CFA TardFi analyst. Seek first the kingdom of God and his righteousness, and all these things will be given to you.
Joined December 2017
- Tweets3.2K
- Following6.2K
- Followers4K
- Likes117K
Pinned Tweet
Macro predictions for 2022:23
1. The Japanese Yen (JPY) is a black swan event that ignites contagion across the globe. The BOJ’s YCC efforts to buoy the infamous USD/JPY carry trade and keep inflation at bay won’t work. USD/JPY at 136/USD at levels not seen since 2002
Treasury just doubled its long-end buyback operations, from $2 billion per operation to at least $4 billion, covering the 10-to-20 and 20-to-30 year sectors, effective September 9 through November 4.
The 30Y sits at 5.181% today, the 10Y at 4.647%. Both printed twenty-year highs the week before Treasury decided the long end needed a hand.
Consensus is calling this QE. It isn't. Treasury creates no reserves; it buys back off-the-run duration and funds it with bills. That's a maturity swap. What it shares with QE is the portfolio-balance effect: duration leaves private hands, short paper replaces it. So the crowd has the label wrong and the direction right, which is the most expensive way to be correct.
The part nobody says out loud is that this is a demand tell. You do not double liquidity support in the 20-to-30 year sector because you admire the bid-to-cover. You do it because the long end is not clearing at a price you can live with, and the marginal foreign buyer is not walking back through the door.
Warsh speaks Friday at Jackson Hole carrying a barbell: pragmatic on the front end, hawkish on the balance sheet, accelerating QT. So the Fed is withdrawing duration support while Treasury adds it. The fiscal authority is running the liquidity operation the monetary authority declines to run.
July PCE landed at 3.7% headline, 3.3% core, the 65th straight month above target. Q2 GDP unrevised at 1.5%. Markets price roughly one-in-three odds of a September hike.
$BTC understood the plumbing before the strategists did, tagging $81,235 Tuesday for its first $80,000 handle in three months and opening $78,528 today. Gold is $4,640. Same trade, different wrapper.
Downstream this is a term-premium story, not a rates story. $TLT gets a mechanical bid it did not earn. $GDX carries more torque than spot metal. $KRE and the agency MBS book get relief that expires with the operations on November 4. $DHI and $LEN rally on a 30Y that has not re-rated.
My lens: long $BTC and $GDX funded by short duration, express the fiscal leg in 5s30s steepeners rather than outright $TLT, and treat November 4 as the real expiry. Invalidation is a genuinely hawkish Warsh on Friday, in which case I am early and gold pays for the wait.
On August 19 the SEC proposed Regulation Crypto Assets, its first formal rulemaking aimed at crypto offerings. Two new Securities Act exemptions: a startup exemption up to $5 million over four years, and a fundraising exemption up to $75 million per rolling 12 months. Both require narrative disclosure; the larger adds financial statements and ongoing reporting.
The part nobody is reading carefully is Subpart D, a conditional safe harbor from the term "investment contract" itself. Once an issuer has completed or permanently ceased all essential managerial efforts it promised, and makes no new promises, the token stops being the subject of a security. Proposed, not final. Comment period runs 60 days from Federal Register publication.
$BTC is up 8.2% to about $69,682 and $ETH up 18.4% to roughly $2,265, total cap $2.45 trillion. Consensus calls this regulatory clarity. It is. That is the problem.
The mechanism: Howey turns on the efforts of others. The SEC just built a documented exit ramp. Promise the efforts, do the efforts, announce you have stopped, and the security wrapper falls off. Coherent legal theory. Also a strong incentive to abandon a project on a schedule, because the moment you stop building your compliance burden goes to zero. Crenshaw made this argument about safe harbors and 2017 years ago. Nobody wanted to hear it then either.
The second-order effect is supply. A $75 million-per-year exempt path with narrative disclosure is an issuance firehose into a market whose bid is fixed. Every new token competes with existing ones for the same marginal buyer.
Downstream, separate the venues from the assets. $COIN and $HOOD get paid on issuance volume regardless of what the tokens do, and $CRCL benefits from settlement flow. Protocols with real revenue improve on a relative basis, which is the $HYPE case. Alt beta with no cash flow gets worse, because it now has vastly more competition for attention. $MSTR is a duration-on-$BTC trade and unaffected either way.
My lens: long the venues and revenue-generating protocols against long-tail alt beta. This is the same trade as the long-end one if you squint. Both are supply arriving into a bid that has not grown. Invalidation is a final rule that guts Subpart D, which would be the first time a comment period changed anything.
