There are three observations I wish to make on the crypto exchange, FTX, and its bankruptcy, likely fraud and asset trading. The observations are around:
Trust
Proprietary trading aka prop trading
Limits of «expected value » and St Petersburg paradox
Codas: why? What falls in status, crazy balance sheets.
Background articles: Matt Levine (Bloomberg, maths on liquidity/solvency problems), Anna Massa (Bloomberg, possible fraud problems similarities to Enron); Atlantic profile, also highlighting the tainting of Effective Altruism; and the WSJ on claims of fraud.
Here is Zvi on the summary:
SBF seeks to make a lot of money to earn to give, founds Alameda.
First project is the famous Japanese BTC price arbitrage, worth 8 figures.
Business is not well-run, cofounders leave because of it, SBF cuts them out.
SBF founds FTX, things initially go well, money is made. FTX goes big: big investment at up to $32bb valuation, big spending, big advertising, FTX foundation, etc.
Alameda transforms over time from a prop trading firm that focuses on market making to a combination of longer term illiquid investments and directional bets, effectively taking super risky large long positions and borrowing short to invest long. In general the firm was structurally permanently long and unhedged.
Some bets go well, others go much less well.
In wake of Terra-Luna Alameda’s loans get recalled, a lot of money is lost and various others need bailouts or there is risk of collapse of FTT, and there is opportunity to get more funds into FTX ecosystem.
SBF decides to use his backdoor to send customer funds to Alameda, (or to allow Alameda to spend customer funds they have lying around for some reason)…
People discovered FTX and Alameda might be insolvent.
SBF and company lie over and over, saying at each point, no, everything is fine.
Everyone sold FTT and withdraw their funds from FTX.
FTT price crashes and FTX runs out of liquid funds.
FTX tries to sell itself to Binance but things are so bad Binance says no. There is no one willing to bail them out, the holes and legal risks are too big.
It is discovered that FTX and Alameda are indeed insolvent, and that FTX sent $10 billion in customer funds to Alameda, and that SBF had a backdoor allowing him to silently transfer funds.
FTX and Alameda file for bankruptcy.
FTX gets hacked, likely from inside. It gets worse.
(Here is more from Matt Levine explaining quite how crazy this blance sheet is and what it represents.)
“…If you try to calculate the equity of a balance sheet with an entry for HIDDEN POORLY INTERNALLY LABELED ACCOUNT, Microsoft Clippy will appear before you in the flesh, bloodshot and staggering, with a knife in his little paper-clip hand, saying “just what do you think you’re doing Dave?” You cannot apply ordinary arithmetic to numbers in a cell labeled “HIDDEN POORLY INTERNALLY LABELED ACCOUNT.” The result of adding or subtracting those numbers with ordinary numbers is not a number; it is prison. “ (Matt Levine)
There is much around finance - and indeed life - that is built around trust. In my podcast with economic historian Mark Koyama, he suggests that the ability to create trust at scale was an important factor for how rich nations developed. (Transcript/link here)
Some scholars believe that the ability of the Quaker people to be established as trust worthy enabled them to do better business.
The importance of trust is so ubiquitous in life today that it often goes without saying. Yet, you can observe nations or societies with more limited trust and observe the problems there.
Trust or the lack of trust still impacts financial firms. If people do not trust a bank to have their money they will withdraw the money and possibly cause a bank run.
Why would people want to break with trust when social norms values trust so highly? I can think of examples but this example around FTX, if I might speculate, is intriguing and in part it rests on the problems of expected value theory. And in part the challenges that trading have always brought.
In the history of financial institutions there are many examples of highly intelligent teams or organisation who lose a lot of money, occasionally everything and more. These teams are often making trading bets or running proprietary trading aka prop trading or have trading strategies in what we call hedge funds. The list is as long as financial history with perhaps in more recent times the notable example being LTCM, Long-Term capital management.
Trading teams have different types of strategies. Many of them at their core have some concept of expected value. The idea of (probability of event happening) x (value) = expected value. This either by computer algorithms or by human judgement on computer outputs. (They also try and arbitrage, which is less reliant on core expected value formulas.)
For rules based games with closed probabilities such as standard backgammon then over time these rules work well.
But, when you enter the messy world of humans and human values then this concept can break down. The philosopher Larry Temkin shows how it can break down at the level of moral philosophy. See my podcast on this.
