Discover more from Theia Blockchain — The Monthly Spectacle
Synthetic Put on UST
How our risk principles, and the principle of fragility in particular, allowed us to pass on investing in Terra and create a synthetic put on UST.
*A note to our investors. This piece can be quite dense, so if you are already familiar with (or uninterested in) the mechanics of the Terra UST stablecoin, we would suggest reading the section titled ‘Fragility and Risk Principles’ and then jumping directly to ‘A Few Good Decisions’.
Fragility and Risk Principles
We have a single fund and that fund has a single mandate — to beat the relevant index by a wide margin after fees and taxes. That single mandate, however, is comprised of two components. We must first meet the index and only then can we beat the index. While we are paid for the second component of our mandate, we have profound respect for the difficulty of the first component. It is hard to simply meet the index over the long term. We are often surprised by how many people entered our industry when Bitcoin was under $10.00 and failed to capitalize on a 3,000x multiple of invested capital. You wouldn’t be surprised by the main culprits: leverage, over-extending during bull markets, selling during bear markets and trusting the wrong counterparties. We expect this will continue and most individuals and funds investing today will fail to meet the market return. Much of our strategy has been built on this assumption.
Our first priority is that our investors are able to benefit from ongoing growth in blockchain adoption, which means we always underwrite with a risk-first approach. We are allergic to everything that does not pass our risk filter. While our risk filter is often governed by strict rules (e.g. no portfolio leverage) and policies (e.g. a well-designed cold storage security architecture), we believe that risk policy is fundamentally the adherence to a well-defined and time-tested set of principles. I mention all of this because it was our risk principles that allowed us to avoid investing in Terra and create an asymmetric put on UST, the associated stablecoin.
The idea of fragility is at the center of our risk strategy. This is an idea that we learned from the writings of Nassim Taleb. Fragility means something breaks with volatility like a vase breaks when it falls off a table. There is an important concept embedded in fragility: the concept of the absorbing barrier. A broken vase is an absorbing barrier since a vase will not miraculously put itself back together once broken. Illness is not an absorbing barrier, since humans can heal, but death is an absorbing barrier. One more example — an incorrect scientific hypothesis (e.g. no life can grow without sunlight) is fragile and susceptible to an absorbing barrier because it can be disproven with a single disconfirming observation (e.g. the discovery of deep-sea vent bacteria). The idea is that absorbing barriers are always reached over enough time.
Terra suffered from this exact fragility. There has been plenty written on the collapse of Terra over the past week, so we will be brief and stick to the essentials. Our focus here is to take our investors through our process, sticking to our initial hypothesis which was roughly correct. We are doing our best to avoid the pernicious effects of the hindsight bias, and sticking to our original models and underwriting documents wherever possible.
Terra and UST from First Principles
Terra is a blockchain operating system like Ethereum, Solana, and Avalanche. It allowed developers to build blockchain applications and deploy smart contracts. Luna is the native token of Terra like Ether is the native token of Ethereum. Most large blockchain operating systems have a native currency, but Terra is unique in that it also had a native stablecoin called UST.
A stablecoin is a blockchain token pegged 1-to-1 with the US Dollar (‘USD’). Since blockchain tokens trade on open markets, this peg is typically achieved by collateral and an arbitrage mechanism. UST, the stablecoin related to Terra, was collateralized by the value of Luna, Terra’s native token. If UST was trading at $0.98 due to a decrease in demand for UST on the open market, then the Terra blockchain would allow arbitrage funds to trade one UST for $1.00 of Luna. This would increase the supply of Luna, diluting Luna tokenholders, and reduce the supply of UST to match demand. If UST was trading at $1.02 because of an increase in the demand for UST, then owners of the Luna token could eliminate Luna tokens equivalent to $1.00 in exchange for one UST and sell the newly minted UST on the open market for $1.02. This would increase the supply of UST to match demand and reduce the total number of Luna tokens outstanding. The mechanism itself is difficult to wrap your head around initially. The point is that the price of Luna went up when demand for UST grew and the price of Luna went down when demand for UST fell.
