Discover more from Alex’s Substack
4 crypto investment theses
What 1.5 years in crypto market taught me about crypto's future
State of the crypto market, May 2023
Crypto’s investable market cap (longable or shortable at institutional scale) today consists of the following groups of tokens:
ETH + ETH scaling solutions: $260B
CEX tokens (eg BNB) & other centralized ledgers: $100B+
Credible ETH competitors: $50B+
Tried-and-failed ETH competitors: $50B+
ETH dapps: $40B
Non-ETH dapps: $15B
Privacy chains & dapps: $5B
There are ~200 Binance perp markets that trade >$5M/day on Binance alone.1
Thesis 1: Governance activism in blue-chip crypto dapps & chains offers enormous unexploited alpha not only for successful activists, but also for web3 ‘management’.
Thesis 2: Building a validator brand centered around professionalized governance activism is an untapped institutional-investor opportunity.
Thesis 3: There’s lots of alpha to be had from shorting failed ETH competitor chains and dapps to zero. These represent massive concentrations of bagholder capital which failed to find product-market fit, who are issuing $150-200M of new tokens every week.
Thesis 4: Since 99% of crypto tokens and projects go to zero, but individual tokens are also capable of extreme localized idiosyncratic volatility, shorting altcoins is another heavily underexploited opportunity, which should be optimized via an index-based approach.
Part 1: A few lessons from 1.5 years in crypto product
Lesson 1: The burden of proof is on crypto to show real-world PMF
2.5 years ago crypto had achieved undisputed product-market fit in 3 things: digital gold, online gambling, and 1-5% for very shady stuff (scams / darknet / organized crime activity, etc).
On the surface, nothing has changed, besides the “extremely shady” percentage of transactions dropping.
As someone who worked on the front lines of crypto product for the past 1.5 years, crypto’s failure to expand its PMF has been a huge disappointment. I don’t think it was all crypto’s fault (Chokepoint 2.0 was devastating for RWA tokenization and anything else trying to integrate crypto with the real world).
The bull cycle of 2021 brought us a lot of business models and “products” that ditched crypto as soon as crypto’s user base receded (web3 games, web3 fitness apps, web3 real world asset tokenization, DAOs, web3 decentralized file storage, web3 decentralized cloud computing, …). It popularized a couple of older products (stablecoins) and created one wildly successful new product (the decentralized exchange, or dex).
Crypto will always retain potent call options on dumb monetary policy, global financial balkanization, the Western banking system being regulated into censoring more and more kinds of explicitly legal activity in the name of regulatory compliance, and a loss of confidence in Western governing institutions.
But dumb monetary policy, absent catastrophe, won’t onboard the next 100 million users. It’s hard to see crypto capturing the imagination of people again if it is unable to utilize its blockspace to provide real products to people that makes their lives better.
Crypto did push the innovation envelope forward in many ways which were drowned out by the macro, regulatory, and self-inflicted contagion of the past 12 months.
After a massive boom and bust, Solana and perhaps Avalanche seem to have achieved self-sustaining high-quality developer culture and user community.
The Cosmos ecosystem survived the Terra financial earthquake and, like Solana, gained significant developer share during the 2022 bear market. The Cosmos SDK is arguably the most credible successor to Ethereum in terms of technological architecture and credibility among elite developers.
Ethereum, of course, executed a massive upgrade with no glitches, and saw the rollout of several scaling solutions which modestly solve some of its uncompetitive bottleneck issues
However, all of these ecosystems feel a ways away from launching groundbreaking products that can broaden crypto’s appeal on their own terms outside of the speculation use case.
On a personal note, as far as variant perception goes, I am especially interested in potential Solana synergies with AI/ML (Solana has very high tech requirements consistent with ML GPUs, and also with decentralized physical infrastructure for purposes of supply chain auditing and tracking of real-world inventory and other assets.) I would expect use-cases that prioritize deregulation/gray markets to stick to either Ethereum or, if unit economics matter more, Cosmos.
Crypto does not need another 20 L1s promising to be the next Ethereum. At a high level, crypto is a 1993 version of Windows running on 2023 hardware. But since institutional crypto is still 99% VC-driven, it’s very frowned-upon to point out that crypto (still) has over $50B of stranded circulating market cap in L1’s which have failed to get traction among developers and/or normie crypto users.
Crypto L1s and dapps essentially have one shot to be successful, to a much greater extent than tradfi entities. They need to rally an extremely multicultural (global) user base around a common idea, work together across very different timezones and countries (~all crypto companies are remote and multinational), and have a very short time to turn lightning in a bottle—an excited community of newly engaged users—into a Minimum Viable Product. If the core team loses the community’s confidence, the token value quickly plummets, the core team has exponentially accelerating incentives to “rest and vest”, grifting the project for millions of dollars while they quietly start up a new protocol or blockchain. Noncompetes and legal constraints do not exist in crypto, so there is nothing stopping core teams from walking away from difficult or unmanageable situations.