Decision theorists and economists can show the problems of the St Peterberg’s Paradox. The stock market regularly highlights challenging events such as “meme stock” price movements. (These don’t fit into traditional expected value calculations).
Expected value can give a good estimation of a certain type of value in many situations but it can often not capture the whole of a value or a decision.
The St Petersburg paradox offers you the choice of 51% chance of some large gain, but a 49% chance of losing everything. For instance, you could double the value of the world or even 10x the value with a 51% chance or a 49% chance everyone in the world is wiped out.
Tyler Cowen gave the CEO of FTX, SBF (Sam Bankman-Fried) the St Petersburg choice. SBF said he would take it. Certain utilitarians and gamers will do so as the expected value is positive. But the paradox is if you are the choice enough times you will after some time inevitably go to zero.
Cowen: …let’s say there’s a game: 51 percent, you double the Earth out somewhere else; 49 percent, it all disappears. Would you play that game? And would you keep on playing that, double or nothing?
BANKMAN-FRIED: With one caveat. Let me give the caveat first, just to be a party pooper, which is, I’m assuming these are noninteracting universes. Is that right? Because to the extent they’re in the same universe, then maybe duplicating doesn’t actually double the value because maybe they would have colonized the other one anyway, eventually.
COWEN: But holding all that constant, you’re actually getting two Earths, but you’re risking a 49 percent chance of it all disappearing.
BANKMAN-FRIED: Again, I feel compelled to say caveats here, like, “How do you really know that’s what’s happening?” Blah, blah, blah, whatever. But that aside, take the pure hypothetical.
COWEN: Then you keep on playing the game. So, what’s the chance we’re left with anything? Don’t I just St. Petersburg paradox you into nonexistence?
BANKMAN-FRIED: Well, not necessarily. Maybe you St. Petersburg paradox into an enormously valuable existence. That’s the other option.
(The whole conversation on crypto and philosophy with Tyler and SBF’s conversation on Oddlots where he suggests much of crypto is a scam now make super fascinating reading)
Many prop trading teams eventually lose a lot of money. These observation has lead many mainstream financial institutions - usually investment banks to close down their prop trading teams or run them pretty small such that any blow up is not a critical blow. Prop trading can make a lot of money for a long time, but it seems at some point either the 49% catches up or a mistake is made. The successful people know when to stop or are lucky enough to stop, but getting out of the game at an appropriate time is, from what I can tell, quite rare. Investment banks have or are meant to have a lot of over sight so they can close down prop traders or cut off trading clients before losses become too big to handle.
Trading teams have many methods of risk control to stop them selves blowing up. Stop losses, limited bets size, diversifying bets, quick control of trading positions and the like, but ultimately a miscalculation or error or an event outside the model or traders experience occurs and large losses occur. In theory perhaps this is preventable but in the messy world of humans and imperfect models this preventention does not occur.
FTX was very close to a prop trading organisation, Almeda. SBF founded Almeda before FTX and had close relations with the traders there.
(This is my speculation) it seems like that Almeda had some large trade(s) go wrong. If Almeda was part of an investment bank, the savvy bank risk control would close Almeda down before losses could balloon. Even if there were no recent losses, at some point Almeda would make losses and if the traders could not control the losses and positions then the correct approach would be to let Almeda fold and not allow Almeda losses to infect any other part of the system.
(This is akin to the ideas that Nassim Taleb writes about in having a system which has lots of small pieces and is anti-fragile rather than a system which has parts which are “too big to fail” or is too centralised)
It is not exactly clear but WSJ and other media reports suggests that to cover Almeda for some reason (possibly bad trades, possibly mis-labelled accounts, possibly some inter-accounting reason; it’s possible Almeda did something even more mysterious that wasn’t simply trading, there is certainly a very complex org chart) FTX used client funds (or, possibly even just FTX or SBF funds) and when there was a rumour something was amiss and clients all wanted their money/crypto back, FTX did not have the funds to give back.
It’s unclear if this is what happened, FTX and, or, Alameda seems to have used its own token - FTT - to support either loans or trading activities which may have complicated matters as well.
In any case, on reflection, a person committed to the 51% bet for trading would inevitably run into trouble. It’s also seems to be the case that a strong unthinking commitment to only expected value models will also lead you to come unstuck. (Interestingly, SBF seems to be a keen gamer where expected value really does generally work)
Putting that all together the lessons I take from this are:
Trading teams inevitably fail. How you manage this is important. Much of Nassim Taleb’s speak to this. Beware ruin risks.
