Tristero's FTX Report

This memo was originally circulated to Tristero stakeholders on November 10th, 2022. We made the decision to release it publicly due to role that current discussion around FTX will play in shaping the future of the sector.

Hello all, 

We're taken aback by the events of the past week. While we were confident that Mt. Gox wouldn't be the last major centralized exchange failure, FTX had an excellent reputation and publicly advocated for more stringent regulation. Ultimately, this highlights fundamental issues with centralized exchanges and the role of trust within the financial system at large. We will attempt to 1) summarize what happened 2) present possible interpretations 3) analyze the potential impact. Everything here is our current opinions based on imperfect information and we're almost certainly very wrong on at least a few of the details. We prioritized speed and rigor in turning this around so ask that you forgive any typos. 

Was FTX a good business?

FTX is not the first exchange to stumble into trouble. The most famous example is arguably Mt. Gox, once the largest bitcoin exchange in the world before collapsing in 2014 and losing some 750,000 BTC in the process. In 2016, $72 million of funds were stolen from Bitfinex, causing the exchange to temporarily suspend operations. In 2018, under extremely suspicious circumstances, Canada's largest cryptocurrency exchange, QuadrigaCX, became illiquid after the private keys for $145m of user funds became inaccessible. The present circumstances surrounding FTX portend a scale of losses an order of magnitude larger than every previous exchange blowup combined. 

FTX was viewed as one of the most trusted exchanges in crypto and, unlike many crypto companies, it was wildly profitable. Assuming the books were not fraudulent, in 2022, FTX made an average of $1.5m dollars in profit every single day. At the end of Q2 2022, FTX had almost 6m registered users and averaged 1.7m weekly active users. Growth of users, usage, revenue and EBITA were all remarkably strong. FTX had grown its trade volumes more than 600% since the beginning of 2020, easily making it the fastest growing major exchange. Both its technology and spreads were viewed as best in class by many engineers and traders. 

In addition to being extraordinarily profitable, FTX was also well capitalized. It had raised $8m in its August of 2019 seed round, $1b during its initial July 2021 Series B, a further $420m during its secondary B round in October of the same year, and $400m during its January 2022 Series C, peaking at a valuation of $32b. The total amount of funds raised were in excess of $1.8b.

Relationship to Alameda

Since founding, FTX has maintained a close relationship with Alameda. Alameda is a prop trading shop that, over time, evolved to operate more like a holding company. During the early days of FTX, Alameda was responsible for just under 50% of their total volume, likely making the majority of FTX’s markets, but by the summer of 2022 Alameda comprised a far more moderate 3% of total volume. 

It is crucial to note that there exists almost no barrier between the two firms. Alameda and FTX have overlapping ownership and even share offices. We have heard credible reports of people working for FTX in the morning and Alameda in the afternoon. We suspect that Alameda experienced a large drawdown in May 2022. As a reminder, the LUNA implosion led to nominal losses of at least $55b and forced funds who had taken levered positions in the crypto markets like 3AC into bankruptcy.

The FTT Token

In May 2019, FTX launched the FTT token. The token was created to give users a way to get discounts on FTX fees, and FTT’s price was supported because FTX committed to spending 25% of revenue to buy back the token. Naturally, FTX and Binnance both ended up with large holdings of the FTT token.

It was common practice to borrow against exchange tokens. We believe that most major exchanges did this in some capacity. Because exchange tokens usually represent a percentage of future exchange revenue, they have been viewed as acceptable collateral. Similarly to the Archegos scandal, the lack of visibility into the net debt load of the borrower combined with the ratio of the float to leverage creates the potential of a false perception of soundness.

Around the time of the collapse of 3AC and Luna, there were large transfers of FTT from FTX to Alameda. We believe that this was to cover losses so Alameda was able to continue to function. Much like investment banks during 2008, Alameda’s ability to do business depended heavily on their perceived credibility. Our understanding is that vesting of FTT may have been accelerated to give FTX access to additional capital. If this is the case, it potentially would have resulted in Binance being given a large, liquid FTT position.

What triggered FTX’s insolvency?

