Finding Edge as a Trader, The Hyperliquid Thesis & Trades For 2026 | Capital Flows

Summary

• One of the guest’s two largest bets is Hyperliquid Strategies (ticker PURR) — long from the “low threes,” now ~$5.50. The thesis: crypto spent the cycle asking how to get tokens listed on TradFi to dump bags, while Hyperliquid asked “how do you get traditional products onto crypto as opposed to the other way around.” With no way for institutions to buy Hyperliquid in a brokerage account, the largest treasury company is “like front running the Bitcoin ETF, which is very different than saying let me start a treasury company post the ETF.”
• His second conviction bet is Oracle — the only AI company that has “leveraged their entire balance sheet, their entire stock, their entire income statement, all their capex,” pulling negative returns into the present (a “50 or 60%” drawdown) and exponential returns into the future. Larry Ellison, 82, owns 40% with float shrunk by buybacks, and in the guest’s view is “trying to get the stock to like 800 bucks.” Bonus: “you can actually buy calls on it.”
• Capital Flows has pulled essentially all his capital out of systematic strategies — “I just don’t find them as very interesting or competitive as they were maybe 5-6 years ago” — and calls agentic trading and machine learning, used as tools by a discretionary trader, the “so much potential that is still not really unlocked” bet of his life. Systematic returns are fundamentally price-based; real discretionary edge lives in what “you can’t derive a signal from price alone.”
• Both speakers agree this is the golden age of the discretionary trader: implied vol blows out more than realized, catalysts can move the market two or three times, and “no one wants to take risk. Like no one” — so the trade is to take more volatility when on-side with lower hit ratio and higher risk-reward, because “the entire industry is set up to do the opposite.” The host’s proof from inside a large hedge fund: the ‘22-‘24 data-matrix trades — with the current Fed chairman serving as an advisor on monthly calls — relied on economic data; today “hedge funds are getting that headline at the same time you are” but can’t move billions fast.
• The guest’s macro frame: all-time-high valuations in every major equity index are “not an AI thing, it’s a global liquidity thing” — dollar plus yen liquidity and accelerating global trade. Citing Brad Setser: no domestic asset-liability mismatch, but a positioning mismatch, with foreigners unhedged on dollar risk. If Trump pushes the dollar down post-midterms, equities might pump 5-10% — but the 2025 tariff-drawdown mechanics (dollar sells off, foreigners sell equities) is the underpriced tail.
• The host is kicking around an explicitly unfinished thesis — “not a steadfast 100% conviction” — that clean regime change in Iran extends US hegemony and is bad for the multipolar trade: “that’s bad for the emerging markets trade, and that’s actually bad for gold as well.” Gold is the purest expression of central banks divesting from Treasuries — “if that stops, then the gold run and the silver run stops” — and if Iran resolves cleanly he’d rotate out of gold into US equities.
• On energy and supply chains: buy uranium miners, not uranium — grids are tapped out (“you can’t build any more data centers in Washington state — there’s no more electricity”) and miners make money on larger contracts “regardless of the price of uranium.” Filter everything through self-reliance: US-based rare-earth companies continue to do well; companies dependent on importing rare earths or intensive work in China are much less likely to succeed than Western-allied peers.
• On where the industry is going: leverage from code, media, and capital is concentrating in individuals — “sometime next year we’ll see someone run a $100 million Hyperliquid vault just by themselves” and make $10-20 million a year in fees.

Deep dive

1. He pulled all his capital out of systematic — the unlocked frontier is agentic

• Capital Flows’ method starts with rates: “I kind of always start with where are we at in the pricing of rates across the curve,” then connects FX and every major economic data point outward from there. The Substack (since early 2023) was an outflow of models he was already running; today he primarily manages his own capital and works with family offices, arguing the strategies he runs now can’t be learned on a sell-side desk the way they could 20 years ago.
• The striking disclosure: “I don’t really have any of my capital in systematic strategies anymore.” The last five years’ debate was “man versus machine, how do you merge systematic with discretionary” — but his current “bet in life” is that there is “so much potential that is still not really unlocked in agentic trading and machine learning,” used to map regime changes in real time as a discretionary trader.

