Nobody is buying the jersey.

The thesis, in one man's closet

The staircase

Jalen Brunson had a season. He also, conveniently for us, had a closet that went to auction in pieces — twelve lots across four different auction houses. Same player. Same body. Much of it the same shirt, from the same team, in the same season. If the object were the thing being sold, these lots would all fetch roughly the same money.

They did not. They formed a staircase.

Nine of the twelve Jalen Brunson lots in our record, across Sotheby's, Goldin, Heritage and Grey Flannel, 2024–2026. The three omitted — All-Star weekend shorts, a warm-up set and a Shooting Stars jersey, $2,560 to $4,096 — sit exactly where the staircase predicts. Bars are proportional on a compressed scale — a true-to-scale chart would render every step below the top one invisible, which is itself the finding. All prices all-in.

Top to bottom, that is a 667× spread on one man's wardrobe. And read it from the bottom up, because the bottom is where it gets interesting. A nameplate with his name on it — a piece of printed card that hung on a locker — is $1,536. A jersey he actually wore, in an actual NBA game, but on an ordinary Tuesday against nobody in particular, is $5,470. That's the floor for the real thing. Everything above that floor is not paying for the jersey. It is paying for which night.

And each step up is a bigger night. Christmas Day, when the league puts you on national television, is worth six times an ordinary game. A trophy — even the NBA Cup, the league's newest and least venerable trophy — is worth seventeen times it. And the Finals debut, the night he became the story of the season, is worth a hundred and eighty-seven times it.

The shirt never changed. The night did. The night is the entire price.

This report is the long version of that sentence, tested against 5,769 lots instead of one closet. It holds. But the interesting part is not that it holds — it's the machinery underneath, which turns out to be simpler and more brutal than the hobby's folklore suggests.

What's inside

Question one

Was it really there?

Here is the uncomfortable fact underneath this entire market: almost nothing can be proven. A jersey is a jersey. Teams make dozens of them. They look identical, they are identical, and the paperwork — a letter from somebody saying this is the one — is a piece of paper that somebody wrote.

The market solves this two entirely different ways, and almost every argument in the hobby is really an argument between them.

Rail one: never let go of it

The strongest proof is not a document at all. It is an object that never left the custody of the people who can vouch for it.

Sotheby's is the official game-worn source of the NBA. A jersey comes off a player's back, into the league's hands, and into a saleroom. It never touches the secondary market, never sits in a collector's closet, never passes through a dealer. There is no chain of custody to reconstruct because the chain was never broken.

You can see exactly what that is worth by reading the biggest lot in this report. Jalen Brunson's Finals-debut jersey sold for $1,024,000. Its entire catalogue description reads:

NIKE, NIKE MESH JERSEY, SIZE 48 (+4 LENGTH). MESH.

That is the whole of it. No photo match, no certificate, no forensic report. The proof is the line printed above the title: Property of the National Basketball Association. Across 72 Sotheby's lots we pulled at random from six 2026 sales — premier and weekly, jerseys and sneakers — 71 carried source-level provenance: 59 consigned by the NBA itself, 12 from the Stephen Curry Collection, consigned by the man who wore them.

And it is not a Sotheby's quirk. NFL Auction runs the same model for American football, and the signature is identical: across its 147 lots in our record, not one cites a photo match, a grading service or a certificate — because the league is the consignor and there is nothing to certify. Two leagues, two houses, the same structure arrived at independently.

Ninety-nine percent. That is not a house with weak documentation. That is a house whose NBA inventory is documented at the strongest level available, structurally, before a single lot is catalogued — because the provenance comes from the partnership, not from the lot. Every NBA Auctions catalogue carries it. Sotheby's non-NBA sales — the Summer Marquee, The Beautiful Game — are ordinary consignments and get read lot by lot, exactly as we read Goldin and Heritage.

Rail two: prove it after the fact

Everything else in this market — the 1994 jersey that surfaced from a collection, the sneakers a ball boy was given, the bat that spent thirty years in a garage — has to have its custody rebuilt. That is what the instruments are for, and it is skilled, expensive work.

The most famous is photo-matching, and it deserves its reputation: a matcher takes high-resolution photographs from the night in question and compares them, pixel by pixel, against the object in hand — the exact pattern of pilling on a shoulder, a pulled thread, a stitch sitting a millimetre off, the precise geometry of a number. Fabric wears randomly, and randomness is a fingerprint. It can pin an object not just to a season but to a night, a quarter, a play.

