Most tokens work fine right up until someone asks what they actually entitle you to. Usually that someone is a lender, an exchange or a regulator, and by then the answer is already fixed. That question, and the structure underneath it, is what I work on. Cross-border structuring since 2009, crypto since 2016. FOR FOUNDERS AND PROTOCOL TEAMS – Token design and tokenomics – What the token legally is, and where it can be issued – Foundation and DAO structuring, opco / foundation / IP splits – MiCA, CASP and VASP perimeter and licensing routes – Tokenised assets: wrapper, custody, transfer restrictions – SAFTs, token warrants, purchase agreements – Token compensation for contributors and core team – Exchange and market maker agreements – Standing counsel where there is no in-house legal FOR FUNDS AND INVESTORS – Token deal diligence: what you own, and what can be changed without you – Term sheet review, and whether the equity and the token are actually linked – Structural risk on a target before you wire – Portfolio support, restructuring, regulatory exposure – Portfolio companies that launched on structures that don't hold – Second opinion where counsel knows the law but not the token frederik@sestina.io / Telegram @fallund_2

Polymarket: $5.37M volume in 24h. Kalshi: $209.5M volume in 24h Open interest? Nearly identical. $10.3M vs $10.5M Same open interest. Same positions.Same market. 39x difference in volume. The allegation is wash trading, meaning that Kalshi's volume is "created". Is it true? I do not know. Polymarket's volume is on-chain. Kalshi's is not. One you can check; the other you can't.

$LAPTOP hit a $144 billion valuation yesterday. The pool setting that price held $48,000. Three million to one. LAPTOP launched on Base yesterday and peaked two minutes later at $401.12. Thirty minutes after that, it was down 95%. It trades at $0.77 today. If you invested $10,000 at the top, you would now have $19 left. The obvious take is that this is peak wealth destruction. Never has so much value been lost in so little time. There is a point here, but in reality, this wealth was only a calculation. Here is what that $144 billion actually was. It was the market capitalisation, not the FDV. The FDV was $401 billion. A market capitalisation is circulating supply multiplied by the last price someone paid. Nothing more. Three hundred and fifty million tokens, one price, one multiplication. And that price was set inside a liquidity pool holding forty-eight thousand dollars. Every fully diluted valuation in every pitch deck is the same calculation. Every tokenised-asset market cap is the same calculation. The number tells you what the last buyer paid. It tells you nothing about what the next seller can get. A valuation without a liquidity denominator isn't a measurement. It's a multiplication. When a lot of people want to buy the token in the first couple of minutes after launch with a liquidity pool this size, the price will go parabolic. But it cuts both ways. When selling pressure mounts, the price will have a similar trajectory in the other direction. Which is exactly what happened. LAPTOP is a meme coin. It does not do anything. There is no substance behind it. A meme coin's price is a function of perception. Its valuation is a function of how little money it takes to move it.

If you invested $10,000 in $LAPTOP a couple of hours ago (as it peaked), you would have just $30 left. A staggering 99,7% loss. Except it would not have been possible, as the official pool held ~$83k USDC and the Uniswap pool ~$380k. Peak wealth destruction.

A seven-year-old blockchain is proposing to switch itself off. The decision took four weeks. It began with a mint nobody authorised. On 12 August, attackers created close to 4 billion ONE that should never have existed, roughly 26% of the entire supply. On 7 September, Harmony proposed retiring its layer one and moving ONE to Ethereum as an ERC-20. Twenty-six days. It is still a proposal, not a decision. But $1.372 million has been reserved to compensate validators who shut down by 10 September, and the snapshot is set for the same day. Money and deadlines are already moving. If you have ONE in a wallet, you're fine. Conversion is automatic. If your ONE is in a liquidity pool, a multisig, or any smart contract, it cannot transfer under this plan. When a system unwinds, the simplest claim is the one that survives. Anything held through something else depends on that something else still being able to act, and in a wind-down, it usually can't. The people stranded are the ones who did the more careful thing. Provided liquidity. Put treasury funds behind a multisig instead of a single key. Locked tokens in a protocol. The plainest holder is the safest. Every layer of structure above direct ownership is a layer the exit doesn't reach. A seven-year-old chain. Four weeks from a supply failure to proposing its own end. The mint didn't just create tokens that shouldn't exist. It ended the thing that issued them. Data: Harmony community proposal, 7 September 2026

The technical question was settled years ago. You can issue a token representing almost anything. The legal question is what the holder is actually entitled to when the underlying stops performing. That one gets asked later, by someone with a reason to ask it.

