The Principles and Nature of Our Firm

I’d like to share some of the lessons I’ve learned about business and startups in the crypto space over the past 7+ years and how we’ve applied these lessons to our firm. In the second half of this post, I describe some of the pressing challenges facing the space today.

Goal-Oriented VS Process-Oriented Behavior

It is imperative that a startup creates a culture of pursuing goals rather than following processes. While we may sometimes fail to achieve our goals, this is preferable to the illusion of productivity that often comes from process-oriented work. Processes are only tolerated at scale when labor coordination becomes too difficult without them, not when taking a company from Zero to One.

Our Goal

The primary goal of our hedge fund, Galois Capital, is to generate alpha for our investors through trading in the crypto space.

The Labor Theory of Value is False

The world does not reward you for how much or how hard you toil, only for the value you provide it. Solving complex problems provides a competitive moat but only if the world actually wants those solutions.

Hedge Funds and Scale

There are two ways hedge funds make money: 1) generate trading profits and take a cut or 2) raise tons of money and charge management fees. Focusing on the first creates a “trading hedge fund”. Focusing on the second creates a “sales hedge fund”. Galois Capital is a trading hedge fund.

A sales hedge fund has greatly misaligned incentives with its investors. Generating returns for their investors becomes secondary to increasing the total assets under management (AUM). Scaling AUM matters more than anything. A trading hedge fund only eats when their investors eat and hunger is the best sauce. Here, scaling AUM matters less.

Moreover, hedge fund startups and traditional tech startups are different in a key way. A tech startup often tries to achieve a monopoly by going deep into the red, burning money to eventually win the entire market. This is usually because of network effects and economies of scale. Hedge funds are different because trading returns on capital have diseconomies of scale. It is easier to turn $1m into $10m than it is to turn $1b into $10b. Trading strategies are capacity-constrained and markets are liquidity-constrained (especially in crypto markets). Because of this and the fact that early cashflows are often black rather than red, there is less pressure for hedge funds to grow into giant, monolithic firms. Still, for operational efficiency and due to labor specialization, hedge funds need to achieve at least some modest scale.

The Startup and Early Employee Covenant

An early employee at a startup often takes a pay cut, shoulders greater risk, and works harder than if he/she had a cushier job at a larger firm. The implicit promise of the startup is that it will have been worth it. Since early employees are issued significant equity in substitute for salary, there is a lower bound to which the firm must scale for that promise to be realized. For example, a sole founder plus 4 salaried employees in a firm could be a good lifestyle business as it is, but a sole founder plus 4 equity-heavy employees might need to grow it to a 25 person firm before it is worth it for the initial 4 equity-heavy employees. Effectively, the earlier a firm starts compensating their new employees with mostly cash, the smaller the firm can be without violating this covenant. However, there are serious tradeoffs to consider here.

Higher variable compensation tied to the success of the firm aligns employee incentives and reduces the need for “management”. Lower variable compensation allows for greater value capture by founders/investors and reduces the pressure to scale. At Galois Capital, all six members of the founding team including both founders take no base salary. At this stage, we believe alignment is the most important.

Learnings from the Street

Why do SIG and other successful prop shops and hedge funds like to hire excellent poker, chess, go, and video game players? Why does Renaissance Technologies hire so many PhDs in math, physics, astrophysics, and computer science as well as code breakers? Why is Jane Street’s interview filled with difficult brain teasers that test a candidate’s ability to think about expected value and risk? Why is Five Rings Capital named after The Book of Five Rings? At Galois Capital, let’s not fix what ain’t broke.

Learnings from the Valley

Why is Peter Thiel a Christian? Why does Ben Horowitz like gangster rap so much? Did Paul Graham’s work in Lisp help frame what he looked for in YCfounders? Why did Elon Musk name his company after Nikola Tesla? Is Valve’s culture responsible for why they are one of the most profitable companies per employee? At Galois Capital, we will not do as they did; we will seek what they sought.

The Necessity of Failed Experiments

Often the market demands a certain class of experiments that are doomed for failure. Whether it is incumbent institutions, burying their heads in the sand, demanding “blockchain not Bitcoin” as a way to co-opt a new technology which could potentially eat parts of their legacy business or greedy investors demanding higher and higher risk tokens born of regretting having missed out on Bitcoin, it was inevitable that the market produced these projects. The market may not be “rational” but it certainly is “meta-rational”.