We like to think about financial products from first principles at Theia. There is little new under the sun, and most mechanisms imitate or slightly alter the mechanisms that have already been tried in traditional markets. This applies to Terra. Our mental model is that issuing UST is equivalent to issuing 0% debt used to buy back shares of Luna. In traditional markets, a company like Apple could easily raise $18 billion of debt at ~4.0% and use the $18 billion it raised to buy back shares of Apple in the open market. Once purchased by Apple, these shares are taken out of circulation and the price of each Apple stock rises, as the remaining shareholders each own a bigger piece of the same pie. This is the same mechanism used by Terra — the market essentially took $18 billion of Luna tokens out of circulation in exchange for $18 billion of UST, a fixed instrument with no coupon. The mechanism worked as intended, and the debt-financed Luna buyback program generated substantial price appreciation while the price of UST remained steady at $1.00.
The incredible thing is that everybody wanted UST. The amount of UST outstanding went from $2 billion to $18 billion over six months. Why was there so much demand for UST? One reason is that the Terra team was acutely focused on expansion, creating partnerships with industry leaders, and adding UST to every decentralized finance protocol on every major blockchain operating system. The more important reason is Anchor, a lending and borrowing protocol that paid 20% on UST deposits. Anchor was a deliberate attempt to subsidize growth through outsized yields. This was only possible because Anchor was funded by the Luna Guard Foundation, which was in turn funded by many of the major funds in the blockchain ecosystem. These funds had a significant interest in driving UST growth to mechanically increase Luna’s price. Everyone wanted a piece of the Anchor yield — the biggest funds in the space were giving away money — and this drove explosive growth in the supply of UST.
By now, you understand the basic mechanics of the operation:
Early Luna investors fund the Anchor 20% yield
Anchor’s 20% yield drives demand for UST
Increase in UST tokens outstanding means a decrease in the number of Luna tokens outstanding
And the price of Luna goes up *a lot*
Let’s illustrate the mechanics of this system with a thought experiment. Imagine that Apple launched a stablecoin pegged to the US dollar. We can call it iUSD. You can always sell iUSD to Apple Inc. for $1.00 worth of Apple stock, so you can always make a profit by buying iUSD at $0.98 and getting $1.00 of Apple stock that you are free to sell on the open market. When demand for iUSD is growing and the price of iUSD rises to $1.02, you can convert $1.00 of Apple stock into one iUSD. This means that the growth of iUSD translates directly into a lower supply of Apple stock (i.e. the growth of iUSD functions as a share repurchase).
Apple could easily handle $18 billion of iUSD outstanding because Apple can easily handle $18 billion of debt. What would happen in a situation where every single person who held iUSD decided it was worthless and wanted to give it back to Apple? Apple, a company trading at a market capitalization of $3.25 trillion and with a stock trading at $140 per share, would simply issue $18 billion worth of stock in exchange for $18 billion worth of iUSD. The total supply of Apple shares would increase by less than 1%. It wouldn’t be a problem because Apple is worth much more than $18 billion excluding any value given to the iUSD experiment.
The situation is different with Terra. Assume that Luna has no value outside of the UST experiment, and you will understand how this goes wrong. While Anchor is paying 20% and the demand for UST is growing, the price of Luna is only going up. You can imply a market capitalization by multiplying Luna’s artificially elevated price by the total number of tokens outstanding, and this market capitalization was at around $30 billion before the depeg of UST. The problem is what happens when the demand for UST goes in reverse. Then every person who is selling UST back to the protocol is issuing Luna tokens, and the price of Luna tokens goes down. Investors panic as they start to suspect that there may be more UST outstanding than the value of Luna can tolerate, and the price of Luna tokens goes down so much that its market capitalization decreases to $10 billion. All of the sudden, you have $18 billion of UST (i.e. debt) and $10 billion of Luna market cap (i.e. total collateral). The holders of UST rush to convert their holdings into Luna while there are still buyers (the only buyers at this point are the people who don’t fully comprehend what is going on), and the supply of Luna increases, further reducing price. Those that were able to sell UST on time can keep a part of their principal, and everyone else is wiped out. Well, this exact process is currently ongoing — the price of UST is $0.06 and the price of Luna is $0.0001835 (down from over $100.00 at quarter-end).
A Few Good Decisions
The most important and difficult thing we did was not invest in Terra (via its token, Luna). As our investors know well, we are acutely focused on the blockchain operating system (‘L1’) thesis and believe that a few winning L1s will deliver outstanding returns to tokenholders. This is a common thesis in the space, as the market power of the dominant L1s is apparent to almost anyone investing in blockchain with an eye towards enduring moats, and the difficulty lies in picking the right L1 assets.