It is not the only one, and the others are not lesser. A MeiGray or league programme tag catches a jersey on its way out of the locker room. A MEARS grade does forensic work on vintage material that predates usable photography. PSA and JSA certify the object and the signature. A letter of provenance from an athlete or an estate is rail one arriving in an envelope. Read the lot, not the label — the Curry sneakers later in this report carry four instruments at once, only one of which is a photo match.

And the market pays about the same for all of them:

At the houses that sell reconstructed material, that gap is the single largest effect in our record — larger than any individual certificate:

The obvious objection — and we spent more time trying to kill this than anything else here — is that documentation isn't causing the price, it's just tracking it. Valuable lots get documented because it's worth someone's while. Cheap ones don't. You would see the same correlation if the paperwork did nothing at all.

So we held the player constant: forty-six athletes who have both a matched and an unmatched lot, each compared only against himself. The premium survives. It drops honestly — from about 5.3× to about 3.5× — and holds for thirty-six of the forty-six. LeBron's matched lots run about eight times his unmatched. Jordan's run 5.7× across twenty-five lots. Not a law. An association, large, and it does not go away when you attack it.

The first document does most of the work

Because instruments stack, you can ask what the second and third are worth. Across Goldin and Heritage, where the lot text is rich enough to count them:

Median all-in price by number of instruments cited

none$1,586

one$7,015

two$9,760

three$7,500

four or more$24,000

For a seller that is a clean instruction: get one credible document, the best one available for your object, and stop. The second is worth having. The third is for your own comfort.

Where it gets useful

When the two questions collide, magnitude wins

Most of the time proof and magnitude travel together, which makes them hard to tell apart. The useful cases are the ones where they point in opposite directions — where you can buy proof of an ordinary night, or an extraordinary night with no proof at all. The market's preference in those cases is consistent, and it is not what the hobby's conventional wisdom says.

Brunson's own closet says the same thing, one page back. His photo-matched, dated, fully-evidenced regular-season jersey made $8,845. His Christmas Day jersey — no match mentioned — made $35,560. Four times more, for four times less proof, because one of them was on Christmas.

So the order of operations matters. Proof is a multiplier on magnitude, not a substitute for it. Proving that an object was present at a night nobody cares about is an expensive way to certify that nobody cares.

Buy the night. Then buy the proof of the night. In that order — the market will not reverse them for you.

The distortion

The room it sells in

Everything so far has been about the object and its history. This section is about something else entirely: the same object, with the same history, is worth radically different amounts depending on which catalogue it lands in. This is the finding that should make the industry uncomfortable, and it is the best-evidenced distortion in our record.

Start with the cleanest case — the same jersey, sold twice in one year.

One pair proves nothing on its own. So we ran it across the whole house. At Sotheby's, lots placed in premier catalogues ran a median of $4,096; lots in the weekly sales ran $896. That is 4.57×, across 229 premier lots and 336 weekly ones.

And when you rank the individual sales, the spread inside a single auction house over a single year is astonishing:

Notice something in that table: the Finals material appears twice, at $15,360 and at $576. Same event, same fortnight, same house — the premier catalogue and then the third overflow catalogue, twenty-six times apart. The market for the event does not decay because the event got smaller. It decays because attention did.

This has a plain reading for a buyer, and it is the single most actionable sentence in this report: the mispriced lot is the good one in the wrong sale. Nothing else in our data offers a comparable edge for comparable effort. You are not competing against the market's judgment of the object. You are competing against how many people bothered to open the catalogue.

The shape of the money

A handful of lots is the entire market

Every finding above sits on top of a distribution that a newcomer should understand before spending a dollar, because it is not the distribution most people assume.

The typical lot in this market is cheap. The median lot at Goldin is $2,074; at Heritage $3,294; at Sotheby's $3,556. These are the prices of a used bicycle, and they are the middle of the market, not the bottom.

Meanwhile, the top tenth of lots holds most of the money:

Share of a sale's total value held by its top 10% of lots

Sotheby's80%

Heritage69%

Goldin59%

Goldin's Summer Game-Used sale is the worked example, and we captured every lot of it — all 563, no price floor, the complete population. It realized $4.1 million. The top twenty-five lots — four percent of the sale — accounted for 41% of that. Seventeen of those twenty-five were photo-matched. The top lot was a photo-matched, signed Michael Jordan pair from his final season, at $183,000.