Reflexivity in Crypto Markets

Fred Wilson published his landmark crypto piece “Fat Protocols” in 2016. While I agree with him on his assessment, this piece may have caused a misallocation of capital into the space. Suppose protocols in crypto/blockchain really do account for 90% of the value whereas applications, 10% of the value. Because most people agree that this is the case, capital starts pouring into protocols. To the point at which 95% of capital goes into protocols, applications actually become undercapitalized and undervalued. At that point, it becomes profitably contrarian to actually invest in applications.

Social Capital Subsidizes Financial Capital

If the market, indeed, clears, an industry which is unseemly or in some other way generative of negative social capital will produce excess financial profits. If we lived in a society where porn or weed had no socially negative connotations, the premia in their profitability would decrease. In the same way, any industry which generates, by association, positive social capital will be over-valued in real capital terms. So Bitcoin will reach long-term fair market price when it is absolved of negative social capital. In the short-term, the various crypto bubbles have shown us that price crests and troughs are directly correlated to how fashionable it is to be associated with the space.

Problems of Incentives in Crypto Custody

Crypto custody is a hot topic these days as the promise of institutional capital entering the space is enticing. Large institutions are unable to create these solutions for themselves not just because they lack the expertise but because the engineers at the bottom of the totem pole who have to implement these solutions will never see the full upside of their work. A team of engineers could successfully build a custody solution that holds billions in crypto and get a sizable bonus but still this would be a rather small amount relative to the value created. If they fail and there is a security breach, they lose their jobs and are blackballed for the rest of the their career. The CEO, at the top, is also in a difficult spot. If the custody solution works, shareholders are happy and the board gives him a big bonus. If it fails and they lose billions, there is nothing the CEO can do to explain why he risked the rest of a perfectly good and profitable business for such a crazy new venture. No one wants to take the fall for a huge mistake that is irreversible, which is why crypto is unlike anything else. The stakes are high and mistakes are permanent.

Should the custody solution require insurance, who would be willing to underwrite it at a reasonable price? The insurance company will certainly have less expertise than the custody company in what it is doing and cannot accurately evaluate the risks. Due to this asymmetry of both expertise and information, the insurance company is forced to price the insurance with greater premiums. To the point at which these premiums are too high, that money would have been better spent actually making the custody solution better.

In the end, the Lindy Effect is a good heuristic. Custody solutions created by the longest-standing un-hacked crypto institutions will be the best, not those created by legacy institutions in finance or tech. Long standing exchanges flush with cash from 2017 should offer custody products and create their own insurance pools for future potential breaches.

Crypto Funds

There are some practices in this space which make me uncomfortable. Many VC-style long-only crypto hedge funds hold an inordinate amount of illiquid tokens. The mark-to-market value of these holdings will be completely different from their mark-to-realized value when investors redemptions force these funds to sell. Liquidity doesn’t matter until it does and then it’s the only thing that matters. If a fund holds 20% of the circulating supply of a relatively illiquid token (say rank 50–100 on CMC), the order books may not be thick enough to support exiting the position without incurring massive slippage.

How about truly illiquid assets held in side-pockets? Many funds hold these assets at cost until there is price discovery from the listing of that asset on an exchange. But if all crypto assets are correlated with Bitcoin and Bitcoin falls 50%, shouldn’t the side-pocket be marked down by at least 50%? I say at least because token beta to Bitcoin is usually greater than 1.

The many funds which started in 2017 are now, in sequence, having their 1 year lockup expire. As investors redeem, there could be a negative cascade effect on the market. Say Alice Fund holds 10% of the circulating supply or float of Token X and Bob Fund holds Token X (5% of the float) and Token Y (10% of the float) and Carol Fund holds Token Y (5% of the float) and Token Z (10% of the float). Redemptions from Alice Fund will pull down the price of Token X, hurting the mark-to-market value of Bob Fund, causing redemptions from Bob Fund which cascades into redemptions from Carol Fund and so on and so forth. In my opinion, the effect will not be that bad until 2019Q1 because most of the 2017 funds have generated huge returns that year so LPs are likely not too unhappy overall even with the drop in 2018. But the 2018 funds are a different story. If this bear market continues, when 2019 rolls around and the 1 year lockup expires for the 2018 funds, we can expect huge redemptions. It would not be surprising to see BTC dominance continue upwards through the early part of 2019.

Kevin |