GameStop stopped being a video game store. It's a card shop with 30% four-wall margins, sitting on $13.70 a share of cash and securities, priced at 4.7x EBITDA. Wall Street won't look because the ticker is embarrassing. We looked. It took 34 slides.
Filings rebuilt from scratch, a dozen store visits, and eBay sales data pulled by hand. Click the right side of a slide to go forward, use your arrow keys, or swipe. Every slide has its own link, so you can send slide 23 to the friend who still thinks this is 2021.
The full written thesis, with every number sourced and every channel check, goes out to The Value Road subscribers. It's free.
Get the written thesisTap the middle of a slide to zoom in and drag around. Prefer words? Read the whole deck as text.
Six things Wall Street is missing, straight from slide 6. Wrinkle-brained readers can check every number in the deck above.
Cards, not consoles. Collectibles went from 6% of sales to 45%, gross margin went from 29% to 44% in twelve months, and gross profit per store doubled to ~$640K after eight flat years. Same box, same lease, different business.
GameStop buys ~$520–560M a year of games and graded cards over the counter, two-thirds paid in store credit that comes back at full margin. eBay can't source a single card. Whatnot can't buy a Charizard from a teenager. 1,600 counters can.
Management says "in excess of $650M" of EBITDA. The first half alone was $340M, and the holiday quarter hasn't happened yet. Repeat Q2 twice with zero seasonality and you're at $688M. Our bridge is ~$885M.
$4.7B of cash, $4.7B of eBay stock, 4,710 Bitcoin, and $2.8B of notes that pay 0% and don't mature until 2030. Even a fire-sale valuation with the retailer at zero lands around $11.50 a share.
No salary, no options, no bonus. $866M of his own money in the stock at a $3.78 average cost, roughly 20% of his net worth, and another $20.4M bought on September 10. He has never sold a share. Diamond hands, but make it a Form 4.
PSA grading at 1,600 counters. A 9.8% eBay stake that is already a look-through tenth of TCGplayer, Goldin, the vault and eBay Live. A partnership worth ~$150–365M of EBITDA without a merger, and a shot at live commerce. At 4.7x EBITDA you're paying for none of it. Each one is a contract, not a $56B check.
The obvious pushback: "Cool, so it's a Pokémon stock, and Pokémon is a fad." We went back 400 years and twenty collectible manias to answer that. The product always crashes. The booth it passes through keeps collecting.
| The craze | Boom-era product | Who kept the money |
|---|---|---|
| Junk-wax baseball cards, 1991 | −90% | Beckett, PSA |
| Comics, 1993 | −90% | Diamond, CGC |
| Beanie Babies, 1998 | −95% | Ty Warner, eBay |
| Pokémon, first wave, 2000 | −85% | Nintendo / TPC |
| NFTs, 2021 | −97% | OpenSea, early minters |
| Funko, 2022 | −80% | Licensors, pre-IPO owners |
| Pokémon, pandemic wave, 2021 | −35% | TPC, PSA, eBay / TCGplayer |
Seven of the twenty cycles on slide 21. Across all twenty, graders and marketplaces kept the gains in 14, the manufacturer in 6, the pure specialty retailer in almost none. Our synthesis of trade, press and index data.
GameStop used to be the specialty retailer in that table, the one that never keeps the money. It's now the other thing: 1,600 counters that buy from the public, a grading intake that is roughly 10% of PSA's volume by our estimate, and store credit that gets spent on the shelf behind the register. It takes a spread on whatever crosses the counter. BookOff, Hard Off, Mandarake and CeX run this model and have lived through every cycle since 1990.
Trade in a card, take store credit, and GameStop gets the inventory today and pays for it later, in merchandise, at retail. That credit sits on the balance sheet as ~$110–150M of interest-free funding, about $65K a store. The company issues ~$350M of it a year (our estimate), some never gets redeemed, and vendors finance another $215M of what's on the shelf. Walmart has vendor float. Costco has membership float. Nobody else has customers walking in to hand over inventory for an IOU. Slide 17.
In FY16–19 GameStop's collectibles aisle was the Funko era, at 33–36% gross margins. Funko peaked in 2022 and boom-era product fell about 80%. GameStop's collectibles business didn't die with it. It went from $718M in FY24 to $1.06B in FY25 and a ~$1.4B run-rate, on cards, at an implied ~56% margin. Different product, same counter, better economics.
One Piece was sold out in all twelve stores we visited and has been the #3 game on TCGplayer for three straight quarters. Magic did $1.7B last year, up 59%. Fanatics is pulling the MLB, NFL and NBA card licenses under one roof. The counter doesn't care which one wins. Pokémon is the current tenant. GameStop is the landlord.
Hyperinflation is the extreme case, and it shows the mechanism clearly. In Weimar Germany people fled the mark into stamps, art and pianos, anything real. In 1970s America it was coins and silver. Savers in places like Argentina and Turkey have spent years moving into dollars, gold, cars and goods, whatever holds value better than a bank balance. In 2020 the stimulus checks landed and Americans bought Charizards.
You don't need a currency collapse for this to matter. In our data, monthly eBay Pokémon volume tracks the M2 money supply at +0.92, gold led by three months at +0.91, and the fed funds rate at −0.78. Sixteen of the eighteen busts we could date followed central-bank tightening within a year. Collectibles behave like a liquidity asset.
