Exodus: Out of Wall Street, into DeFi
Reflecting on 12 weeks in cryptolending, after 12 years on Wall Street
Out of Egypt
I haven’t posted in a long time because around three months ago, I transitioned into the crypto-lending industry after 12 years in the hedge fund world. And despite the lower pay and prestige—what a breath of fresh air.
When I joined the HF world, the archetypical hedge fund felt like a crypto-corporation (a “DAO,” or Decentralized Autonomous Organization) does today.
Respect for others, without HR hall monitors treating everyone like a child.
Respect for laws and ethics, without compliance hall monitors treating everyone like a criminal.
5-30 total employees.
A high shared appetite for risk, not just to get rich, but for love of the game: we aren’t like those corporate/sell-side pussies. You want to do something cool, you better a) be willing to risk something valuable, or b) be willing to invest a lot in a product with a low probability of initial success.
A high appetite for unconventional perspectives.
Getting hired for a middle-management role consisted of a quality reference and at most 3 in-depth interviews, not an initial HR filter followed by 10-step, 10-week decision by committee involving lots of non-stakeholders.
A product (high-alpha returns) that you genuinely felt proud to be a part of.
Sure, TradFi and HFs have “gone corporate” in many annoying ways, like the rest of society, over the last 15 years. But what broke the HF model, and TradFi in general, is that the industry no longer believe in their product, end-customer, employees, or innovation in their core product (company- and industry-specific research and forecasting). Hedge funds today that have achieved “scale” (>$1B of unlevered assets under management) are not only uninterested in innovative research, they will actively refuse to underwrite it 9 times out of 10.
Example: You want to get some real, scalable edge on Snapchat (SNAP), which you’re the lead analyst for. After studying the company, you determine that its ad load in English-speaking markets is quite high, i.e. its incremental growth opp in those markets is fairly low, but its ad load in key emerging markets like the Middle East, France, and Brazil is very small. Therefore, SNAP’s ability to hit its annual revenue targets is very dependent on how quickly it can scale its advertising in these key emerging markets.
You pitch your portfolio manager on creating 20 fake Snapchat profiles in different geolocations in Dubai / Mecca / Beirut / Istanbul / Rio / etc, and gathering all ads on each profile’s Discover page every day, to see how SNAP’s ad load is trending those key emerging markets—thus gaining a key edge on SNAP’s revenue momentum over the next several quarters. You use outsourced proxy servers in a foreign country, such that your ultimate identity is impossible to discover without the company serving a warrant to that country and forcing that company to cooperate—thus reducing your de facto legal exposure to zero. You figure it will take around 50 man-hours of work, spaced out over several weeks, with a couple of engineers to spin this up. All in, it might cost $20k worth of time, servers, etc to build a running machine that has a 50/50 shot at creating real, scalable alpha in SNAP, or just being noise.
Your PM will likely dismiss your idea because, besides the uncertainty of the payoff, it will take several weeks of building, and 3-5 months of deployment, before you can make high-conviction stock calls based on this data—and your PM probably has 2 quarters’ visibility at most into whether he will still have a job at that fund by then.
Your employer’s compliance officer will be completely dismissive because creating “fake profiles” violates a Terms of Service clause on each profile’s SNAP app signup, which theoretically exposes the organization to a less-than-minuscule reputational liability in the less-than-minuscule hypothetical scenario that SNAP a) sees what your fake accounts are doing, b) links them all together, and c) decides to blacklist you as an investor for showing such a different level of interest in their company. Never mind that none of what you are doing is remotely ethically questionable or illegal: you’re just trying to do some differentiated goddamn research!
The compliance officer genuinely doesn’t give a shit if he is totally wrong in terms of case law, legal theory, or practice. His incentives are identical to a government employee’s: Cushy job. Risk nothing, lose nothing. Therefore, say “No” to 99% of unconventional things.