2. Discretionary edge lives where price can’t see

• His delineation: systematic returns “are primarily going to function around price” — multivariate trend following, cross-sectional momentum, “all the lead-lag correlation guys.” So for a discretionary trader, “I think it’s kind of a fool’s errand to say, well, let me just use technicals” — that only works if you run a genuinely different time horizon. Real edge is in “what elements are very challenging to quantify simply in price alone.”
• The non-price inputs: economic and fundamental data connected to “agents in the market who are going to be forced to take action based on that data,” and information mapped “across the spectrum of uncertainty to certainty” — the Fed’s actions five years out carry more uncertainty than two meetings out. Netting out balance-of-payments and liquidity flows matters most: liquidity is “probably the most misunderstood right now.”
• The merge logic: know the hit ratio, risk-reward, and frequency of your systematic insights, then stack a discretionary insight not derivable from price — the way you’d combine several low-Sharpe strategies — “so that the whole is greater than the sum of the parts.” His dip-buying example: back-test how many times you must run the trade for positive expectancy; if a signal tells you the bottom chops four or five times before rallying, that changes how you tolerate the losers.

3. No one wants to take risk — that is the whole opportunity

• Market structure has changed: implied vol blows out more than realized because players are bigger and slower — ranges widen intraday and intraweek “but not intramonth, we always come back to the mean.” And the order book is different: “mean reversion is used as a liquidity provision mechanism in a much larger way” — everyone goes to market, no one works limits like before. Distinguishing a fundamental move from an execution-liquidity move that will mean-revert “gives you a clear signal-to-noise ratio”; he uses those events to get on-side, then holds the risk on a larger view.
• Catalysts carry outsized, repeatable significance — VIX expiration “set a lot of bottoms in 2021 and 2020,” then CPI and NFP events. “If you could know if a certain catalyst is going to have a higher probability of moving the market than not… you can actually know how to enter positions and get on size a lot easier.”
• The behavioral core: “everyone in the industry right now, no one wants to take risk. Like no one” — everyone cuts as soon as they’re up on their basis. So the discretionary trader’s edge, especially a technical one, is to take more volatility when on-side, run a lower hit ratio with higher risk-reward — “the entire industry is set up to do the opposite.”

4. Inside the big fish: why the small fish now win

• The host’s testimony from inside “one of the biggest fish on Wall Street” — a large hedge fund: every repeatable money-maker in ‘22-‘24 was ingesting economic data through pre-built matrices — “if number comes in X, then we short this much; if number comes in Y, then we go long this much” — executed instantly on release. And the information was walled: “the current Fed chairman was an advisor that we would get on a call with once a month… he gets lunch with Jerome Powell like once a month. You’re not competing with that.”
• Today the game favors the small: headline risk hits every month — Iran mining the Strait of Hormuz and oil jumping $10 — and “hedge funds are getting that headline at the same time you are, basically,” but with billions to move. The host knows oil traders who simply couldn’t hedge the recent rally; they were too large. One down month and subscriptions get cold feet — which is exactly the skittishness showing up in implied vol.
• The guest extends it to structure: leverage from code, media, and capital is concentrating in tiny teams and individuals. “I think sometime next year we’ll see someone run like a $100 million Hyperliquid vault just by themselves… they’ll probably make 10 or 20 million dollars a year if they’re good at trading.”

5. Information diet: headlines on X, long-form research, and “why are they wrong?”

• The guest’s system: Twitter strictly for breaking headlines, most of the day spent reading long-form research and taking notes — 13D Research “is phenomenal” (plus Capital Flows: “he actually didn’t pay me to say that”). The job is to “figure out what the mega trends are at any given moment and really just shove my capital into it,” while staying informed enough to instantly interpret any headline for any asset.
• The sell-side exercise, worth keeping verbatim: “You read it and you just think, why are they wrong?… it’s up to me to figure out why it’s incorrect and can I prove that it’s incorrect” — training the brain to pick up patterns so that “you’ll see something that rhymes with it in 6 months and then you’ll be able to take that trade again.”
• The guest’s case for thinking alone: at funds, “it’s very hard to have the intellectual freedom to think and come up with high-quality ideas beyond just taking the other side of a positioning move” — everyone is collecting carry, everyone hears the same desk chatter, and “everyone is fading each other.” Solo, his iteration speed “has increased like 10x over the last 6 months. It’s honestly surprised me even.”

