The Best and Worst Contracts, Then and Now

In the summer of 2004, Dallas let Steve Nash leave. Phoenix signed him, and he won the next two MVPs. That contract returned $783M more than Phoenix paid for it, in today’s money, the most of any NBA contract since 1999-00.

Every July, every signing gets a grade within the hour, and almost nobody goes back to check. So here is every negotiated contract since 1999-00, checked: the best and the worst over their whole lives, then the best and the worst of the ones still running.

The rules

A contract’s value is what the player was worth on the open market, season by season, minus what he was paid.¹ Each season counts in today’s money, by its share of that season’s cap, so a dollar of 2003 cap space counts the same as the same share of today’s.² Rookie-scale deals are left out, because the league sets those, not a negotiation. An extension counts as a new contract, starting the season after it was signed.³

For a contract that is over, that’s a record. For one still running, it’s a forecast: what the board expects the player to be worth over the seasons he has left, against what he’s owed, with player and team options priced the way they get used. The tables also show what his club can expect from the contracts after this one, each counted only with the chance he re-signs there.⁴

For a baseline: the typical negotiated contract since 1999-00 returned $8.7M more than it paid, and two in three came out ahead.⁵

The best, then

In today’s money. † began in 1998-99, the season before the board starts; that season isn’t counted.

Four of the ten are extensions: deals signed by a team that already knew what it had, before the player could find out what the market would pay. The best rate on the list belongs to one of them. Stephen Curry’s extension, signed in 2012 when his ankles were the worry, returned $180M a season over its four.

The worst, then

The bottom of the list is mostly one era. Nine of the ten worst started before 2004, and 28 of the worst 50 before 2005, from a stretch that signed 19% of the contracts. Deals were longer then: 3.4 seasons on average before 2005, against 2.7 since 2011. A long contract has more seasons to go wrong.

Not every name here is a bad decision. Grant Hill played fewer than half of Orlando’s games in four of the seasons on his deal and missed another entirely. That isn’t a scouting error; it’s an insurance claim, and the list charges it to the contract, not to him.

The best, now

Forecast over the seasons left on each contract, in today’s money; with re-signing, plus the contracts after it, weighted by the chance he signs them with the same club.

Shai Gilgeous-Alexander leads by a distance: $394M over the five seasons left on his deal, almost twice the next one. He is paid the most the rules allow, and it isn’t close to what he’s worth. One Celtic makes the list, and it isn’t who you’d guess: Neemias Queta, $179M over five seasons, on $46M in all.

Count the contracts each player is likely to sign next and the top changes. Nikola Jokić, with two seasons left on his deal and outside this top ten, would be second: $564M in all, most of it after this contract ends.

The worst, now

Forecast over the seasons left on each contract, in today’s money; with re-signing, plus the contracts after it, weighted by the chance he signs them with the same club.

Near the bottom is Jaylen Brown’s, which Philadelphia took on in July. I wanted Boston to keep him, and I still think a team owes its long-serving players something the model can’t price. That isn’t the model talking; it’s me. The model’s number is $60M under water over his three seasons left.

Re-signing barely moves this end of the list: a club is unlikely to keep a player who returns this little. These are forecasts, and they already allow for missed games: the board projects how much each player will play, not just how well.

Why the best are bigger than the worst

Then and now, the shape is the same. The best contract since 1999-00 returned $783M; the worst lost $192M. Of the ones running, the best is expected to return $394M and the worst to lose $78M. A bad contract can cost a team at most what it pays, and none of the ten worst lost more than that. A great one has no ceiling, because the player’s pay is capped and his play isn’t.⁶ That’s the case for swinging big in free agency.

Every running contract is on the Value Board. Next October I’ll check this year’s forecasts against what the season actually did.

Figures: every season on the players board, 1999-00 to 2025-26, with contracts from the contract ledger; running contracts as the Value Board projects them; in today’s money, 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. Fine, the economics: the difference is what economists call surplus. A season’s worth is the player’s Marginal WAR (mWAR) at the price an open market pays for a win; the method has the price. ↩
2. Each season’s surplus as a share of that season’s cap, times the 2025-26 cap of $155M. The cap was $34M in 1999-00. ↩
3. The contract ledger records an extension as part of the deal it extends, which is right for pricing trades and wrong for this. Split at the extension, the new money starts the season after it was signed; a veteran extension signed two years early is credited a year too soon. ↩
4. A player with an option leaves when he’s worth more than it pays, so the club keeps only the seasons he’s worth less; a club with an option declines when he’s worth less, so it keeps only the seasons he’s worth more. Each is priced over the forecast’s uncertainty. Players in their first three seasons are left out, since most are still on the rookie scale. The contracts after this one run in four-season deals to age 40: at the end of each, the chance he re-signs with his club, from what players like him did, and his pay the max if he’s likely to get one or what players like him re-signed for. The Trade Lab and the Value Board price re-signing the same way. ↩
5. Median, over 2,419 contracts. ↩
6. Fine, the economics: a capped downside and an uncapped upside is a convex payoff, which is why a team should take more of these bets than the downside alone suggests. ↩