Robinhood made $727 million from trading last quarter. $129 million of it came from stocks. The full split: ✅ Options: $342m ✅ Event contracts: $156m ✅ Equities: $129m ✅ Crypto: $100m Robinhood is a stock trading app. Stock trading is its third-largest source of trading revenue. Options have quietly been the real business for years, more than double any other line, and 2.65 times equities. That part isn't new. What's new is the line that just moved above stocks. Event contracts (prediction markets) went from $10 million a year ago to $156 million last quarter. Fifteenfold. Robinhood calls it the fastest-growing product line in its history, and it now accounts for roughly 12% of total net revenue. Now the part that is not said out loud. That product is currently the subject of lawsuits in six states, brought under statutes written to let people recover gambling losses. Robinhood's defence is that federal commodities law preempts state gaming law. The Third Circuit accepted that argument in April. Washington barred sports contracts anyway. If the preemption argument holds, Robinhood has a product that is already 12% of net revenue and still compounding. If it doesn't, that 12% has a jurisdictional map problem, and the fastest-growing line in the company's history might become its most expensive one. Nobody knows which yet.

Seems like the hackers have refunded 3,600 BTC. The rest might be a "negotiated" finder's fee. It will be interesting to follow in the days to come.

Someone printed $320 million of Bitcoin that didn't exist. Nobody could tell it was fake. The system was built not to be able to. On Saturday, the wallet backing Liquid's Bitcoin token went from roughly 4,200 BTC to 197. A flaw in the software allowed previously verified cryptographic proofs to be reused, minting L-BTC that was backed by nothing. The attacker took 4,000 of those tokens to SideSwap and requested a peg-out, the ordinary process for converting the token back into real Bitcoin. SideSwap processed it. Burned the tokens and instructed the federation to release around 3,996 BTC. SideSwap wasn't breached. Its key wasn't compromised. In its own words, it could not distinguish those coins from ordinary L-BTC. Nobody could. Liquid uses confidential transactions, which hide amounts. That is the network's flagship feature, and it is the reason the counterfeit was undetectable. You cannot audit a supply you have deliberately made unobservable. No key compromised. No system breached. Contracts operated exactly as written. Structures don't fail when someone breaks the rules. They fail when someone follows them.

#Pons did $9.05 million in fees in 24 hours, beating #PumpFun. Pons is eight weeks old. First, the numbers for the last 24 hours: Pons - $9.05 million PumpFun - $ 3.55 million Pons is a memecoin launchpad that went live alongside #Robinhood Chain in July and has now out-earned PumpFun over 24 hours. Pump has led that category since early 2024. Pons is now ranking second among all crypto protocols by fees, only behind Tether. Around 27,000 tokens launched through it in those 24 hours. Roughly 530,000 since July. Two things worth taking from it. ✅ First, where the money is Not with the chain, and not with anything beneath it. An application sitting on top out-earned all of it. This is a clear example of value accruing at the application layer rather than with the infrastructure. ✅ Second, the clock. PumpFun spent two years becoming the default. Pons took eight weeks to out-earn it on a given day. So yes, value accrues at the application layer. That has been true for a while, and this is the clearest evidence of it I've seen. But notice what else it means. If a category leader can be displaced in eight weeks by something that didn't exist in June, then application-layer value is enormous and almost entirely undefended. Infrastructure captures little. Applications capture a lot, briefly. Which leaves us with a question nobody has a good answer for. Not where the value goes. Where it stays.

Europe wrote the world's first stablecoin rulebook. Two years later, euro stablecoins are 0.25% of the market. To be fair, last month was a record. Euro stablecoins reached an all-time high of $776 million, up 6% for the month and 68% year over year. Circle's EURC passed €400 million for the first time. That is real growth, and it is accelerating. The total stablecoin market is $311 billion. USDT and USDC alone account for around $257 billion of it, or roughly 330 times the entire euro stablecoin market, combined. The assumption behind MiCA optimism was that the constraint on euro stablecoins was legal. Nobody would build one while the rules were unclear, so write the rules and a market appears. The rules have been in place since June 2024. The market did not appear. Because the constraint was never legal. A stablecoin is a claim denominated in a currency, and demand for the claim follows demand for the currency. Look at what people actually use stablecoins for: 👉 getting dollars where dollars are hard to get 👉 settling in dollars across borders without a correspondent bank 👉 posting dollars as collateral in crypto markets 👉 trading and market making in crypto markets None of those is European problems. Europeans already have frictionless euros. SEPA works. Instant transfers work. A euro stablecoin solves, for a European, a problem that was already solved. This is worth a thought, because it generalises well beyond stablecoins. Regulation is very good at deciding what may exist, but it has no mechanism whatsoever for deciding what people want. Europe has built, by some distance, the best-regulated market in the world for a product with limited demand. And then there is the detail from last month that says it more neatly than I can. Revolut, Europe's flagship fintech, launched its euro stablecoin in August. It is issued by Bridge, which Stripe acquired last year. Europe's regulatory achievement, denominated in euros, running on American infrastructure. None of this is an argument against MiCA. Clear rules are better than unclear ones, and the firms operating under it are better off for the certainty. It is an argument against a specific hope: that writing the rulebook first would make Europe the centre of this market. The rulebook did exactly what a rulebook can do. It made euro stablecoins legal. It was never going to make them necessary.

