What Does 20.1 WAR Actually Mean?

Nikola Jokić was worth 20.1 wins above replacement in 2025-26. If you follow the public win metrics, that number may have made you spit out your coffee. On the familiar public scales, a great season reads somewhere around 10. Is the model broken? Is Nikola Jokić secretly two people?¹

Neither. It’s a different ruler. Here’s how to read it.

Replacement of what?

“Wins above replacement” is only as meaningful as the replacement. Here it’s the kind of player any team can sign in an afternoon: the end of the bench, the two-way contract, the veteran on a minimum deal who is always, somehow, available. The model puts that player at −1.77 points per 100 possessions against an average player, and it estimates the level rather than assuming it: across 37,895 games, every point of a lineup’s expected margin moves its win probability by 0.045.

Put five of those replacement players on the floor for a whole season and they win about 8 games. That’s the zero on the ruler. An average team wins 41, so an average roster has to supply about 33 wins above replacement between all of its players.

Now reread the headline. Nikola Jokić, by himself, supplied 20.1 of them. That is 61% of what an entire average roster produces above the waterline.²

How a season becomes a number

The recipe has three ingredients: how good you are per possession, how much you play, and where replacement sits.

Nikola Jokić rated +7.57 per 100 in 35.1 minutes a night over 65 games. That’s 9.3 points per 100 better than replacement, sustained across about 2,282 minutes. Nobody who played regular minutes rated higher.

The same arithmetic explains why a brilliant player who misses half a season can finish behind a solid starter who never sits. Giannis Antetokounmpo posted the 4th-best rating among regulars, +5.13 per 100, and finished with 6.8 WAR, because he played 36 games. WAR is a quantity, not a quality. It’s how much, not how good.

What a normal number looks like

For calibration: the typical rotation player, meaning 20 or more minutes a night, produced 2.5 WAR. 13 players cleared 10. Only two cleared 15: Nikola Jokić and Shai Gilgeous-Alexander. And 53 players finished below zero, which is the model’s polite way of saying their teams would have done better with whoever was available that afternoon.

What it’s worth

The last step is money. Estimated from 2,542 contracts, the open-market price of a win above replacement was about $10M last season.³ By that price, Nikola Jokić’s season was worth $192M. The entire salary cap was $155M. He was paid $55M.

Read that again, because it isn’t a typo: the model thinks one player was worth more than a whole team’s cap. That is not the model going feral. It’s the maximum contract doing precisely what it was built to do, which is make sure the very best players are never paid what they’re worth. DEN kept the difference: +$137M of surplus.

So when you see 20.1 WAR, read it like this: 61% of an average team’s wins, from one person, at about 29% of his price. Pretty good player.

Figures: the 2025-26 season on the retrospective players board, as of this build. Wins are on this model’s own scale and dollars at the open-market price of a win; the method explains both.

1. Unproven. ↩
2. Public metrics tend to set replacement higher, closer to a real if marginal NBA player, so there are fewer wins to hand out and every number comes out smaller. Neither choice is wrong. A ruler marked in centimetres is not lying to you about your height. ↩
3. “Open market” is doing real work in that sentence. Salaries are capped, so what teams actually pay understates what they would pay. The model reads the true price off teams that pay the luxury tax, where signing a player costs the salary plus the tax, and a team only does that if the player is worth at least as much. The method page has the estimator, and the honest caveat that a reader who rejects the argument should shrink every dollar figure on this site. ↩