The SEC calendared a meeting for August 14 with one agenda item: whether to propose rules creating "a tailored offering regime for certain investment contracts involving crypto assets." It announced the meeting August 10, unusually short notice. Then it canceled, citing an unforeseen scheduling issue.
Be precise about what was on the table, because most coverage is not. This was a vote on whether to propose a rule. Not a final rule, not an exemption in effect. A proposal opens a comment period that runs months. Nothing was changing for anyone this week.
It reportedly contains a startup exemption letting developers raise an illustrative $5M over four years, a fundraising exemption capped at $75M a year, and a decentralization safe harbor letting sufficiently decentralized tokens exit securities status.
The third is the whole ballgame. Every token launch since 2017 has been a private negotiation against an undefined standard. A written safe harbor makes it a checklist. The caps matter far less; $5M and $75M are rounding errors next to what gets raised. The cancellation landed days after the Senate punted the CLARITY Act, suggesting the agency is holding its calendar open until Congress signals direction. It has chosen to go second.
Downstream this is a primary-market story, touching issuance venues rather than $BTC at $63,000 or $ETH at $1,876, which trade on ETF flows and rates. $COIN would get a listings pipeline, $HOOD the retail distribution. Offshore perp venues that never needed US registration keep their edge meanwhile, but that edge is not free money. $HYPE ran about $178B of 30-day perp volume with open interest near a record $11B, and produced $41.69M of 30-day fees, $29.02M of that protocol revenue. Gross protocol revenue has fallen four straight quarters, roughly $357M in Q3 2025 to about $202M in Q2 2026, down 43%. Record volume, shrinking take. HIP-3 lets builders keep up to half the fees on markets they deploy: good for growth, awkward for the token capitalizing them.
My lens: I am not long $HYPE at $59 on a fee story running the wrong direction. Roughly 9.92M tokens vest into circulation monthly against $53M to $83M of buybacks, so supply arriving is several times the protocol's own bid. I would rather hold regulatory optionality in $COIN than buyback math in $HYPE. Invalidated if HIP-3 grows the fee pool faster than it dilutes the take rate, making the revenue decline an investment and me a man who read a growth cost as a leak, which I have done before.
$BTC opened at $63,410.39, down 0.2%. The daily, weekly, monthly and yearly trends are all negative simultaneously. $ETH opened at $1,878.08, still roughly 62% below its August 2025 high near $4,950.
Gold is up nearly 10% over the same month.
We have now had two consecutive soft inflation prints. CPI at 3.4% headline, 2.5% core on Wednesday. PPI final demand unchanged, core at 0.2% this morning. That is the setup crypto twitter has spent a year describing as the trigger. The trigger fired twice and $BTC moved less than a percent.
Consensus treats this as coiling. My read is simpler and less flattering: the asset is no longer being bought by anyone whose decision depends on the rate path. The ETF complex has stopped adding, spot $ETH funds have logged four straight weeks of outflows, and what is left is a leveraged book trading a chart. Gold gets the sovereign bid. $BTC gets the residual.
The one place the sector still has a defensible number is revenue. $HYPE trades at $57.44 for a roughly $12.8 billion cap against roughly $758 million of annualized protocol revenue on about $172.6 billion of trailing 30-day perp volume. Seventeen times revenue for the venue taking most of on-chain perp flow is a valuation you can argue with. It is not a valuation you have to take on faith, which currently separates it from most of the top ten.
Two catalysts have collapsed onto one date. The Senate's cloture vote on the CLARITY Act is set for September 15. The FOMC meets September 15-16. The sector's regulatory framework and its rate path resolve inside forty-eight hours, and September vol is not priced like it.
Downstream: $COIN is the cleaner CLARITY expression than $BTC. Miners $IREN, $CORZ and $WULF have functionally decoupled from spot and now trade as data-center real estate, which makes them an AI capex position wearing a crypto ticker.
My lens: long $HYPE over $ETH on revenue per dollar of market cap, long September vol into the 15th, and I would rather own $GDX than $BTC for the debasement leg until ETF flows turn. That last leg and my gold position are one trade, not two. I have called ETH capitulation early twice, so discount my timing accordingly.
DRAM spot prices are up roughly 700% over the past year, per Bloomberg's July data. DDR5 contract pricing has more than doubled. HBM capacity for 2026 is sold out, with manufacturers turning away orders.
Core CPI printed 2.5% this morning.
Consensus holds that AI is a disinflationary force. Eventually, probably. Right now it is running one of the sharpest goods-price shocks in the semiconductor era, and it does not appear in a single number the Fed reacts to.