This can also happen via hacks and rugpulls, in which one developer creates a vulnerability which is later exploited from a newly anonymous address to steal capital from e.g. a DAO treasury. Many of the “hacks” in crypto are in fact inside jobs by a discontented member of the core team.
Adding to the above issue, crypto developers trend idealistically clueless when it comes to adversarial game theory (which is really bad news when you are dealing with crypto’s smaller numbers of high-ASP transactions and high concentration of scammers).
Lesson #2: Stop wasting time on problems where crypto is at a disadvantage
In my brief time in crypto I have met many builders who throw their passion and intellect at difficult problems without understanding what crypto can and can’t do better than existing web2 infrastructure. This would cause them to waste a lot of effort attacking problems which blockchains have no business trying to solve.
A blockchain is an expensive cloud-based database, where anywhere from 20 to thousands of validators independently validate a transaction before that record is appended to the blockchain. This makes a crypto database exponentially more expensive to run relative to, say, an AWS database, which means you shouldn’t be using crypto for that product unless the use case of that product benefits from decentralized validation in some compensating way from the much higher infrastructure cost per use case. We haven’t even touched on the abysmal state of crypto’s UX yet, or how the fragmentation of different blockchains is a UX nightmare for any app builder trying to reach different tribes of crypto users.
Those use-cases are many, and many of them are still untapped. However, they revolve around higher-ASP use-cases involving regulatory uncertainty, tax uncertainty, political uncertainty, or other kinds of high-ASP-meets-high-distrust situations.2
A blockchain should probably never be used to trade carbon credits, for example: that is a very high-trust use case between very large institutions and governments, where hundreds of additional “validation” steps per database update amounts to red tape dressed up in ones and zeroes. But in crypto, you constantly see money thrown at use cases like this, when existing TradFi legal/financial infrastructure is actually better suited to do the job.
Similarly, in areas where latency and “frame rate”/transactions per second matter a lot for UX, it’s odd to think crypto has a particular advantage. “GameFi” sticks out like a sore thumb here.3 Gamers care about a smooth, sensory-captivating gaming experience. Gamers don’t care at all about decentralization. To the extent that existing crypto users represent a sort of captive audience for blockchain developers, that’s one thing, but the web3 stack seems to offer far more liabilities than assets to the gaming market.
Lesson #3: BTC/ETH can accrue value on current use cases in the absence of new PMFs, but it will become a much more boring asset class, akin to traditional art
Crypto can be thought of as a giant casino of libertarian gamblers who derive comfort from self-custodying their assets. However, the house’s (whales’) winnings are not recycled out of the crypto casino—they’re recycled into BTC and ETH. Large BTC and ETH holders consider BTC & ETH safer as well as Sharpe-superior assets to cash or even stocks. The Michael Saylor mindset is fairly common among this group.
At a meta level crypto resembles the market for expensive art more than gambling—a highly volatile slice of global discretionary income speculating on mimetic momentum, while crypto-specific fundamental criteria (governance cohesion, community quality, efficiency/decentralization tradeoffs, legitimacy among other developers) drives longer-term outperformance. As with expensive art, crypto also offers very significant tax advantages over traditional assets.
The process for valuing crypto L1s/L2s/dapp tokens (not just NFTs) has a lot more in common with traditional art than with traditional securities.
Lesson #4: Thus far, crypto has failed to align stakeholder incentives intelligently
One of the biggest unforced errors in web3 over 2021-23 was the failure of web3 to properly align incentives between founders, investors, the broader community, and other stakeholders.
The foundational crypto-corporation primitive of the 2021 bull market was the DAO. Its main differences with a corporation were: more horizontal hierarchy, more anonymous leadership, more ‘inclusiveness’, and Animal Farm-meets-Lord of the Flies idiocracy.
DAOs’ top contributors would tend to become hindered or abused by the bottom 80% of the DAO. Many DAOs saw splits between a minority of “doers” and a large majority of weak contributors who leeched onto the project in its early days
The hypercyclical nature of crypto accentuates this problem. Crypto tends to draw in gushers of high-passion, low-quality users.
DAO hierarchies were generally based on time joined, not work contributed.
DAOs weren’t all bad. Successful DAOs leaned hard into community engagement and brought out the best of above-average communities, like the Maker DAO and the Curve DAO. However, in aggregate, the model not only failed, it caused much more value destruction than is commonly appreciated. Something more specialized and hierarchical, but also proof-of-work based, is needed.