When you, apparently, do not read your own book…

1,685 people were told the money was in their own wallet. It was. But someone else had access to it. On Friday, roughly $500,000 was drained from card balances at Avici, a Solana-based crypto neobank. Nobody's key was stolen. The attacker submitted a crafted signature bundle, made itself an administrator of the escrow accounts holding user funds, and withdrew them. Three steps, all permitted by the contract. The users thought they had self-custody; that was only partially true. The flawed contract wasn't Avici's. It came from Rain, the card-issuing partner. 1,685 people chose Avici. Not one of them chose Rain. Most had never heard the name until their balance hit zero. Holding three different crypto cards feels like diversification. If all three run on the same infrastructure, it isn't. Avici has pledged to refund every balance in full, and I have no reason to doubt they will. But look at what compensates those users 👉 Not insurance. 👉 Not a deposit guarantee. 👉 Not a legal claim against anyone. A company deciding, voluntarily, to absorb the loss, and saying so in a social media post. It worked because Avici could afford half a million dollars and chose to spend it. Had either been untrue, 1,685 people had nothing to fall back on. Because there was nothing there to fall back on. The money was in their own wallet. The safety net was a tweet.

245 crypto platforms were hacked. 147 of them had been audited. Only 11% of the money was lost through anything the audit was looking at. That's from @CoinGecko's security report covering January 2025 through July 2026. Across those 245 incidents, $3.63 billion was stolen. Six in ten of the platforms that lost the money had passed a security audit first. And here is the number that explains it: of that $3.63 billion, roughly $396 million left through a vulnerability that was actually inside the scope of a completed audit. About one dollar in nine. The other eight went somewhere nobody was checking. CoinGecko's own breakdown: 👉 Infrastructure and supply-chain failures: over $1.8 billion 👉 Smart contract exploits at applications: $546 million 👉 Centralised exchanges: mostly compromised private keys One example: The second largest hack was Kelp, at $292 million. Reviewers went back through the code afterwards and found it correct. Every contract did exactly what it was written to do. The loss came from how the bridge had been configured on deployment day, and from whose servers its verifier happened to be reading. The audits weren't wrong. They were just not answering the relevant question. An audit asks one thing: does this code do what it claims to do? That is a real question, and the answer has real value. Here is what it does not ask: ✅ Who holds the keys, and what happens if one laptop is compromised ✅ How this was configured on the day it was deployed ✅ Whose servers the price feed and the verifier are reading from ✅ What a third-party vendor's next software update is permitted to do ✅ Who can mint, and who can change who can mint These are not technical questions. They are structural ones. And they often get decided once, early, by engineers who aren't framing them as risk decisions at all. Then they are never revisited, because no one owns them. The audit has a defined scope. The insurer has a policy with exclusions. The board has a report it doesn't read closely. It is the gap between them that lost the $3.2 billion. Afterwards, everyone points at the audit. But the audit was never built to carry that weight. The question to ask isn't whether you've been audited. It's what your audit was scoped to look at, and who, by name, owns everything outside that boundary.

Tokenised stocks just became the fourth-largest asset on-chain. 2.25 million people hold them. Almost none of them own a share. The growth is real. In a single month, holders rose 179%, trading volume rose 364%, and the category passed $2.5 billion. The average position is $1,138. This is ordinary people buying what they believe are shares. Here is what they actually get. Buy a tokenised Tesla, and you do not own Tesla stock. You own a token issued by a platform. That platform holds, or says it holds, shares through a broker or a custodian. Tesla's share register has never heard of you. Which means: 👉 You have no vote 👉 You have no direct legal claim on the company 👉 Dividends reach you only if the platform's terms say they do 👉 If the platform fails, you do not own shares to reclaim. You are a creditor in someone else's insolvency, standing in a queue You didn't buy a claim on a company. You bought a claim on a company that has a claim on a company. Now look at the price. The token trades at the price of Tesla. Not at the price of Tesla minus the risk of the firm standing in the middle. Everyone can see the first number. Almost nobody is calculating the second. This is not a blockchain problem. It is the oldest structure in finance wearing new clothes. Every time a market has inserted an intermediary between an owner and an asset and then treated the resulting paper as equivalent to the asset, it has worked beautifully. Right up to the moment the intermediary is the problem, and everyone finds out what they were holding all along. It doesn't have to be built this way. A handful of companies have put their real shares on-chain, where the token genuinely is the share and your name is on the register. That version works. It's a small fraction of the market, because it requires the company itself to agree, and almost none have. 2.25 million people think they bought a share. What they bought is a promise from a company most of them can't name. Those are the same thing right up until they aren't.

