Is Anybody Actually Good at This?

In Boston we called him Trader Danny, and about half the time we meant it as a compliment. Danny Ainge would trade anyone, for anything, at any hour, and for most of two decades I was sure it made the Celtics smarter than everybody else. I am from Boston. I would like that to be true. This column is me finding out whether it is.

The question is bigger than Ainge, so the rules come first. I’m not asking who made the best trade or drafted the best player. Somebody always did. I’m asking whether anybody is consistently better: whether knowing who made a decision tells you anything about how it grades, once you account for the year it was made in, and whether that difference is big enough, and clear enough, that it isn’t just the dice.¹

There are two ways to be good. One is process: making the decision the numbers liked on the day. For a trade, that’s the Trade Lab’s grade. For a pick, it’s what the draft board thought the player was worth to the club that took him, against a typical pick in that slot. The other is beating the process: what actually happened, minus what decisions graded like yours usually produce. That second one is where scouting, development and luck all live, and it’s where a front office would show up if it knew things the model doesn’t.

And there are two ways to sort the decisions: by franchise, and by the person running it on the day.² A franchise over 27 seasons is several front offices, and a skill should walk out the door with the person who has it.

Trades: nobody

2,763 trade sides since 1999-00. One side’s grade swings by about $98M either way, in today’s money. The true spread between franchises, once the dice are out, is $3.8M a trade. Random dealing spreads the teams this far 28% of the time.³

By executive it’s cleaner, and by cleaner I mean emptier: 73 executives with at least 15 trade sides, and no spread left at all once the dice are out. Knowing who ran the team tells you nothing about how its trades graded.

Now the part I was dreading. Boston is second of 30, at $21M a side above the league, and on its own that clears the usual bar for “not luck”. So does Miami. With 30 teams you expect one or two to clear that bar by luck, and the test that looks at all 30 at once is the one that knows it. So I can’t tell Trader Danny from a lucky coin. (Sigh.) I would like the record to show that the coin came up heads a lot.

Each franchise’s average trade grade against the league’s, per trade side, in today’s money, with the range that covers 95% of what the dice alone would do. Top five and bottom five.

What about beating the grade, the teams whose trades turned out better than trades graded like theirs? Nothing. The true spread is zero by franchise and zero by executive, over trades through 2020-21.⁴ This is the last trades column one level up: the gap between a trade’s grade and how it turned out belongs to the dice, and so, it turns out, does every team’s.

The draft: close

Here it gets more interesting. By franchise, the board’s value of each pick against a typical pick in that slot spreads out further than random dealing manages 94% of the time. That misses the usual bar, which is 95%, but it’s closer than anything in trades. The first-round version, how close each pick was to the board’s favourite of the players still available, lands in the same place, and it’s the one thing in this column that repeats itself: a franchise’s average in odd-year drafts predicts its average in even-year ones, at a correlation of 0.55.⁵

Miami and the Lakers take the board’s favourite of who’s left most often; Washington and Utah least. The caveat is a big one: “good process” here means agreeing with my model. A franchise that drafts like the board might be good, or might just like the same kind of player it does. And by executive the signal disappears, which is the opposite of how a skill should behave.

Beating the board: three players

Then the one test in this column that clears the bar. By executive, how much a club’s picks returned beyond what the board projected is significant: under 1% by the usual test, 2% by the reshuffle, with a true spread of $23M a pick. The top three are Tim Connelly, Neil Olshey and Sam Presti. That’s a real result. Here is what it’s made of:

The ten picks since 2007 that landed furthest from what the board expected them to return the club that drafted them, all ten above it: classes through 2021, over the seasons each has played, in today’s money.

Take out Nikola Jokić, the 41st pick in 2014, and the result still clears. Take out Nikola Jokić, Damian Lillard and Russell Westbrook and it doesn’t (11%). Take out all ten and it’s nowhere (51%).⁶

So the one front-office skill in 15 drafts is having taken a superstar the board didn’t see coming. That may well be a skill. The trouble is that most of these executives did it once, and one draw can’t tell you whether somebody is good at finding stars or was standing in the right place.⁷

The exception is Sam Presti, with three of the ten: Kevin Durant, Russell Westbrook and James Harden, in three straight drafts. Doing that once is a great story. Doing it three times starts to look like a person.

Also on the list, sixth: Jayson Tatum, the pick I was furious about in 2017. The board had him third, where he went. Nobody had him that good, least of all me.

My call

I’m with the dice. Nobody in this league is consistently better at trades than anybody else, by process or by result, and that includes the man we called Trader Danny. The draft is closer, and the one real difference is a handful of stars nobody saw coming, mostly one to a customer. The exception gets a ledger entry: I think Presti is the one real thing in this data, and I’ll check his next five drafts against the board to see whether he keeps beating it or I’ve been flattering a coin.

Next time somebody calls a GM a genius, ask how many Jokićs that is. Then go grade a few trades yourself on the trades page. It’s more fun than being right.

Figures: every trade side on the trades page, 1999-00 to 2025-26, and every pick on the draft board, 2007 on; executives from Basketball-Reference’s team pages; 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. Formally: a fixed effect for each franchise, and separately for each executive, net of season fixed effects (and pick-slot fixed effects in the draft), tested jointly with an F test and against 2,000 reshuffles of who made which decision within each season. Executives with fewer than 15 decisions are pooled together. ↩
2. The executive is whoever Basketball-Reference names on each team’s season page, the president of basketball operations or the general manager. In a season that names two, trades before March go to the one who was there the season before, and the draft goes to the one who is there the season after. ↩
3. The F test gives 0.27. The reshuffle and the F test put every test in this column on the same side of the bar. ↩
4. Later trades still have contract years to run, so their results are partly projections. ↩
5. Against reshuffled teams, a correlation that high turns up under 1% of the time. The true spread between franchises on that measure is $15M a first-round pick. ↩
6. It is not fragile in the other ways: it clears at every cutoff for how many picks an executive needs, and leaving out any single draft class never pushes it past 6%. It is fragile in exactly one way, which is the way that matters. ↩
7. The economics, briefly: telling skill from luck means estimating how much a person’s results vary around their own average, and you can’t estimate a spread from one observation. Ten more drafts each would help. Nobody gets ten more drafts. ↩