6. The Iran thesis: a clean regime change is bad for the gold trade

• The host faded the 12-day war on pattern memory: Iran’s 2024 retaliation against Israel dipped markets 2-3% and immediately rebounded, so “I’m going to fade this as well” — concluding it won’t escalate into a regional war and “everything is still a buy here.”
• The bigger thesis, offered with its hedges intact — “this is not a steadfast 100% conviction… I’m still working on it”: post-Iran regime change “so severely handicaps the apparatus that China has built” that “the US has extended their hegemony. That’s bad for the multipolar world trade, bad for the emerging markets trade, and that’s actually bad for gold as well.” Europe, which viewed the US as an unstable partner (Greenland saber-rattling, building independent reserves), “is probably going to come back to the US” post-Iran; China remains isolated. Expression if it resolves cleanly: “diversifying out of my gold positions into US equities again.”
• Within metals, gold is singled out: it’s “the purest expression” — stockpiled by central banks as divestment from US Treasuries — “and if that stops, then the gold run and the silver run stops.” Copper is going up for industrial reasons and US-based rare earths continue to do well; he’s “not necessarily super keen on fading” the rest.

7. The guest’s frame: it’s not an AI thing, it’s a global liquidity thing

• His read on the gold/EM move: it’s the combination of dollar liquidity plus now yen liquidity at the same time global trade accelerated. The only crash scenario is trade shifting back to the US — which becomes plausible if autonomous manufacturing changes PPP between countries. That, to him, is why the AI race and compute matter: China knows it’s “like 3 years away, 4 years away… and if that shifted, then the US has all the cards.”
• On valuations, citing Brad Setser: “we don’t have a domestic asset-liability mismatch, but we have a positioning mismatch, which is why every major equity index in the world is at all-time-high valuations. It’s not an AI thing, it’s a global liquidity thing.” The value investors who’ve called the crash for five years are “the boy that called wolf — you can’t really listen to them anymore.”
• The tail scenario he’s thinking about but not betting on: Bessent — and “especially Kevin Warsh… when he comes in” — know they have two years, and Trump is running it straight into midterms (“he genuinely just does not care”). If they push the dollar down against the yuan and force more cuts into the curve than priced, equities might “pump 5% or 10%” — but with foreigners unhedged on dollar risk, the 2025 tariff-drawdown mechanics (“dollar sells off and equities sell off… the primary reason was foreigner selling”) make the downside “a lot bigger than people have assumed.” Post-midterms is “very very consequential.”

8. Uranium miners, not uranium — and only Western supply chains

• The host’s energy call: long uranium because every grid is tapped — “you can’t build any more data centers in Washington state because there’s no more electricity” — and nuclear has to power the buildout. The instrument matters: “you want to buy the miners” because he has no view on deposit discoveries; miners get “much larger contracts to produce more uranium, and with larger contracts, regardless of the price of uranium, they’re going to make money.”
• The filter across all of it is self-reliance: US-based rare earth producers continue to do well, while “any companies whose model rests on importing rare earth minerals or intensive work in China” are much less likely to be successful than Western-allied peers — because even if America reasserts itself as number one, the US-China fracture keeps growing.

9. The 2026 book: PURR and Oracle, everything else is defense

• One of his two largest bets: Hyperliquid Strategies, ticker PURR — long from the “low threes,” now about $5.50. The setup: crypto “shifted from creating value to let’s just get these tokens listed so we can dump our bags,” becoming its own establishment — “oh, you should buy Bitcoin. If you don’t, you’re like stupid and poor.” Hyperliquid inverted the question: “how do you get traditional products onto crypto, as opposed to how do we get crypto into TradFi?” Once Hyperliquid is added to a US regulatory framework and CME/Kraken run 24/7 perps, arbitrage compresses funding rates and more capital flows in.
• Why the treasury wrapper: there is essentially no way for an institution to get long Hyperliquid in a brokerage account today, and PURR is the largest treasury. Everyone got burned on Bitcoin treasuries last year and lumps this in with them — but “this is like front running the Bitcoin ETF, which is very different than saying let me start a treasury company post the ETF.” He sees this as different from the Bitcoin treasury setup.
• Second bet: Oracle. Larry Ellison — 82, “one of the absolute savages who actually takes risk, not like all these other tech bros” — has levered “every single part of a company that you could leverage”: balance sheet, stock, income statement, capex. That pulled the negative returns into the present (a “50 or 60%” drawdown) and the exponential returns into the future; the guest thinks a bottom is forming (the stock was up a little; the host noted a nice after-hours pop). Ellison owns 40% after years of buybacks — “there is no float in a sense” — was richer than Elon Musk last September, and is “trying to get the stock to like 800 bucks.” “Oracle’s nice cuz you can actually buy calls on it.”
• Everything else is defense — rate trades and hedges against the two big positions. The host’s sign-off, tongue in cheek: “If you buy PURR and Oracle, you can retire a billionaire at the end of this year. Capital Flows has said it.”