#Strategy sold $2 billion of its own stock last week. It bought zero Bitcoin with it. Its shareholders now own less Bitcoin than they did in January. Between 17 and 23 August, Strategy issued 18,261,118 new shares $MSTR for roughly $2.01 bn net. Where it went: 👉 $300 m into the USD Reserve, now $5.10 bn 👉 $1.59 bn into a newly created "USD Cash" pool 👉 $136.4 m to buy back its own $STRC preferred shares 👉 $0 into Bitcoin The announcement focused on three numbers: 👉 4% of total bitcoin supply 👉 0% net leverage 👉 840,447 bitcoin:native at an average cost of $75,385. All true. All of them measure the company. None of them measures how this works out for the shareholders. The number that measures this is Bitcoin per share, and Strategy publishes it: ✅ 220,900 satoshis — 25 May 2026 ✅ 188,628 satoshis — 23 August 2026 Down roughly 15% in three months, and now slightly below where 2026 opened. On Strategy's own scoreboard, the year for the shareholder, measured in BTC, is negative. Meanwhile, the company has sold 6,916 BTC in 2026. The proceeds went to buying back preferred stock, not into the treasury. The flywheel only ever ran in one direction. Issue equity at two or three times net asset value, buy Bitcoin, Bitcoin per share rises, premium justified, repeat. That premium is gone. Enterprise mNAV is now reported around 1.0. But the machine cannot stop. The company now has roughly $6.75 bn of debt and $14.94 bn of preferred, with the STRC coupon lifted to 12%. Those payments are contractual. So the issuance continues. It just no longer buys anything per share. And "0% net leverage" is only true if you count $14.94 bn of preferred as equity. Preferred paying a 12% cash coupon and ranking ahead in liquidation is debt from a common shareholder's perspective. Selling $2 billion of stock and buying no Bitcoin is not a treasury Strategy. It is debt service.

Bitcoin just had the biggest short squeeze in its history. Everyone's calling it strength. It's the opposite. It's a measure of how much fragile leverage had quietly piled up. Here's what actually happened. For six weeks, Bitcoin did nothing. It sat in a tight range around $62,000–67,000, with volatility at a record low. And this is where the story starts, because quiet markets don't clear risk; they accumulate it. Through those flat weeks, leveraged short positions stacked up, and a dense band of forced-liquidation levels built between $65,000 and $67,000. By the time it broke, more accounts were positioned short than long. The calm wasn't calm. It was fragility building in plain sight. Then came the spark, and notice it wasn't a crypto event. The US Treasury announced it would double its long-dated bond buybacks, pulling the 30-year yield back from a near-two-decade high and weakening the dollar. Gold and equities rose the same day. This was a macro liquidity signal, and Bitcoin was the most leveraged asset at the time. Once price nudged into that $67,000 liquidation band, the structure did the rest. Closing a short means buying. So each liquidation forced a purchase, pushing the price higher, triggering the next cluster of liquidations, and forcing more buying. More than $1 billion in Bitcoin shorts were closed in a single hour. Over 24 hours, roughly $3 billion in crypto shorts were wiped out against about $260 million in longs. More than ten to one. The largest short-liquidation event on record. And this is the part the "Bitcoin is strong again" takes miss. Almost none of that buying was demand. It was margin calls. Traders weren't buying Bitcoin because they wanted it. They were forced to buy it back to close losing bets. So, does this mean that the Bitcoin price will not continue upwards? As always, the future is difficult to predict. But this is where we are now: 👉 Open interest hasn't rebuilt. 👉 Spot demand hasn't confirmed the move. 👉 More than 44,000 BTC were sent to exchanges as longer-term holders took profit into the spike. Someone put it this way: the rally is now hunting for real buyers. So here's the structural read, against the headline one. The size of the squeeze isn't a measure of how strong the market is. It's a measure of how fragile it had quietly become. How much complacent leverage the boring weeks let accumulate. The same conditions that produce cascade crashes produced this melt-up. Positioning, not conviction, set the size of the move. A rally built on forced buying isn't demand. It's borrowed. And borrowed moves have to be repaid.

Price leads narrative. Always. Now that $BTC has pumped, narratives change. Not the other way around.