The NBA recently announced the addition of a new performance award for the 2022-23 season: the Jerry West Trophy, to recognize the NBA Clutch Player of the Year. The NBA has yet to release many details about the award. We do have this not-at-all-scripted quote from the award's namesake, that the NBA included in their announcement:

There are few greater thrills as a basketball player than coming through for your teammates and fans when they need it most. This new trophy will be awarded to the player who best delivers in those moments.

So, armed with this guidance, how would one try to quantify "coming through for your teammates and fans when they need it most"?

The nice thing about a win probability model is that it allows you to quantify exactly what "when they need it most" means. For many (many) seasons, I have published player-level win probability added stats on this site. The basic idea, which I did not invent, is to measure how a player's actions impacted their team's win probability.

And in general, this aligns with what most would consider "clutch" performance. Clutch performance is associated with big, needle-swinging win probability plays. Its antithesis is garbage-time stat padding, where a team's win probability has already flatlined. A win probability added approach properly inflates the former and ignores the latter.

That being said, creating a fair and comprehensive win probability added metric for the NBA is not straightforward. My official definition only includes the impact of the following: made/missed field goals, drawing fouls, made/missed free throws, and turnovers. This roughly aligns with how we define "usage" in the NBA, and I chose this approach because it felt "symmetric" in that a player could get WPA "credits" (making field goals, drawing fouls, etc.) as well as "debits" (missing field goals, turnovers, etc.).

But this is far from a comprehensive measure of player performance. There's a reason it's the eighth most trusted player metric in the NBA. So, for the tool I am launching today, it will allow the user to create their own bespoke version of a win probability added metric, bringing in all available stats from the play by play logs: field goals, drawn fouls, free throws, turnovers, assists, rebounds, blocks, and steals.

While win probability added better aligns with clutch performance than traditional box score stats, it's not quite a one-to-one correspondence. For example, hitting a three pointer in the middle of the 2nd quarter will improve your team's win probability (typically +3-4%), but is not what one would consider clutch performance.

To zero in on true clutch performance, we need to narrow our focus to situations where the win probability impact is outsized, relative to a typical game situation. For example, when Jerry West hit a 60 foot buzzer beater in the 1970 NBA Finals to send the game to overtime, the impact of that shot was roughly +50%, far above the typical +2% bump you get from a normal two pointer.

You can find a detailed explanation of Clutch Win Probability Added here, but in a nutshell, it measures only "excess" win probability added, over and above a typical game situation. For example, a turnover typically costs a team -2.1% in win probability. But if a player turns the ball over with 19 seconds left with their team down by 1, the win probability impact is much larger: -20.8%. In this situation, the clutch win probability added (or subtracted, rather) would be -18.7%.

And conversely, if the turnover took place with your team up 24 points in the 3rd quarter, the win probability impact is -0.5%. Clutch win probability ignores this play because -0.5% is smaller than the typical -2.1% value.

I have compiled typical (i.e. median) win probability values for all play types, they are summarized below. If you want to know how much (and if) a play qualifies for Clutch Win Probability Added, just consult this table.

play typeeWPA
turnover-2.1%
missed field goal-1.6%
missed free throw-1.5%
getting 1 foul shot1.6%
getting 2 foul shots1.0%
getting 3 foul shots2.7%
defensive rebound0.7%
offensive rebound1.6%

### The Jerry West Clutch Player Tracker

Hopefully, this is all straightforward. You have your choice of stats at the top, and you're free to enter the weights that best align with your definition of clutch performance. The default weights correspond to my current definition of win probability added.

Don't like giving a player credit for drawing fouls in clutch situations? Enter a lower weight into the "DF" column (or zero) and click the calculate button.

Want to include "clutch" assists? Change the "AST" column from zero to a weight of your choice (note that applying a weight of 1 here would likely overvalue the contribution assists, but you could probably make a case for a weight of 0.25).

Using the default definitions, we see that DeMar DeRozan has a healthy lead in Clutch Win Probability Added. Note that DeMar would have been the runaway favorite for this award last year

This tool will update daily. Let the arguments begin.