In particular, we are enthusiastic about a subset of L1s that have been able to make progress in building a dual-sided marketplace with committed developers and consumers while still trading for less than 10% of Ethereum’s valuation. The market is pricing in a monopolistic outcome in a market we believe will be a duopoly or an oligopoly. This is why it was difficult for us not to invest in Terra. Market consensus in late 2021 and early 2022 was that Solana, Terra, and Avalanche were the most serious contenders to take real market share from Ethereum, and Terra was the most beloved of the three. Terra had the second most collateral locked in the system (a metric that measures L1 adoption) after Ethereum, and the number of wallets in the Terra ecosystem had been exploding all year.
The hindsight bias that makes Terra’s collapse seem inevitable did not shine so brightly in early 2022, and John and I were constantly asking ourselves if we were ‘missing something.’ Many leading funds in our space were vocal investors, and we were seeing at least one article per day with a title like ‘Can UST Grow to $1 Trillion?’ At the height of the siren call, it was our risk principles that kept us out of trouble. Under-collateralized by design, Terra was fragile and that was the end of the story.
We even started working on a post where we explained the merits and demerits of Terra to our investors and explained why we decided to pass on the opportunity. We felt that our investors, who had heard us sing the praises of the L1 thesis, deserved to know why they were missing out on Luna’s stellar performance (see YTD performance above). The piece was going to explain that we would rather miss a few good opportunities than risk making a large mistake.
As we started gathering data on the Terra ecosystem for this ‘pass piece’, we became convinced that UST was headed towards an eventual depeg. Everything we looked into ended up being much worse than we expected. How was Anchor funded? We knew it was partly through operations and partly through venture capital funding, but when we did the math we learned what ‘partly’ really meant. Only 10.9% of Anchor’s funding came from operations and 89.1% of its funding came from unsustainable venture capital. How many more days could Anchor pay a 20% yield without receiving additional outside funding? It was burning $7.6 million per day and had $202.7 million left in the capital reserve. Anchor could only afford to pay 20% for 30 more days. How important was Anchor to the demand for UST? $14 billion of the total $18 billion of UST supply was locked up in Anchor — imagine a single bank with 78% of the money supply.
Most importantly, how critical was Anchor to the entire Terra ecosystem? It was everything, or as close to everything as a single protocol can be. Anchor accounted for between 20 - 50% of the traffic on Terra, and many of the other large protocols depended on Anchor. Pylon Protocol, another application on Terra, would allow a consumer to make recurring payments by depositing 5.0x the annual payment amount in Anchor and redirecting your Anchor interest rewards to the merchant as payment. So if you wanted to pay $240 per year for Netflix, you simply had to deposit $1,200 in Anchor and Anchor would make your monthly $20 payment to Netflix for you. If you wanted to unsubscribe from Netflix, you could simply remove your principal and the entire experience was ‘free.’ The ecosystem was full of these protocols that were based entirely on the promise of a sustainable 20% yield coming from Anchor. Our final estimate was that at least 70% of the Terra ecosystem depended directly on Anchor’s 20% yield, and Anchor’s 20% yield was unsustainable.
How Many Days does Anchor Have Left?
Let’s go back to our mental model. Luna is equity and UST was debt yielding 20% for a brief period. That in itself is fine, as long as people know that UST is debt (they didn’t) and the amount of outstanding debt is less than the value of the collateral excluding the stablecoin mechanism. Remember that in the Apple-iUSD case, Apple’s equity can absorb an $18 billion iUSD contraction by simply diluting its equity holders. What we realized in writing our ‘pass piece’ was that the ability to issue new debt (UST) and the value of the collateral (Terra’s ecosystem) were completely based on the false assumption of the Anchor protocol, and the market would soon see that assumption vanish into thin air. It would become apparent that the value of UST outstanding was greater than the collateral value of the Luna token, and the death spiral mechanism would kick off.