Part 2: Where are the market cap disconnects?
I see 3 large pockets of disconnect between crypto PMF viability and current market caps.
ETH’s dominance among smart contract L1s won’t last in a growthy or flattish environment: Technologically and politically, ETH seems unable to achieve consensus around easily solvable problems (MEV sniping, the re-entrancy vulnerability of the ERC-20 token standard). Scaling solutions will damage ETH’s composability to the extent that it mitigates ETH’s cost/congestion issues, and ETH rollups are one SEC subpoena away from melting down. ETH is far too expensive to accommodate the next X million users and doesn’t have the political will to materially change these circumstances. ETH has chosen a path of extracting maximum rent on its current market, and has conceded a major opportunity to Cosmos, Solana, and/or some other chain.
Thesis: ETH rollups will dilute ETH’s composability without growing ETH transactions in a way that will accrete to ETH holders. Over time, the ETH blockchain will morph into a chain of rollup-chains, aka a highly inferior implementation of the Cosmos chain-of-chains architecture (much less security, less memory efficiency, higher tx costs). Rollups will not “save” ETH’s PMF absent other fundamental changes to how ETH operates.
ETH is very socially decentralized, and building consensus around major roadmap changes takes a long time.
Thesis: ARB’s US domicile is a gaping regulatory vulnerability. OP (UK) and MATIC (India) don’t have this vulnerability; however, if ARB’s founders are subpoena’ed, expect a run for the exits, with extreme logistical chaos, for all ETH rollups
Thesis: SOL’s super-low transaction costs give it a unique edge in regulated, high-frequency, decentralized physical infrastructure (dePIN) and real-world-asset use cases, where offchain regulations already apply and decentralization has the least marginal value.
Decentralization is a double-edged sword with regard to regulatory risk. If the US government sues you, you need to have money to fight back while you maintain some sort of legal status. The US government has made clear that it will sue anybody. A “decentralized” L1 without the money or willingness to cover L1 devs’ legal exposure is much more vulnerable to state pressure than a centralized L1 with intelligent legal structures and capital reserves.
At the same time, most of the other Ethereum alts – over $50 billion of market cap – are 12 to 24-month death spirals.
(anecdotal) Higher-end blockchain developers show the highest marginal preference for the Cosmos SDK, assuming their golang/rust skills are equal to their solidity skills
It seems very hard to find good Rust developers though
Because it’s clear to many people that ETH is very poorly optimized to foster mass adoption and there are no barriers to launching new blockchains, there’s always a deluge of pseudo-credible ETH alts, which fragments liquidity across many insufficiently-differentiated competitors.
The dilemma of an ETH alt is: how are you different enough to build a sticky core of talented blockchain developers who will build new apps around your operating system?
In crypto there are only 3 sticky differentiators that can’t be copied from one blockchain to another: culture, decentralization, and the base programming language, which determines developer learning curve as well as intrinsic efficiency of that blockchain.
A new-and-materially-more-popular programming language becomes a differentiator over time relative to more established communities, but isn’t a moat in and of itself
Launching a blockchain is very easy, but getting a blockchain off the ground is extremely capital intensive and requires tens of millions of dollars in working capital.
The rates of these alts’ spirals will be driven by how distributed they are, ie how much of the token supply is still locked up and in founder hands today. ADA, for example, is extremely widely distributed, so its underperformance would be much more gradual than, say, DOT’s.
Thesis: Around 15 ETH alternative chains with circulating market caps of $500m-$15bn, and another 50+ chains circulating market caps of $100m-500m, are zeroes.
Governance activism in dapps is a major untapped opportunity: Any team that has managed to build a successful dapp on a DAO model has something rare and precious. DAOs introduce a level of drama that needs to be experienced to be understood, which has no comparison to corporate life. It requires rallying people from dozens of different countries around a common vision, putting up with an unheard-of level of abuse from retail investors who often have no clue what they’re talking about, refusing the temptation of grifting a huge community that’s extremely clueless about inner workings of the product yet extremely opinionated about it, and tuning these different tribes out in order to economize focus. Good teams always need help from value- and product-focused activist investors, and those investors in turn get front-row seats into major political shifts which have enormous consequences for token value.
Teams which can manage DAOs, tune out the noise, and keep shipping have a level of talent and cohesion that’s extremely rare. They will probably make you money whatever they do next.
Inflation. Like other asset markets, crypto markets price inflation in advance. By the time inflation is actually happening, it’s not bullish for crypto (just like it isn’t bullish for stocks or bonds) unless that inflation is getting worse faster than policymakers or the real economy are countering it.
