That isn’t the point.
The point is the 46–33 and the 49–7. Rather than saying “The Rockets are 64–16, and the Thunder are 47–34, but Paul has missed 22 more games than Westbrook,” all we need to say is “The Rockets are 49–7 with Paul, and the Thunder are 45–33 with Westbrook.”
I’m using the most basic stats to illustrate the approach, but you can do this more accurately. You can look at point differential and minutes (The Rockets have outscored opponents 4,315-3,918 in Paul’s minutes, while the Thunder are up 6,501-6,124 in Westbrook’s minutes, per nbawowy!) or net rating in non-garbage time possessions (Houston is at plus-12.6 with Paul on the court, while OKC is at plus-5.5 with Westbrook, per Cleaning the Glass).
It doesn’t matter how you slice it: We now have the proper framework for how to compare Paul and Westbrook’s seasons.
From here, we can determine each player’s respective impact on their teams success. We can look at box score stats, on/off data, eye test, etc. But we are starting by framing a player’s season in the exact appropriate way, no subjective cutoffs or demerits needed. You simply get credit for what you did on the court.
At this moment, it may seem like this system benefits guys who play less. They are often better rested (particularly those missing time due to DNPs and reduced minutes rather than injury), and may get lucky enough to compete against a weaker part of the schedule or play in more favorable situations game-to-game.
However, so long as you operate by the chief tenant, this is irrelevant.
Remember, we are not crediting guys for any time they missed, or projecting their performance out over a full season. We are not saying “Paul was this good in 57 games at 31.9 minutes, so he would have been just as good in 79 games at 36.4.” Rather, we are acting as if Houston would be 49–33 had Paul played all 82. We are only giving Paul credit for the positive contribution he actually made, just as we are doing with Westbrook.
To use a famous example, many rallied against Joel Embiid’s 2016–17 Rookie of the Year candidacy due to him not playing “real basketball.” The argument was that, by skipping back-to-backs, playing 25.4 minutes and only appearing in 33 games, Embiid’s numbers should be taken with a grain of salt. They were easier to achieve and harder to sustain.
This logic is not wrong — the grind of an NBA season is real. The problem is that this awareness put Embiid on the wrong side of an arbitrary cliff, one that he may have been on the right side of with 10 more games. Or 15, or 25, or 30, depending on where people drew their subjective lines.
The right approach was not to punish or disqualify Embiid, but rather to take his contribution for what it was: 33 games and 786 minutes of high level basketball. His impact on the Sixers in those minutes should have been analyzed and contrasted with Malcolm Brogdon’s impact during his 78 games and 1,982 minutes, and whoever impacted their team’s overall record more would become clear without even needing to worry about games or minutes.
(Note: This approach can be thought of as more robust in approach, but similar to Win Shares in concept. We are giving a player credit for what they do on the court, and neither punishing them nor falsely crediting them for time missed).
Arriving at a conclusion about a player’s overall impact in the time they played is not the be-all and end-all of determining awards. There are several other criteria to consider.
For example, I discussed in Part I the importance of looking at whether a rookie’s impact was more translatable to all-star level success down the line, or whether it seemed more based on favorable fit and circumstances. Embiid would obviously destroy Brogdon in this criterion, and thus could still fairly earn your ROTY vote even if Brogdon had the edge in absolute impact.
We do not need to agonize over the Games Played conundrum. Really, there is no conundrum. All we need to do is credit players for what they have done in the games and minutes they spend on the court. It takes a little more statistical leg work than just speaking from the gut and saying “55 games is just too few” or “an eight-game difference is negligible,” but it will lead to more accurate and less arbitrary voting.
Check back tomorrow for Part III, in which we will address the most complicated question surrounding the NBA Awards: Who gets the credit?