Alameda's balance sheet was leaked, showing a risk of illiquidity if the FTT token dropped in price.  Subsequently, unusual activity was detected in their hot wallets.  Binance announced their intent to liquidate all of their 23m FTT (worth $552,000,000 on Nov. 6). FTT fell 90 percent over the subsequent 2 days. This was potentially triggered by Genesis liquidating FTT collateral after the price fell below a prespecified threshold. 

How big is the deficit? 

There is no credible, public number around the size of the deficit. This is likely due to challenges in determining future liquidity requirements as well as fair market value for illiquid balance sheet assets. 

The best, but unlikely, case is that the debt load is around 2 billion. The Wall Street Journal has reported that Alameda owes FTX 10 billion which is hard to reconcile with Alameda's alleged balance sheet. It is worth noting that Alameda's debt load as represented in the leaked balance sheet was 8 billion which in theory sets a hard cap on the size of the hole. If there is off balance sheet debt (i.e. power barges) it indicates a high likelihood of criminal wrongdoing and also draws into question all other figures conveyed by the FTX/Alameda team to date.

How is insolvency possible?

What we know for certain at this stage is that FTX has a large amount of bad debt. The origins of this debt have two possible sources. Either the debt was created internally via Alameda or the debt was generated through the assumption of external liabilities. 

In the scenario where it was generated internally, billions were lost on Alameda's side and these "bad assets" were transferred to FTX. One way this could have happened is that Alameda loaned billions from FTX secured against Alameda’s FTT tokens and was unable to repay it. FTX would now have a balance sheet that is only liquid so long as the FTT token remains above a certain price. Many believe that such a multibillion dollar loan from FTX to Alameda was implicitly created because FTX failed to liquidate Alameda. Alameda may have had a special account which was not bound by the cross-margining engine and therefore was never liquidated, leading to ballooning debt. 

The assumption of external liabilities may have been less than voluntary. If Alamada did have a large amount of debt against FTT, any large sum being sold on the market could have put the firm underwater. This may have contributed to the bailouts in which FTX/Alameda participated. 

The line between touching and not touching customer deposits can get blurry at an exchange that allows margin trading. Anytime someone is allowed to borrow, someone else is assuming that liability. There is always a risk that the exchange cannot liquidate the collateral efficiently enough to cover the loan. One can argue that it’s not possible to run an exchange that allows margining that will also never lose money on certain liquidations. Such ambiguity introduces a degree of moral hazard that we believe played a critical role in the implosion. There is reason to think that FTX’s cross-margining engine would have been able to cover loans through liquidation if applied. Further, because Alameda was simply a user on the platform, their claims are not junior to other users.

Regardless of if the debt was internal or external, had FTX simply let Alameda implode, or even properly fenced the risk between the two entities, we think it unlikely that FTX could have gotten into a position where insolvency was even possible.

How severe was the fraud?

We think there are three options:

1) FTX was always fraudulent. It is possible that every asset deposited was booked as revenue. Mechanically, all of these funds would have been lent to Alameda on receipt, who would have liquidated upon and effectively bet on their ability to beat FTX's clients to pay back later withdrawals. This would  have been illegal due to the shared control of both firms. It is unclear how this could have happened in the context of the caliber of FTX's investors. In theory, the lending of assets alone may not pose a risk if collateral requirements are sufficient. Lending client assets has a very similar risk profile as managing a margin engine effectively and there are regulated parallels in the American equities market. Obviously, it is highly unlikely that risk is well managed if the assets are lent to oneself. 

2) FTX crossed the line in response to the Luna-3AC crisis. FTX may have chosen to access client funds in response to the liquidity crunch in the spring. In this scenario, they simply prolonged their eventual failure through the use of client funds. In many ways, this option is the most likely and is squarely between 1 and 3. 

3) FTX built up a large amount of debt against FTT and was effectively short-squeezed. In this case, FTT would have been used as collateral for Alameda on FTX and a sudden drop in the price of FTT would have lead to Alameda becoming insolvent. The exchange withdrawals may have been paused because depositors are junior to other lenders. Allowing withdrawals would have been in breach of contract. If Alameda was using FTT as collateral on FTX, regardless of whether FTX chose to liquidate, a precipitous drop would lead to a hole emerging. Liquidating without demand and with the loan underwater would have further undermined the assets to liability ratio. 