Trust. Social capital is very valuable. Destroying has many negative ramifications.
Sophisticated people forgo standard due diligence in the presence of strong charisma (this has endless examples). Classic red flags on a check list are there for a reason.
Do not trust a trading team that is closely intertwined with a client fund entity. Flags were raised on Alameda and it was obvious SBF was very close. Why exactly SBF decided to cover Alameda with client funds (if that seems to have happened) we might never know, but it seems like a bet (maybe a 51% bet?) gone wrong.
The future… the number one crypto exchange Binance does not have a clean past, much of crypto remains full of scammers, Effective Altruism is going to be tainted perhaps to some degree permanently and the idea of “super earners” earning to give away money (basically a billionaire earn to give strategy) is going to (perhaps rightly) going to take a major ding.
Coda:
Work on ergodicity is one way of thinking that perhaps overcomes some problems with expected value paradoxes. See the work of Ole Peters.
Larry Temkin’s work questions the foundational axioms behind expected value for moral reasoning and social choice. The mainbook is quite complex (Re-thinking the Good).
Tyler Cowen argues that this is not terminal for EA and “Intellectual movements can be quite influential on small sums of money.”
For financial comparisons this is a mix of MF Global (mis use of client funds; not speration of client funds) and Enron (fraud, complex subsidaries), with maybe a tiny touch of Lehman (bank run).
Coda 2:
Ben Thompson notes:
There are a few different ways to interpret Sam Bankman-Fried’s political activism:
That he believed in the causes he supported sincerely and made a mistake with his business.
That he supported the causes cynically as a way to curry favor and hide his fraud.
That he believed he was some sort of savior gripped with an ends-justify-the-means mindset that led him to believe fraud was actually the right course of action.
And… (the conspiracy theories that he was actually some sort of secret agent sent to destroy crypto, a close cousin of the conspiracy theory that Musk’s goal is to actually destroy Twitter; I mean, you can make a case for both!)
(Me: Thompson makes a very intriguing point about “folllowing the product” not the narrative. And points to AI being de-centralising and crypto being centralising)
I note that Micahel Lewis was around SBF in the last 6 months. Movie coming soon….
Tyler Cowen notes:
Rises
Common-sense morality
Common-sense investing rules
American corporate governance
Boards, and nervousness about related-party transactions
Coinbase
Seen as stodgy and bloated for much of the past year. But run in the US, listed in the US, and properly segregating customer funds.
Elon Musk’s ability to judge character
Vitalik and Ethereum
Circle, Kraken, and Binance
Anthony Trollope, Herman Melville, and the 19th century novel. Books more generally.
U.S. regulation of domestic exchanges – it is one of the things we seem to do best, and they created little trouble during 2008-2009, or for that matter during the pandemic
CBDC, and sadly so
Crypto forensics
Twitter and weird anon accounts
When would the trouble have been exposed if not for Twitter? And much of the best coverage came from accounts with names like Autism Capital.
Some critics (like Aaron Levie), too.
Bitcoin
After a cataclysm for the crypto sector, it’s down about 15% over the past month. That’s less than the S&P 500 lost during the worst month of the GFC.
Falls
Effective Altruism
A totalizing worldview that has enabled some undesirable weirdness in different places.
Valorizing “scope sensitivity” and expected value leads people violently astray.
Being unmarried (and male) above the age of 30
Being on the cover of magazines
And if you see “the next Buffett”, run.
Appearing with blonde models
Buying Super Bowl ads and sponsoring sports and putting your name on arenas
“Earn to give” as both a concept and a phrase
Mrs. Jellyby
The concept of self-custody
Weird locations for corporate offices
Venture capital
Our ability to see crazes for what they are in the moment
This is not just, or even mainly, about crypto
Drugs
Adderall and modafinil, perhaps stronger stuff also played a role.
The children of influential faculty
Do they grow up witnessing low-accountability systems and personality behaviors?
Coda, Personal stories:
The “why” is still interesting dramatically. Why do this? Also is there a love story of sorts with Caroline?
Do sign up to the substack newsletter below.
More coda:
Zvi has a very detailed summary of what we likey know here.
Scott / Astral Codex covers the drug abuse angle.
No one is reall covering the possible love story, but maybe there isn’t really one. I think there is still, sort of.