The mechanism is capacity, not demand curves. Every wafer a fab redirects to high-bandwidth memory removes roughly three times as much equivalent commodity DRAM from the market. So an AI server order does not just consume HBM. It withdraws ordinary memory supply at a three-to-one rate, and the price of a laptop module gets set by an inference cluster it will never touch.
That cost lands in capex, not consumption. Meta raised 2026 capital spending to $125-145 billion from $115-135 billion and the CFO named higher component pricing as the primary reason. That is a ten billion dollar increase attributed largely to buying the same silicon at a worse price. The five largest hyperscalers are now committed to roughly $660-690 billion combined for the year.
So the inflation is real, it is large, and it is invisible to CPI because intermediate goods are not in the basket and consumer electronics is a hedonically adjusted sliver of it. It shows up instead as margin compression and as a depreciation schedule nobody wants to discuss.
Downstream: $MU and $WDC are the direct beneficiaries and are no longer cyclicals in the way the multiple assumes. The victims are the box assemblers who cannot pass it through, so $DELL, $SMCI and $HPE gross margins. $ANET is insulated by mix. Power remains the true bottleneck behind memory, which keeps $VST, $CEG and $TLN structurally bid, and $DLR and $EQIX collect rent on the whole thing.
My lens: long $MU against short $SMCI as a pure pass-through spread, long $VST outright, and I am skeptical of any hyperscaler guiding to flat gross margin while writing checks at these component prices. What kills the trade is a memory capex response arriving faster than consensus thinks, which is the one thing this industry has historically never failed to overdo.
If you post it manually, spot-check that the eleven cashtags render as blue links — $TLN and $IREN are the two I'd least expect to resolve cleanly.
Gold $GLD is above $4,400 and through its 100-day. Bitcoin $BTC opened at $63,912 this morning, down 1.4%, after failing to hold $65,000 on Monday. $ETH opened $1,871, down 2%.
Same macro. Same inflation print coming. Same debasement thesis. Opposite tape.
Consensus in crypto is that $BTC and gold are one trade with different volatility - two expressions of the same monetary-premium bid. I think 2026 is the year that got falsified in public, and most people holding the thesis have not updated.
The mechanism is who is actually buying. Gold's marginal bid right now is central banks. The PBoC added roughly 20 tonnes in July after about 15 in June, the largest monthly addition since October 2023, and estimates put 30-35% of this rally's appreciation on official-sector buying. That bid is price-insensitive, unlevered, and does not care about a 4.7% 10Y. No reserve manager is buying bitcoin.
Bitcoin's marginal bid is levered, taxable, and funded at the front end. So in a cycle where September hike odds are live and the 10Y is at the month's highs, $BTC does not trade like a reserve asset. It trades like long-duration tech with a 3x beta. That is not a flaw in the asset. It is just not the thing people describe it as.
Downstream: this splits the complex. Miners with an AI pivot - $IREN, $CORZ, $WULF - now trade the data-center lease, not the hash price, and they should keep decoupling from spot. $COIN and the venue complex are volume-levered, and chop at $64k is the worst take-rate regime there is. $HYPE and the revenue-generating protocols hold up better than alt beta here because they earn fees rather than needing a bid. Meanwhile $GDX and silver are getting the flow crypto thinks belongs to it, and the stablecoin float keeps quietly buying T-bills, which is the one leg of crypto that benefits from a hiking Fed.
My lens: long gold and GDX as the monetary trade for this leg, not bitcoin. I'd own BTC as the pivot trade, not the debasement trade - its torque is real but it needs the front end to break first. Long $HYPE against alt beta. Invalidation is BTC reclaiming and holding $70k while real yields stay here; that would mean an unlevered bid I cannot see, and I'd take the other side back.
The most underpriced risk in US markets right now is not the Fed. It is Japan.
10Y JGB is at 2.78%. It touched 2.901% Thursday. The 20Y hit 3.90%. Those are levels not seen since 1996, and yields are up over 70bp this year. The BOJ is at 1.0%, the highest since September 1995, after hiking 25bp in June. The July summary of opinions had a board member arguing the pace of hikes should accelerate. September is live.
For thirty years Japan exported capital because there was no yield at home. That machine is reversing. When a Japanese lifer can get 3.9% nominal in 20Y JGBs with no currency risk, the case for owning 30Y Treasuries or European credit at a hedged pickup collapses. Japan is the largest foreign holder of Treasuries. This is a slow, structural, price-insensitive seller emerging in the long end at the same time the US is running historically large deficits.
That is a term premium story, and it is why I don't think the US long end rallies even if the Fed turns dovish. The bid that used to absorb duration is going home.
