Credora: What We Saw (Part 2 of 3)
FTX Disaster & Contagion
The recent FTX and contagion events are devastating for the industry. Multiple Credora borrowers have been impacted. We have a huge amount of respect for how many of them are navigating the challenge, and we have been able to see in real-time their prudent risk management ensuring ongoing creditworthiness. A timeline of key events is included below:
- 11/2: Alameda balance sheet headline numbers are published by CoinDesk.
- 11/6: Binance signals sale of FTT holdings.
- 11/8: FTX processing significant withdrawals ~$6bn.
- 11/8: FTT price declines from ~$22 reaching intraday low of ~$3.
- 11/8: SBF announces LOI with Binance, confirming market fears of a material problem in the ability to process further withdrawals.
- 11/8: FTX halts client withdrawals.
- 11/8: FTX liquid asset gap is revealed, initially ~$6 billion (currently ~$9 billion)
- 11/10: BlockFi freezes customer withdrawals.
- 11/11: FTX files for bankruptcy. APIs are deprecated, and FTX experiences a hack of ~$620 million.
- 11/13: Crypto.com and other exchanges face increased scrutiny, leading to large capital withdrawals and the acceleration by many towards proof of reserves.
- 11/16: Genesis freezes customer withdrawals for the lending business, resulting in a freeze of customer withdrawals on Gemini Earn.
We continue to monitor in real-time the rapid deleveraging of the system, and like many others, we are digesting the real implications for the industry and credit market specifically. Trading firms had already undergone a significant deleveraging following the Terra and 3AC incidents earlier this year. According to our data, average leverage across a selection of borrowers is ~25% of what it was in Q1. In the past few months, although lending appetite resumed post 3AC, limited volume and trading opportunities restrained borrowers from incremental borrowing.
As for the current market risks we remain vigilant to:
- Borrowers who had a significant amount of capital on FTX
- CeFi lenders who have credit exposure to Alameda and FTX, or other financially impaired borrowers
- Prime Brokers who had a significant amount of capital on FTX, in their own control
- Other possible exchange ‘bad actors’ (the market is swiftly pressing those who have built less trust)
One interesting point regarding #2–4. Our platform primarily services trading firms who are net credit borrowers. The reality of the market is that there is a higher expectation for due diligence placed on trading firms as borrowers, versus their lending firm, prime broker, and exchange counterparts. Ultimately, any capital allocator to a lending firm or depositor on an exchange is taking credit risk to that institution. In many instances, there was ZERO due diligence done on these companies’ financials and operational policies (even from a simple financial reporting perspective).
Credora has historically attempted to work with lenders to validate their financial position. From a real-time data perspective, it is a more challenging exercise as the majority of lending occurs OTC, and therefore there are no third-party sources of truth about where those funds are going. Regardless, the main blocker was that their positioning in the market allowed them to operate without transparency. They were empowered to require information from their borrowers, while revealing little information about their own financial position. This is a double standard that we do not expect the industry to tolerate further.
During these events, Credora made three modifications to the methodology.
- 11/8: FTX exchange equity discounted by 50% on balance sheet factors. This includes equity, leverage, and current ratio of a trading firm. Amidst uncertainty, the choice was made to not yet realize the loss of capital on FTX.
- 11/9: The 50% FTX exchange equity discount was also applied to income statement factors. Although challenging to quantify, there were indications that the market believed the price of recovery to be in the ~50% range.
- 11/10: FTX exchange equity was discounted by 100% across all borrowers. Probability of any timely recovery was deemed negligible.
Borrower scores and ratings were impacted according to the relative exposure to FTX. Some scores more materially declining versus others depending on the strength of factors before the incident. For example, the shape of scoring curves more negatively impacts borrowers who already had relatively weaker metrics (i.e. the change from 2.0 x to 1.5x current ratio results in a smaller point decline than the change from 1.5x to 1.0x).
The below visuals display the changes in scores over the period, across borrowers that had published scores externally. Following 11/10, there were no remaining AA borrowers, and multiple borrower ratings declined from A to BB.
During the first week of November, there was approximately ~$200m of unsecured exposure across Credora partner protocols. Of the outstanding exposure, 98% was already repaid and there are thus far no defaults across the borrower group. We continue monitoring remaining exposures, keeping partners informed as appropriate.
Our Maple pool recalled outstanding open term loans on November 8th and was able to get fully repaid in less than a week. The pool has no outstanding liabilities, and will aim to re-deploy capital after the dust has settled. As a pool delegate, we operate a liquid and transparent pool, and although we have a unique advantage in understanding who remains creditworthy, we elected to recall all outstanding loans as we expected the following:
- Material re-pricing of credit risk
- A swift drain of liquidity from the borrowers
- Demand for capital from lenders
- Difficulty quantifying second order impacts
We have historically described our platform as having a reverse lemon ‘problem’. Only transparent borrowers are willing to complete our processes, and so if you generally believe more transparent borrowers are better borrowers, then this is a good problem to have. Furthermore, borrowers do not publish their score unless it reflects positively on them. As a result, it appears externally as though our methodology scores all borrowers highly and uniformly, but in reality, many borrowers have started our processes and elected to attempt to borrow without publishing. The requirements for information validation and real-time transparency are more than they are willing to provide, and the scoring reflects this (Alameda never published a score). The reality can be visualized by the below graphic.
Our processes in evaluating borrowers are incredibly transparent. We deliver a breakdown of the scoring, and how they can improve the score. Typically, score improvements can be achieved through the following:
- More or higher quality financial information
- More real-time data validation
- Improvement of a particular metric (i.e. performance, interest coverage)
- More of a borrowing history
The final part of our three-part blog series, Where We Are Going, will be released on Monday.
We are actively looking to expand our credit and development teams. If you are data and model oriented, and believe that credit risk can be accurately quantified and assessed in real-time, we want to hear from you!
Additionally, we invite industry capital allocators to reach out and discuss our methodologies.