And, most importantly: The fund manager appreciates that his compliance officer is making bad risk-adjusted research process decisions. The vast majority of fund managers believe much more in asset management fees than in performance fees, and potential liability of lost assets from bad press dwarfs the potential asset of creating real alpha in researching a particular stock.1 Voila, a conspiracy of bad incentives has turned a business of calculated bets has turned into an asset-management fee-farming business.
In broader TradFi, the analogies are the same. Banks walked depositors’ yields down to 0% during the bogus permanent crisis of 2010-2015. Banks consolidated. Interest rates gradually rose again during the Trump years. The too-big-to-fail megabanks all still offer 0% on deposits (plus plenty of random invisible fees), but don’t call it cartel collusion.
Large banks themselves were quasi-nationalized by Dodd-Frank and are unable to take many types of risk, like offering HELOCs in many states. But they can loan hundreds of billions of dollars to Blackstone and private equity at a 4% rate / 80% LTV to buy apartments, put new finishes on the units, and flip them.
The typical saver depositor is getting 0% yield when official CPI is 6.5%. As I’ve proven in other inflation-themed posts, the CPI today is at least 3 points too low due to the government’s inexplicable calculation of rent (30% of CPI). Add in random fees for things like overdraft charges, and a depositor today is getting an effective APY of -10%.
My fascination with DeFi was born last summer when I first saw @SebVentures’s Dune dashboard showing the explosive profitability of the Maker DAO, warts and all, during a time of severe famine-to-feast period in crypto (2Q20). I grew more fascinated still after reading @hagaetc’s post, “The Revolution Will Not Be Reported Quarterly.” That post, more than anything else, was the “light bulb moment” for me: it showed a fully functioning corporation’s financials in real time. No quarterly earnings calls (although they do Twitter AMA’s and updates much more frequently anyway). No gated access. It showed a functioning bank with 19 employees, $5 billion (if I remember correctly at that time) of assets under management, all kinds of lending data, and way more profitability than a TradFi bank. This is the future of banking.
More transparency, more scalability, no suits, no status tokens (expensive offices): This is the future of money. This is the future of financial data analysis.
Who would need screen scrapers and investor conferences and millions of dollars to sell-side banks and all that other expensive, gated, mediocre-value-added, buy-side-country-club shit, when you can just see the corporate ledger in real time?
Who will need to schedule 50 calls with investor relations every quarter, to write up a pointless document about your conversation with IR next to your conversation with the same compliance-whitelisted industry expert that everybody else also talked to, trying to show how your analysis was truly variant relative to “the whisper” and “the fast-money consensus” and “the vanilla consensus”?
Who was going to hire “senior analysts” 10 years from now, whose job today—regardless of their innate creativity—mostly consists of being a very highly paid stenographer of managements’ earnings calls, IR calls, and other boring, public interactions? What was the point of trying to be different when you can only do the same compliance-gated commoditized analysis as all your competitors?
Then I thought some more.
If data was on-chain, and governments paid in cryptocurrency, we wouldn’t need to know which collection of nonprofits were favored with the $700m of “mental health” that Chirlaine McCray, wife of NYC’s corrupt mayor, had responsibility for, which nobody can find today. We’d just be able to see where the money went. How bearish would blockchain adoption be for government incompetence/corruption, which is incidentally a huge passion of mine?
If the private sector gradually adopted cryptomoney, we would just devalue all the debts, all the stupid choices the Boomers and GenXers made for us. They would still be owed their shitcoin debts, but we’d have a more valuable currency by which to repay it. We’d actually break free of the enslaving web of collective obligations which has created so much strife for all of us, and so little freedom for the young. We wouldn’t have to trust preening, corrupt, alwayspromise-neverdeliver GOP politicians to shrink government. We could just … financially secede on our own. (Theoretically.)
It took me a long time to learn the jargon-laden language around DeFi, and a long time to understand DeFi’s nutty presentations of “yields” (which do it no favors in marketing to wealthier people, since so many of them come across as Ponzi schemes). But over time I did realize that, yes, 5-10% yields for depositors at very low risk are truly possible; yes, higher-risk “liquidity mining” in the lower-to-mid-double-digits are possible; yes, savers can actually save without being fucked at the same time.