The second-order effect is the one nobody positions for. The yen carry trade funds a meaningful share of global risk-taking. Every BOJ step higher raises the funding cost on that trade. When it unwinds, it does not unwind gently, and it does not respect fundamentals. August 2024 was the rehearsal. The assets that get hit hardest are the ones highest on the risk curve with the thinnest real bid: small-cap growth, momentum baskets, and crypto beta outside the majors.
My read: the carry unwind is the real tail risk for crypto, not a CPI print. Long BTC, short high-beta alts. Bitcoin now has a structural bid that a Solana-ecosystem token does not, and BTC dominance at 57.2% is telling you the market already knows this even while people keep calling for alt season. Short duration in the US long end for the same reason the JGB curve is steepening. This is one trade, not two.
Weak jobs data is pulling forward Fed cut bets, the liquidity backdrop BTC has tracked closely. BTC holding $65K+ on it. My lens: long BTC / short alts as the trade on this liquidity thesis; hot CPI is the risk to it.
Whoa - more than half @hyperliquid fees within a few years
$hype so much higher
A few notes after integrating both read and write priority fees on HL:
- Many have been wondering why priority fee revenue hasn't ramped faster. It's a combination of two things. Makers need time to integrate, but the bigger issue is the validators, they have to explicitly enable the gossip priority config to respect the read auction ordering. Not all validators have done this yet, which means winning the gossip auction today doesn't guarantee you actually get prioritized access to the mempool. Rn you might end up burning your precious HYPE for nothing
- Many retail traders don't know this, but before this upgrade you needed to pay validators to get access to sentry peering or other kind of dedicated setups in order to have competitive latency as an API trader, which had a cost of a few tens of thousands of dollars per month depending on the provider. This upgrade directly targets these setups, and will internalize most of these fees into HYPE burning. It's hard to estimate the exact numbers, but I would say that this upgrade just added $500k-$1m monthly buying pressure on HYPE, with potential for write fees to grow much more.
- This is a first step from the HL team to try to internalize more revenue from latency and infra investments that most HFT teams spend. This is something traditional exchanges have been trying to do for a while (IEX introduced a 350μs speed bump and refused to offer colocation entirely, NYSE/CME keep building bigger colo facilities to capture more of this spend), and the biggest risk is destroying your microstructure because the exchange becomes too expensive for makers, or too slow. But the risk is definetly worth the reward. BIS research estimates the HFT arms race extracts $5B/yr in global equities alone, or about 0.5bps on all volume. Growth mode assets on HL charge 0.45 - 0.9bps in taker fees, so if priority fees capture even a fraction of that 0.5bps equivalent, that's potentially doubling protocol revenue on these markets. I'd estimate monthly infra spending from HFT teams on HL today to be in the ballpark of $5-10M between sentry peering, server optimization, and brainpower. Priority fees are how the protocol captures a share of that. My bold take is that trading fees will compress over time as competition forces HL to stay in growth mode for HIP-3 markets. Priority fees won't since they're driven by the arms race itself, which grows with more volume. My guess is this revenue line becomes more than half of HL's revenue in a few years if they manage to capture more flow from TradFi.
The most boner-ific thread I’ve read on HIP-3 so far
A market, an oracle, and a dream
One area where crypto has unmistakable PMF is in the creation and trading of new markets.
Over the last few years, tokens, bonding curves, and AMMs have been the core primitives for this — great for minting net‑new, onchain native assets, and leading to an explosion of experimentation with memecoins, content coins, social tokens.
However, we’ve only barely begun to scratch the surface of new markets enabled by another crypto primitive — perpetual futures.
It makes sense why: the token deployment stack has matured over the past few years, with protocols like @dopplerprotocol continuing to push the envelope. But the permissionless perps stack is only now arriving.
Enter @HyperliquidX's builder deployed perps (HIP3).
Hyperliquid exposes its orderbook and matching engine so builders can launch perp markets without rebuilding an exchange from scratch. Deployers define the market (specs, oracle, leverage limits), run it (maintain the oracle), and can iterate on UX and fees.
In other words, the technical challenges of creating new perp markets has been reduced to near zero, and new perp market creation is now primarily a coordination problem of flows, liquidity, and capital.
If you can find a group of people who want to speculate on a number going up or down and connect it to the real world with a credible oracle, you can create a market.
Market + oracle + demand. That’s the HIP3 recipe.
What kinds of perps markets will we see?
Zoom out and the HIP3 opportunity set looks like a distribution w/ a fat head, chunky middle, and long tail:
Fat head: The obvious high-volume stuff – S&P/Nasdaq style indices (@unitxyz), major FX pairs, core commodities. These already have robust reference prices and deep market‑maker coverage.