This is when we decided to create a synthetic put instrument on UST (Terra’s stablecoin). The way our instrument works is that we use AAVE, a borrow-lend protocol on the Ethereum blockchain, and posted USDC (Circle’s stablecoin) as collateral. We used that collateral to borrow UST and sold the borrowed UST for USDC on Curve, a currency trading platform on the Ethereum blockchain. We were then able to post the newly purchased USDC as additional collateral and increase our borrowing of UST. We were able to run this loop three times. In the end, we had $2.50 of USDC collateral and $1.50 of borrowed UST for every $1.00 invested. The idea was that the $1.50 of borrowed UST would disappear, and we could keep the entire $2.50 of USDC. The instrument was beautiful. Our carry cost was about 0.87% per month, and we only needed to hold the position for one or two months to see our thesis play out. In keeping with our conservative natures, we had the position at a 60% loan-to-value ratio. UST, a stablecoin, would have needed to rise to $1.50 for us to need to place collateral. We don’t consider many things impossible in blockchain markets, but the probability of a stablecoin with a functional mint mechanism going to $1.50 is as close to zero as things can be. The Terra ecosystem could have become the most profitable investment in history and the outstanding supply of UST could have grown to $500 billion within six months, and UST would still not have risen above $1.10 for a brief instant. To top it all off, we executed and custodied the position in a single isolated wallet we have for decentralized finance. The position was completely separate from the rest of our portfolio.
This is a good time to explain our position on shorts and put options. Shorting volatile assets with your portfolio as collateral is immensely fragile. You may be right about every short for twenty years, but then you happen to step on a landmine by shorting GameStop and you suffer serious permanent capital loss. Our short positions need to be isolated and capped downside investments. Our UST trade could realistically only cost us the carry cost for six months (5.23%) despite being a position that could (and did) give us 10 points of overperformance relative to our index.
We also require that our shorts compete with our longs on an expected value basis. Many investors short at an expected value loss to dampen volatility. Many of our peers have their savings in Bitcoin and simultaneously have shorts on Bitcoin within their funds. We understand the desire to dampen short-term volatility, but it is not the way we invest. The synthetic put on UST was certainly attractive on the basis of expected value. We were break even from an expected value perspective if we thought UST had a 1.5% probability of losing its peg in the next two months, and we could make between 2x and 3x our investment within the same period if things played out as we expected. The synthetic UST put was certainly competitive with our long positions as a standalone investment.
Our UST synthetic put option was what we considered to be a bounded downside, probable, and asymmetric investment with a near-term catalyst. It was the type of investment we look for at Theia. In the end, we did not need to wait for Anchor to run out of funds. The narrative started changing around Terra within a week, and major publications started to mention the upcoming need to fund Anchor. A couple of days later, UST lost its peg. There are all types of narratives — speculative attacks, coordinated campaigns — it doesn’t matter to us. That’s the principle of fragility at work. If we see a delicate vase on a sailboat crossing the Atlantic, we won’t try to predict how it will break but will try to predict that it will eventually break.
In a classic bank run, Anchor lost $5 billion of deposits in a single day. The market cap of Luna went below the market cap of UST (i.e. equity went below debt) a day later and the UST peg went as low as $0.35. The protocol entered the so-called death spiral as arbitrage funds were buying UST for well below $1.00 and using it to mint $1.00 of Luna, which caused the supply of Luna to grow from 340 million to over 6.5 trillion over the next couple of days. Luna’s price fell from $119.20 as recently as April 5th to $0.0001621 today. UST went from $1.00 to $0.06 during that same period.
We exited our short position at $0.20 since we had made a sufficiently good return and did not want to risk our $0.80 ‘bird in the hand’ for an additional, higher risk $0.20. You can see our results in the table above.
It’s about Fragility
This entire episode has been unfortunate for our industry and a devastating life event for millions of people around the world. We are not excused — we did not sound any alarms and did not attempt to discuss our views publicly. Going forward, it is important that we learn the right lessons from these events.
The first lesson is that we should always come back to the first principles of microeconomics and risk management. These principles have been refined through thousands of years of humans interacting in markets. We should at least try to justify decisions on first principles, and we don’t believe this attempt was made with Terra. A stablecoin is not a new and magical invention — it is liquid debt with no coupon payment. We are willing to hold USDC, the stablecoin issued by Circle, because we appreciate its liquidity characteristics and can verify that the debt is fully collateralized. We were not willing to hold UST, regardless of its liquidity characteristics, because there were $18 billion of claims on much less than $18 billion of assets. We expect that this incident will cause investors to minimize risk when it comes to their stablecoin holdings. If you need to hold debt but won’t receive a coupon, then you should move to the lowest-risk debt.
The second lesson is about fragility. We believe this lesson is currently being discussed at the wrong level of abstraction. The consensus opinion is that ‘Terra taught us that we can’t have algorithmic stablecoins.’ We think the right lesson is that we should not create fragile assets in volatile markets.