Global real discretionary income. Assets other than BTC and ETH are extremely out-of-the-money call options on various concepts related to untaxed, deregulated, borderless capital flows. Until those assets incubate businesses that generate real world utility, these altcoins’ premium is decaying in multiple ways: they are falling behind chains/dapps who are building real things (many are), these altcoins are issuing tens of millions of dollars of new tokens per week.
Regulatory. The US regulatory assault on crypto threatens to set crypto back 2-4 years. The USD doesn’t maintain its dominance because the USA has more aircraft carriers or nukes than everyone else; the USD is the only truly global “financial operating system.” This applies especially for cross-border transactions, which is what matters for crypto, since crypto is itself borderless. If crypto cannot be interoperable with the USD, it is cut off from 99% of its attainable real-world use cases. It is of existential importance for crypto to win its current battle with the US government. None of us has the time to wait 50 years for a non-USD reserve currency to take over. We need to interoperate with the real economy to access diversified cash flows.
The outlook for the 2024 election, at this time, is not good.4 The only regulatory relief in sight before Nov. 2024 is if SCOTUS materially guts Chevron deference, the 1984 judicial precedent that licensed regulators to do virtually anything within their regulatory domain as long as the law doesn’t prohibit it.
Crypto altcoins have extreme bouts of idiosyncratic volatility while also having very high correlations with one another over time.
Daily trailing 30 day correlations are stratospheric in bear markets as you’d expect,
… but are pretty high in bull markets too:
So, diversification is difficult to find on a minute or hourly basis. Diversification over days, however, is significant, and comes in a few ways.
Genuinely differentiated product offerings
Quality/pedigree of core dev team
Like the hedge fund industry, “good educational pedigree + youngish + not filthy rich yet” is the way to go
Different rates of token issuance / dilution / balancing stakeholder incentives
Decentralized ownership (more decentralized ownership means more people have bought into the narrative in some way and have a vested interest in sticking around; this means tokens with terrible/missed-their-shot PMF like ADA trade very differently from other tokens with terrible PMF)
Conferences dedicated to that specific L1
Quality of community - every crypto community is represented by influencers, who act like tribal shamans trying to rally part of a community around a set of quasi-religious beliefs. Some influencers are good. Most are grifters. The level of their grifting / BS is a good insight into the IQ of the community and the likelihood that they’ll be further grifted. Further grift = token go down.
A perp (perpetual future) is an option-like instrument popularized by crypto which has mostly taken the place of the equity option, although markets for American options on crypto also exist. A perpetual future is speculation on where the spot price will close in X hours (usually every 4 or 8 hours). To the extent that there are more longs than shorts or vice versa, the crowded side of the perp trade refunds the uncrowded side every X hours, when the perp is reset to the spot price. This instrument ordinarily charges a ~12% annualized funding cost per turn of leverage used, and goes higher if one side of the trade gets more crowded. However, the vast majority of perps skew long relative to spot, so as a short-seller you are refunded that difference. For non-crowded shorts, net borrow costs are basically zero, or actually skew positive.
Perps are significantly more intuitive for retail users than US equity options, have extremely narrow (<.1%) spreads, and have comparable funding costs. They are also not recognized by the SEC, so Binance and other CEXes do not officially offer perps for American customers. Perps are needed to have institutionalized access to crypto short-selling. Spot short-selling is also possible, but access to borrows is more complicated hence very suboptimized relative to perps.
A list of those use cases could include:
Real-world asset tokenization (Japan, India/SE Asia, LatAm)
Moving beyond the “DAO” governance structure, which turned into an Animal Farm-esque failure
Some kind of KYC passport, enabling publicly-anonymized participation in permissioned pools
Online pharmacies and other gray-market businesses providing a clearly legal service within their own jurisdiction to people in other jurisdictions
Once a network like Helium were physically built-out more, real-world asset tracking and supply-chain auditing will become much more viable.
Every crypto investor talks up web3 gaming. I am sure crypto will have its Candy Crush product at some point (games don’t have to be hardware-intensive to be fun), but thus far, GameFi has amounted to “pay kids in Vietnam and Pakistan in in-game items, aka shitcoin-NFTs, to play your shitty game 24/7.” Paying early adopters to use your product is terrible practice when it comes to assessing PMF, and gamers hate the NFT-ization of game items. Shanda Interactive and Netease were managing markets for their in-game items for 10 years in China before crypto was a thing anyway. You don’t need to be a crypto company to have a secondary market for rare virtual items in your game.
The GOP’s midterm performance against Biden in 2022 was the worst midterm opposition performance in my lifetime, considering the macroeconomic environment and the incumbents’ track record on inflation, crime, and Covid/schools.