With HIP3, they can exist as perps without onchain spot inventory — all they need is a reliable index and counterparties willing to quote around it. And for many of these assets, perps are actually easier to list than spot because the oracle handles the tethering rather than needing to wrap the spot assets into tokens.
Chunky middle: This is where HIP3 starts to really shine – think private companies (@ventuals), luxury goods baskets, GPU compute cost per hour, or a city’s median price per sqft for real estate.
The common trait is that a significant number of people and/or pools of capital care about a number that updates in the world and can be indexed. They want to get exposure, hedge, or speculate.
Long tail: The permissionless frontier – the price per oz of bluefin tuna in NYC, the resale price of a Tesla Model 3, the Google Trend score of an e-commerce brand, the average nightly rate of a 5 star hotel in London, or other more wacky ideas.
This is the area of experimentation where few CEXs would be willing to venture out into. But it's where builders with a strong grasp of a niche may see latent demand where others don't.
One standard, many markets
Across this entire spectrum, HIP3 keeps the mental model simple. The oracle ties the perp to a reference number, funding nudges the perp towards that number over time, and margin tiers shape the risk that traders can take on.
The deployer is accountable for the oracle definition + ongoing publication and puts up a meaningful HYPE stake, which gives traders a credible assurance that markets are being run with clear responsibilities.
And because operators have levers for differentiation, we should expect many parallel subdexes to spring up on Hyperliquid’s rails – each competing on price, depth, and UX. This competition will be great for end users.
Why is this bullish for Hyperliquid?
While HIP3 is an amazing unlock for builders, it’s also very good for Hyperliquid. Successful HIP3 markets add flow, depth, and attention to the network. The standard ascribes a meaningful share of fees back to the protocol itself, reinforcing the HYPE value accrual flywheel.
HIP3 commoditizes the exchange layer and shifts innovation to market design and demand origination. Instead of a single venue trying to list everything, Hyperliquid becomes the base infrastructure for many specialized venues.
Expect an explosion of markets and rapid natural selection by users and market-makers for which venues and perps they want to trade.
And there’s a cultural unlock here, too. Anyone with a market and demand thesis + the ability to index it can build an exchange for users to express directional views.
The next billion users
Hyperliquid’s HIP3 opens up a vast design space that didn’t exist before — markets that appeal more broadly to normies because they're tethered to *something* in the real world.
Undoubtedly some markets will fail; that’s fine. But the winners across the fat head, chunky middle, and long tail will compound volume, liquidity, and attention back into the Hyperliquid ecosystem.
As an industry, we started with tokens, bonding curves, and AMMs to create new internet native assets — some of these experiments have worked out spectacularly. For most others, the jury is still out.
With the permissionless primitive of perps, we can create new internet native markets for virtually anything that can be measured.
It'll be exciting to see what builders come up with.
HIP3RLIQUID
The first big domino to fall
Higherliquid
Phantom is integrating Hyperliquid perps.
That’s 15 million MAUs now seeing a perp trading experience front and center on their home screen. Phantom is one of the most widely used wallets in crypto with 15 million monthly active users, $25 billion in self-custody assets, $20 billion in annual swap volume, and about 12 app opens per day per user. The Phantom integration means users will be able to trade Hyperliquid perps directly from the wallet interface, without any bridging or additional wallet setup required. If just 10% of Phantom’s user base converts into Hyperliquid traders, that’s 1.5 million wallets entering the Hyperliquid ecosystem, which would nearly double the addressable user base for onchain perp trading overnight.
On Binance, Bybit, and OKX, spot volumes make up roughly 27% of perp volumes. Phantom does $20 billion/year in swaps (spot trades). If that ratio holds, it implies ~$74 billion in potential perp flow latent in Phantom’s user base.
What’s interesting is that Phantom picked Hyperliquid Builder Codes and not Jupiter perps. In a chain abstracted world, products are no longer limited by their native stack and can focus instead on liquidity and fee economics. Builder code lets partners like Phantom add a monetization layer on top of their UX, while offering users near-institutional pricing, equivalent to Binance VIP1, which normally requires $15M+ in monthly volume.
Phantom wins by monetizing distribution. The user wins by getting low fees and better execution. Hyperliquid wins by growing its user base and order flow.
Looking forward, Phantom has significantly derisked the idea of integrating Hyperliquid. I expect more wallet frontends, interfaces, aggregators, and Tier 2/3 exchanges to follow. As more frontends integrate Hyperliquid, the protocol’s liquidity deepens, its fee model strengthens, and its backend becomes even more attractive for new interfaces.
Hyperliquid.