What that means practically: this is not one market, it is two wearing the same coat. There is a thin market at the top where moments are priced in multiples and the analysis in this report applies with real force. And there is a wide, shallow market underneath it where a jersey is a souvenir, prices are set by whoever showed up, and no model — ours or anyone's — will help you much. The mistake newcomers make is reading headlines from the first market and buying in the second.

The reach

The same engine runs off the court

A reasonable objection to everything above: this is a sports story. Sports fans are irrational about championships, athletes are famous, and none of it generalises.

It generalises. When we folded in the entertainment sales — Julien's and Heritage Entertainment, 1,790 lots — the same machine was running at the same violence of multiple, with the vocabulary swapped out.

The pattern repeats. Ace Frehley's Budokan stage-played Les Paul made $512,000 — magnitude, the specific famous night. A transom nameplate from the shark boat in Jaws, photo-matched to the film, made $192,000 against a $30–50,000 estimate — which is question one, photo-matching, wearing a movie costume and behaving exactly as it does in a basketball arena.

In sport we call it photo-matching. In entertainment they call it provenance — a documented chain of custody, an estate, a single-owner sale. Different word, identical job: proving the object was there when it mattered. That is why this report's beat is sport and entertainment. It is one market with two vocabularies.

The part most reports leave out

What we got wrong

A record that only contains wins is not a record, it's marketing. Five corrections stand, at the same size as the findings.

We misread our own central signal.

After the Goldin sale we published that the moment premium was weaker than we'd thought, on the evidence of three Stephen Curry lots that came in under our calls — a Finals Game 1 pair, a 51-point night, an MVP season. That was exactly backwards. Those are not moments; they are a good game, a big number, and a long stretch of time. Game 1 is not the Finals. Fifty-one points is a box score. The market didn't shrug at the moment — it correctly declined to pay a moment's price for a stat line, which is our own thesis working, not failing. We had it in writing and we still misread it.

We mistook Sotheby's cataloguing for Sotheby's proof.

This is the worst error in the edition, and it took three attempts to see. We searched lot titles for the phrase “photo-matched,” found it rarely at Sotheby's, and treated its absence as absence of proof — publishing 1.01×, then 3.0×, then 0.85×, none of them stable. All are withdrawn.

Then we read the actual lot pages. Sotheby's is the official game-worn source of the NBA: 71 of 72 lots we sampled are consigned directly by the league or by the athlete who wore them. Brunson's $1,024,000 jersey carries no photo match because it does not need one — the line above its title reads “Property of the National Basketball Association.” We ran a string search against the house with the strongest provenance in the set and called the result evidence. A text search is not an analysis, and the absence of a word is not the absence of a fact.

Timing is not a proxy for the moment.

We wagered that since the market prices moments, it would price timing too — material still warm from an event running hot, a quiet off-season sale running cool. Goldin's Summer sale closed on 8 August, dead in the pre-season, and ran hot. One strike, n = 1, against our own falsifiable line. It stays on the board.

Two older ones stand withdrawn.

A $176,900 Heritage “ceiling” computed off a part year that the full population blows past — there is a $2,623,000 lot in it. And “four of four six-figure vintage jerseys below estimate,” built on truncated titles that concealed what the lots actually were. Both withdrawn, both logged with reasoning. The rules that came out of them — never compute a trend from a part year, never trust a truncated title — are now rules rather than lessons we keep relearning.

In closing

The object is just the receipt

This market does not buy things. It buys evidence that a moment happened and that this particular object was there when it did — and the object is merely the receipt. A Finals jersey, a coach's name on a shoe, a curated room, a movie star's union card: each is a different way of proving the same thing. Where the proof is tight and the night was big, the market pays in multiples. Where either is loose, it shrugs, and two identical objects sell five times apart on the same evening.

Buy the night, then the proof. Mind the room. Never confuse a great performance with a valuable one.

Lot & Hammer is independent and subscriber-supported. For the underlying dataset, a valuation on a specific lot, a correction, or to put an upcoming sale on the tracker — and for press, who are welcome to quote this report — write to ryan@lotandhammer.com.