When it comes to individual awards, like NBA MVP, are we valuing contributions to wins correctly?: Partnow

For as long as I can remember, the Christmas Day slate of games has represented the start of the NBA season in earnest. There is enough of a sample on every team to have a sense of the degree to which preseason hype has translated into justifiable aspirations for the season. While the first 30ish games might tell us a little more about teams, it’s still sufficient to have a decent idea of who is really bringing it this year.

As such, it’s just around the time the annual awards talk starts in earnest. Really, we tipped off with this season’s first installment of an MVP straw poll last week. Full disclosure, I was among the voters, with a top 5 of:

  1. Jayson Tatum
  2. Giannis Antetokounmpo
  3. Steph Curry
  4. Luka Dončić
  5. Kevin Durant

I made my submission before Curry’s recent shoulder injury, which would have dropped him from my top 5 with either Zion Williamson or more likely Nikola Jokić as the replacement. While this largely aligned with the overall results, there is a vociferous minority wondering how one could not have either Curry or Dončić at the top, given how absolutely essential each is to their teams.

Certainly, the last week of Golden State’s play lends some credence to that argument. A 1-3 record and a league-worst 17.0 non-garbage time net rating over that span (per Cleaning The Glass) are certainly illustrative of the troubles the non-Curry Warriors face. But the 1-6 record with Curry not in the lineup only proves so much.

Golden State is only 14-12 with Curry, a 44-win pace. Their +2.7 average net Rating in those games suggests slightly better performance, but only by a win or two. Similarly, Dallas is 0-3 in games Dončić has missed, which still leaves them only 16-13 (45-win pace) with a +3.0 Net Rating in the games he’s played.

I bring those numbers up in the context of one of the great schisms in these award contests. How much should team success matter?

On one hand, some argue that for an individual award, we shouldn’t blame a player for having worse team members than another candidate. On the other side, in basketball, a singularly talented individual can exert perhaps the highest degree of influence on wins and losses of any major team sport. Further, great players on mediocre teams might benefit from a degree of “double-counting” in the “How bad would they be without him?” line of thought. It is not that Curry and Dončić are putting up incredible individual statistics and their teams fall apart without them. Rather, the lack of overall talent aside from those stars is (in part) the cause of both those gaudy totals and the gob-smacking on/off differentials.

(We could also bring up the notion that awards represent an attempt to quickly tell the story of a season, and a player on a team without much relevance to the title picture doesn’t really fit that picture, but that point seems unlikely to sway those who have already emphasized the “individual” in “individual awards.”)

As should be somewhat clear from my straw poll submission, I believe team performance should be included, but lack thereof should not be disqualifying. Part of my reasoning is I think we are valuing wins all wrong.

From the standpoint of players creating wins for their team, analysis has largely treated wins as fungible and linear. This assumption of all wins being equal is largely out of necessity. Modeling player value is hard enough already! But that doesn’t make it true. Much like the Player Tiers, where a single Tier 1 player is more impactful on ultimate success than several Tier 3 players, wins 51-60 are more impressive than wins 41-50, which are more impressive than wins 31-40 and so on.

This is implicitly understood in large part; going from a bottomed-out lottery team to making the playoffs is easier than proceeding from that playoff appearance to becoming a legitimate title contender. We know win 60 is tougher than win 50, but how much so?

In looking into this question, I took every NBA season where 30 teams have played 82 games, which means 2004-05 to 2018-19 as well as 2021-22. I gave a team’s opponent in each game a “win value” corresponding to the number of wins that opponent ended the season with; last year’s Suns would have a win value of 64 while the Rockets would have been 20. I then ordered every teams wins from “worst” — coming against the lowest opponent win value — to “highest.”

Taking the average of opponent win values for each successive win starting a 1 and ending at 65+, I found that as the number of wins increase, the opponent difficulty of the marginal win increases at first rapidly, but slows down after a team hits 40 or so wins:

Of course, this is an imperfect way of looking at it. In practice, teams don’t “use up” wins against bad opponents completely before advancing to the next level. But the number of wins against terrible opponents dries up pretty quickly.

I divided the same sample of teams into bands of seven wins, centered around a .500 team at 41-41. That middle group include teams finishing with between 38-44 wins. The next worst band included 31-37 win teams, while the next best was 45-51, and so forth. I then looked at the average number of losses a team in any one band had against teams in each of the bands to identify where a team at a certain level had the most opportunities to pick up additional victories:

So a truly wretched, “we hope to have good lotto luck” team loses around 4.5 games to other similar teams on average while dropping just more than 26 to teams winning between 38 and 51, and so on. One of the things this illustrates to me is how swiftly the well of available wins against really bad teams dries up. By the time a team hits solid playoff status of around 45 wins, there might be a game or two to pick up by not taking their eyes off of the ball in a back-to-back against a tanking squad, but other than that, they are already taking care of that business. By contrast, they still have around 11 more chances to dig out an extra win against the contender-level 52+ win teams.

The upshot is that while teams going from 25 to 40 wins pick up those extra wins against, on average a 45.5 win opponent, to then progress from 40 to 55 means beating 53.1 win teams on average. In other words, a team you might face in the first round of the playoffs compared to a conference finalist.

While all of the above is pretty crude methodology, and I haven’t really solved the question of how much more is the 50th win “worth” than the 40th, I think it does have some bearing on why I believe it is fair to reward a player more for bringing extra wins to a better team. I can’t prove that a Jayson Tatum or Giannis Antetokounmpo driving a 45ish-win team to 55 or 60 is doing more than Dončić taking an otherwise 25-win roster outside of him to 45, but it’s at least a reasonable argument to make.

(Photo of Jayson Tatum and Giannis Antetokounmpo: Adam Glanzman / Getty Images)

Leave a Reply

Your email address will not be published. Required fields are marked *