Our view is that the liquidity is coming: 7%-of-GDP deficits, more than $1 trillion a year of interest, CPI re-accelerating to 3.4%, central banks buying a record 289 tonnes of gold in a quarter, and a dollar that has lost 87% of its purchasing power since 1971. When the next round gets printed, some of it flows into things that can't be. The Pokémon Company will never reprint a 1999 1st Edition Base Set, and GameStop is the physical exchange for that kind of asset.
The honest version: twenty months of correlation is a pattern, not proof. The Fed hiked on September 16, and a toll booth still collects less when traffic drops. Our own model has EBITDA falling from ~$885M to ~$646M in the year after the anniversary. The claim is that the counter survives and rotates, not that it's immune. All eight risks are here.
Think $885M of EBITDA is fantasy? Think the card shop deserves a liquidation multiple? Fine. Drag the sliders and see what's left. The balance sheet is from slide 7; everything else is yours to break.
Scenario arithmetic on thesis-date balance sheet figures ($4.70B cash, $2.80B of 0% notes, 504.3M shares). It is not a price target, a forecast or advice.
Anyone can post rockets. These are the specific things the thesis says will happen, with dates, from slide 31. When each one lands it gets marked hit or miss, and the misses stay up.
| When | What we said | Grade |
|---|
Subscribers get our read on each of these as it lands, including the ones we get wrong.
Get the catalyst updatesA thesis without the risks is a pitch. These are slides 32 and 33, in full: eight ways we lose money on this, using our own numbers against us, and the signals that would tell us it's happening.
After the 25th anniversary in 2021, modern Pokémon prices fell 30–40% and The Pokémon Company printed 9 billion cards into the correction. Japan leads the U.S. by six to twelve months and already shows the shape: high-end singles down 75%, shops in Akihabara closing, sealed premiums compressing from 6x to 2.4x. Our FY27 builds in the hangover: EBITDA ~$646M against ~$885M. A correction is priced. A bust is not: collectibles −40% at a 5x multiple is worth ~$15–16 a share.
GameStop is one of five direct accounts. TPCi bought its printer in 2022 and its largest U.S. distributor in February 2026, explicitly to control allocation and retail pricing. Its goal is a $49.99 box reaching a customer at $49.99. GameStop listed that box at $150 on launch day, in the press, under a meme-stock brand. Roughly 25% of revenue is Pokémon-linked (our estimate), and the sealed half of that is whatever the licensor decides to ship.
A card bust costs 25–30% of the stock. A bad deal costs 40–50%, and the deal is the one thing management is actively trying to do. The eBay bid as structured needs ~$22B of new debt at 6–7x leverage, hands eBay's holders ~70% of the company, and spends the $13.70 floor: ~$15 a share on our math. The record: NFTs (closed), Bitcoin bought at $106K (−40%), now a hostile bid. Cohen is elite at raising and cutting capital and unproven at deploying it in size.
The grading program is PSA-only. PSA paused every tier under $80 in June, sits on a 12.4M-card backlog, and is losing share (72% to ~65%) to CGC, which grades the same card for $15–20. Its parent faces an antitrust suit, and its CEO sits on GameStop's board, which makes adding CGC at the counter awkward. Every collector who left over the price hike is a slab GameStop's counter can't buy.
TCGplayer, the hobby's price feed, the point-of-sale software in the card shops, Goldin, the vault, eBay Live: all eBay's. Whatnot owns live commerce at ~$15B of GMV. Every hit pulled on a stream is sold for a platform fee instead of traded in at a counter. If the eBay episode ends in acrimony instead of a contract, GameStop is a sealed supplier at ~35% margins on someone else's toll road, not the exchange where the 55–60% resale happens.
16 of 18 datable busts came within a year of central-bank tightening, and the Fed hiked on September 16 for the first time in three years. Gold is 18% off its January high, and that +0.91 correlation works in both directions. If the sequence is squeeze first and capitulation later, the squeeze lands on FY27's numbers and the capitulation on somebody else's holding period.
GameStop doesn't disclose category margins, Pokémon's share of sales, PSA volumes, per-store costs, or the sealed-versus-graded mix. Every one of those is our estimate. The balance sheet, the guide arithmetic, the segment margins and the insider buying rest on filings; the grading network, the anniversary sizing and the eBay synergies rest on inference. If GameStop's PSA share is 5% rather than 10%, the intake story is half as large.
Retail holder base, no earnings calls, no comps, a single-number guide, and a CEO who communicates through social media and hostile bids. The arbitrageurs who own $2.8B of the converts are short the stock against them. If a squeeze took the stock through $32 before October 30, 59M shares of warrant dilution would arrive at the top. The balance sheet protects the value. It does not protect the price.
We're long, which means we're biased, which means you should check the work. That's what the 34 slides and the calculator are for. If you find the hole, tell us.
Deep value means being early, and being early feels exactly like being wrong. Hold the button for 12 seconds while the tape does what the tape does. Let go and you're paper.
X won't attach images from a link, so save it and drop it into your post.
The full GameStop thesis is going out on The Value Road: every number sourced, the store visits, the 400-year collectibles study, and the model. Free to read.
Get the full written GME thesis from The Value Road. Free.
Subscribe