If vesting of FTT was accelerated in the spring to keep Alameda solvent, there may have been other parties whose stakes also became liquid. In theory, if there was a large loan at a lender like Genesis, a small decline in price could force the lender to liquidate a very large position. In the past, Genesis was less aggressive around liquidating and ended up losing large amounts of money. This third scenario demonstrates egregious risk management and blatant conflict of interest but meaningfully less criminal intent as, in theory, a similar situation could have emerged without the conflict of interest. 

While the amount of enterprise value that ever existed obviously varies widely in all three of these scenarios, the lack of a Chinese wall between and joint ownership of Alameda and FTX crossed lines. If there is large amounts of debt not on the balance sheets of FTX or Alameda, it would almost certainly imply accounting fraud. 

How does this impact the sector?

There is a chance that this leads to $100b of value ceasing to exist. This would force a re-accounting by many investment firms and likely impact strategies going forward. This will accelerate the current liquidity crisis as many firms had much of their liquid portfolio on FTX. Any remaining lenders will be much more aggressive around margin calls due to their experiences so far. Given that this hole was at least exacerbated by relaxed margin requirements, this is rational and to be expected. 

Longer term, this will drive decentralization forward. This will lead to greater discipline among market actors and higher standards for managing credit. The risks that got the industry to this point are the same ones we are seeking to resolve at Tristero. Furthermore, many viewed FTX s participation in the space as mercenary. Many perceive their lobbying efforts as being only in the firm's interest rather than the sectors. To cite a version of our deck, "Centralized exchanges represent the worst of crypto and the worst of tradfi.” 

What does this mean for Tristero?

On a long timeline, this is good for Tristero. Unfortunately, these developments are not inconsistent with our perspective on the sector. People only tend to value decentralization because of crises like these;  airbags are only appreciated after accidents. In the medium term, the story is more complicated. Crypto, like finance, runs on credit being extended between firms. As noted above, Genesis liquidating FTT possibly played a large role in this collapse. The amount of credit extended between firms will likely continue to shrink for at least several months. We believe that these developments could suppress non retail trading volume for some time. The more large blocks are being moved, the more value we can create and the more revenue we can generate. 

At its core, Hourglass relies on instantaneous, direct atomic swaps, which represents an improvement on the pool model from a risk standpoint. Even in the case of total smart contract failure, the funds lost would be a small fraction of the daily traded volume. We believe that the market conditions are ripe for a platform like Hourglass that offers low slippage, capital efficiency, and a lack of counterparty risk. In the long term, we believe that Hourglass is well positioned to play a central role in the future of liquidity as the industry moves towards a more risk-averse and transparent culture.

Appendix: Timeline of Events

November 2, 2022

07:44 PST: Coindesk reports, based on a leaked balance sheet from Alameda Research, that most of FTX’s reserves are based in FTT.

November 6, 2022

06:32 PST Caroline Ellison of Alameda tweets clarifications around Alameda's balance sheet. 

07:47 PST: Zhao of Binance tweets that his company plans to sell all FTT holdings, which date back to an early investment, and compares the FTT token to the failed Luna. Bankman-Fried spars with Zhao in response.

12:47 PST: FTX employee with Telegram handle @gregftx notifies the FTX public chat that non-fiat withdrawals have been suspended.

November 8, 2022

08:09 PST:  Zhao tweets that “FTX asked for [Binance’s] help” and announces that Binance has signed a non-binding LOI to acquire the company.

November 9, 2022

09:22 PST: Bloomberg announces an SEC investigation into FTX and Binance’s acquisition attempt.

13:00 PST: Binance corporate account tweets that, due to financial issues unearthed during the due diligence process, it would no longer pursue acquisition of FTX.

November 10, 2022

06:13 PST: Bankman-Fried announces that Alameda Research is winding down its trading.

10:35 PST: FTX.US warns users that trading may be halted in a few days.

11:08 PST FTX tweets that limited withdrawals are permitted again.

18:45 PST: Reuters reports that FTX is seeking $9.4b in rescue funds from investors. Liquidity issues abound as a deluge of FTX customers withdraw their fiat holdings.