Wall Street closes at 4pm. Onchain markets don't. Between 24 Sep and 23 Oct, about 60% of stock-token trading on Robinhood happened outside regular US hours, and roughly $532M changed hands on weekends and holidays. This week Hyperliquid added 24/7 Nifty 50 perps through HIP-3, and Kraken's parent is lining up CFTC-regulated perps on the same order book with an optional allowlist. The demand for markets doesn't stop when the closing bell rings. The real work now is pricing the hours when the underlying is shut, and that's where onchain venues get to prove themselves. Building for the 24/7 trader isn't a niche anymore. It's the default.

TBA is a game I'd actually play. Battle Royale... So good!

Bitcoin is chopping around $80K and everyone's staring at the Fed. Meanwhile, the rails are quietly being rebuilt on Hyperliquid. In the past week alone: Ripple Prime started offering institutions 24/7 swaps on 100+ real-world asset markets that reference Hyperliquid perps. Kraken's parent Payward says it will create, clear and settle CFTC-regulated perps for US users on HIP-3 markets. And Uniswap is launching on Elysium, the Hyperliquid L2 going live on 9 November. Regulated brokers, a major exchange and DeFi's biggest DEX are all choosing the same venue to execute on. Price is noise this week. Who builds where is the signal. The traders who learn to use these new markets well will have a real edge, and that's exactly what we're building HyperFlow for.

For four years, doing your homework on altcoins didn't pay. Real Vision's Jamie Coutts backtested the top 200 tokens since 2022. The highest-quality third, with real fees, tight supply and actual payback to holders, fell 50% while Bitcoin rose 82%. This year that same group is up 49% while Bitcoin is down about 6%. One year isn't proof, and Coutts says so. But it matches what's happening on the ground. PaperTrade pulled in $140M of deposits, then made its token non-transferable, and one tracker showed staking revenue fell 93% in a day. Delphi's Jason Pagoulatos has the cleanest test I've heard would the business work without the token? If yes, the token can share in real revenue. If no, the token is the business, and that's a fragile place to build.

Stablecoin freezes are losing the race to hackers. Blockaid looked at 60 crypto exploits from January to August and found that stablecoin issuers' blacklists recovered nothing for victims. Controls built into chains and protocols held 99% of the $91.2M that was frozen or recovered. This week showed why. When Initia's bridge was drained, it halted its own chain, but the stolen funds had already crossed to Ethereum. The stablecoin issuer's freeze landed about 89 minutes after the money was swapped away. Tether's freeze on CoinDCX outflows came about 31 hours later and caught roughly $2.2M of about $16M. The lesson for builders: don't outsource incident response to someone else's compliance desk. Pause switches, rate limits and monitoring need to live in your own protocol, and be tested before the day you need them.

Papertrade just taught every founder designing a token a hard lesson. Its house-pool perps exchange on Hyperliquid pays losing traders in PAPER, a share of the house's take. This week the team said PAPER won't be transferable for the foreseeable future, and outside venues delisted their PAPER perps. The reasoning is the interesting part. If the pool holds $10M and draining it costs an attacker $15M, the attack loses money. Add a $10M PAPER short on another exchange, and the crash pays for the attack. A token that's a claim on your treasury is also a hedge for whoever wants to hit it. So before you launch, ask one question: if someone shorts my token, what's the cheapest way to make that short pay? If the answer is "attack the product," fix that first.

AI agents can already pay. The harder problem is proving they were allowed to. This week OpenAI said an internal model derived keys from public recovery phrases, checked 362 addresses and paid test-network fees from an account it had no right to use The same week, fixes started shipping. World launched human checks for Zoom and for agents. Mysten Labs and Google Cloud announced a verifiable record of what an agent did and on whose authority. Circle and Proof demoed an agent presenting a verified identity before paying in USDC. Mastercard opened early access to Agent Pay for Machines m. For founders building with agents payment rails are the easy part now. The product people will pay for is the audit trail. Who the agent acts for, what it may do, and proof it stayed inside those lines.

Design wise I’m a big fan of this. Very murcielago. Would like one.

A year after the 10/10 liquidation crash, the most telling data point isn't price. Galaxy Research notes that only eight of the top 50 tokens (ex-stablecoins, wrappers and LSTs) trade above their 10/10 levels, and most share two traits: a durable underlying business, and a mechanism that ties that business's value to the token. Meanwhile Blast and Blur are winding down, and more projects are announcing shutdowns. The market is getting better at telling a token from a business. For founders, that changes the brief: token design isn't a launch-week marketing exercise. It's the contract between your revenue and your holders. If you can't explain in one sentence how your product's success reaches your token, assume the market already has an answer, and it isn't the one you want.

When trading perps, bridging, or just swapping spot, there is no reason not to use @HyperFlow_fun Its fast Cheap and you earn FlowXP Every trade counts toward collective SIZE making every trade cheaper. Simple