It also took me a while to get past crypto’s formidable learning barriers. I resigned myself to burning a couple thousand dollars just to figure out how to move money in crypto, which can turn insanely expensive insanely quickly if you don’t know what you’re doing.
But the more I learned, the bigger the implications were. As “late” as it feels to get into crypto, it is still incredibly early on a 5- or even 3-year basis. Global fixed income is a $33 trillion market yielding negative real returns. Global equities are priced for perfection relative to global equities, but are priced very attractively relative to global sovereign bond markets. Gold, which has marginally better scarcity than ETH and equivalent scarcity to BTC (but none of BTC/ETH’s fungibility and none of ETH’s programmability), is a $12 trillion global market. Money must flow where depositors can get yields on deposits. The market cap of all major banks and payment companies worldwide is
(if I remember correctly) something like $15 trillion $10 trillion.2
The market cap of BTC, ETH, and its more credible competitors? Around $2 trillion.
It took some embarrassing rejections. I remember applying to Messari, which wants to be a kind of sell-side research clearinghouse for crypto. The interviewer asked me if there were any projects I was interested in. I said Solana. Cringe.
But after another 9 months of doing another TradFi job (which I hated) while still hacking away looking for a crypto job that didn’t involve high-octane coding or high-frequency trading (way too much talent there, and while I can code a little, it’s no strength of mine), I lucked into a very old friend whom I’d fallen out of touch with, who had become a relevant player in DeFi, and after I asked him to work evenings and weekends for free for a month, he pitched his new startup and offered me a good seat there.
Suddenly for the first time in years, I loved working again: the unexpected parts, the expected-but-fun parts, the shitwork-out-of-nowhere parts. For the first time in a long time, 70-hour work weeks felt like 60-hour work weeks, and researching my industry sucked up most of my free time.
Oh, and I started making money immediately: stable-and-sorta-safe, 95%-gross-APY LP yields on Anchor.
You might be wondering at this point how “alpha” shops that cut their teeth in equities put up these cray-cray numbers year in and year out. The open secret is that a) they don’t actually make money in fundamental equities anymore and b) they need to employ large-to-gigantic of leverage to achieve returns which, even in Citadel’s case, are thoroughly mediocre on an after-tax basis for any taxpayer.
Point72 makes its money mostly from aggressive market-timing bets, which they’re best-in-class at.
Citadel, despite employing 9x leverage, makes little P&L and has astronomical turnover within its fundamental-equity businesses—Global Equities, Surveyor and Ashler; as well as Ravelin and Aptigon, which were liquidated after bad performance.) Most of its returns come from equities market-making, fixed income (a much more insiders-club business where trades still happen over the phone), and commodities market-making.
Balyasny is probably the most genuine “equity alpha factory” model at this point, but its 5- and 10-year returns have been much more inconsistent than peers’.
Millennium’s equities business has been an underperforming shit-sandwich for years, and MLP makes most of its money from very concentrated fixed-income bets.
In an “ordinary” year, Citadel will make 15-20% net of a 30% performance fee, a formal management fee of 3 percent, and a cost passthru of 2-5 extra percent on assets (overall probably a “6 and 30”). That 15-20% return will be taxed at ordinary income rates if you are a taxpaying US citizen, leaving you with 8-12 percent of net return on a business employing 8-9 times leverage.
A key factor in this change was the Trump tax cuts, which removed tax deductibility of HF management and performance fees. Pre-Trump Tax Reform, a HF investor could deduct both management and performance fees as expenses, thus recouping at least half of the management fee.
Very well-connected hedge funds can create OTC tax swaps where they magically classify ~70% of short-term trades as long-term trades and dare the IRS to sue. That option, which comes with significant legal risk to the enabler and hefty fees, isn’t available to startup hedge funds. It can theoretically bring the blended tax rate on hedge funds’ gains down to ~35% instead of 50%+.
https://companiesmarketcap.com/financial-services/largest-financial-service-companies-by-market-cap/ (minus BTC/ETH)