Understanding FTX’s crimes
In the aftermath of SBF’s convinction, there have been a few posts trying to make sense of FTX. Some people are trying to figure out what happened, and some people are interested in trying to find clever defenses.
I’m in a much more boring position: I am confident SBF is the fraud the world believes him to be. I hope this post can provide reasoning transparency on why I think this, and perhaps serve as an easy link for others who feel similarly but don’t want to get bogged down in a point-by-point.
Posted anonymously as some protection against future employers Googling [1].
I have divided this post into a summary of the major crimes and my basis for believing they occurred, an ‘FAQ’ dealing with some common misapprehensions I’ve seen on this forum and elsewhere, and an appendix explaining some crypto exchange basics / jargon for those who don’t have a background there.
Crimes
Misappropriation of Funds
Summary
This is the big one. Customers who deposited to FTX believed their assets were being held separately to FTX’s own funds. In reality, their funds were available for Alameda to use freely.
For practical purposes, there wasn’t any separation between these two companies; FTX’s money was Alameda’s money and Alameda’s money was FTX’s money. Since SBF was majority-owner of both Alameda and FTX, this overlap is not obviously [2] illegal, though it is highly inadvisable. However, if customer money is FTX’s money and FTX’s money is Alameda’s money and then Alameda invests a ton of that money, all while SBF tweets to customers that their money is safe and uninvested [3], that’s a problem.
Detail
If a troubled company has a few days to beg potential investors for a bailout before it files for bankruptcy, and it sends those investors its balance sheet so they can consider investing, and they all pass, and then the company files for bankruptcy, of course the balance sheet was bad. That is not a state of affairs that is consistent with a pristine fortress balance sheet.
But there is a range of possible badness, even in bankruptcy, and the balance sheet that Sam Bankman-Fried’s failed crypto exchange FTX.com sent to potential investors last week before filing for bankruptcy on Friday is very bad. It’s an Excel file full of the howling of ghosts and the shrieking of tortured souls. If you look too long at that spreadsheet, you will go insane.
-Matt Levine, FTX’s Balance Sheet Was Bad
There’s a ton that could be written here, but since intent seems to be the main point of contention I think the most interesting data points are the ones that suggest how big of a problem the insiders thought it was before facing criminal charges:
Here’s SBF’s balance sheet that he circulated to investors during the panic of November 2022. As far as I know this sheet has never been verified, and it’s from a convincted felon, and it was essentially a sales pitch. So it could be expected to present a perhaps-too-rosy view of the world. What it in fact presents is an unmitigated disaster. Matt Levine’s reaction above was the same as my first reaction, and I recommend that piece as a whole if you want to get a vibe of how insane this was.
Roughly the sheet lists assets as follows, rounding to the nearest $100m and using the ‘October / Before This Week’ values; note that many of these took substantial hits during the November panic.
$6bn FTT
$5.4bn SRM
$2.2bn SOL
$1.5bn ‘Other Ventures’
$1.2bn GDA; this is Genesis Digital Assets, a mining company
$500m Anthropic
$865m MAPS
$600m HOOD; Robinhood shares owned personally by SBF
$1.5bn of other assets of varying quality
Total of $19.6bn
There isn’t as much colour on the liabilities, but based on the section at the bottom they were $8.9bn, all owed to customers.
One might naively think that $19.6bn of assets against $8.9bn of liabilities is fine. It is not. The most obvious way it was not fine is in the two largest assets; FTT and SRM. These were both ‘SamCoins’. FTT functioned somewhat like stock in FTX and so naturally became unsellable as problems took hold. SRM had a floating market cap of $200m. How can FTX hold $5.4bn if the market cap is $200m? Because the vast majority of SRM tokens were not floating; they had been minted by SBF for SBF, were not sellable, and in any case would not have been sellable for anything close to $5.4bn.
Revaluing those two holdings to zero, as would have been appropriate before the fact, is already sufficient to make FTX insolvent. But many other holdings are also wild, and again would have been wild before the fact:
MAPS is another SRM situation; $865m of ‘value’ while having a $6m market cap.
What are SBF’s personally-owned HOOD shares doing on here?
Did Anthropic and GDA realise their $500m / $1.2bn was effectively FTX’s customers investing in them, and that those investments might need to be sold if those customers came knocking?
In fact, who authorised FTX to build a $1.5bn ‘Other Ventures’ porfolio with customer funds? And so on.
Most importantly, the dollars, stablecoins, BTC and ETH that presumably made up the vast majority of customers’ original deposits are barely present.
One thing that was different before the fact, as seen right at the bottom, was that FTX had an additional $5bn in liquid assets, and a corresponding extra $5bn in customer liabilities. That $5bn was returned to customers during the November panic.
But while this explains why SBF may have thought they could survive in October 2022 - they’d never seen anything close to $5bn of withdrawals before! - it does not in any way excuse the clear imbalance and misuse of funds demonstrated by this sheet as a whole.
Caroline Ellison also held an all-hands meeting at Alameda during the panic. Unbeknownst to her, she was being recorded. Here’s the full recording if you want to listen, and a summary of some key exchanges. Note that at the time, FTX thought it was going to get bought out by Binance and customers would be made whole in short order[4]; at one point Caroline notes that she wasn’t sure if the details would ever get out publicly. Again, this is Caroline talking about Caroline, so one might expect the story to be slanted in her favour. Yet Caroline acknowledges that:
Alameda ‘borrowed’ a lot of FTX user funds during the crypto credit crunch of May / June 2022.
This was done via special privileges that only Alameda had.
Alameda is insolvent.
The resulting hole on FTX is closer to $6bn than to $1bn.
She talked about the June 2022 decision with Sam, Gary, and Nishad.
Alameda, and Alameda alone, was always allowed to borrow FTX customers’ funds. While Caroline highlights June 2022 as an inflection point, when asked directly she does not claim there was no misappropriation before that point. I’ll revisit this in the FAQ, but here’s the key exchange:
DIANA MA: Who made the decision on using user deposits?
CAROLINE ELLISON: Um...Sam, I guess.
MA: And when was it, like first touched?
ELLISON: I think, like, FTX, like, basically always allowed Alameda to, like, borrow user funds, as far as I know.
MA: Just Alameda?
ELLISON: Yeah.
In mid-2021, SBF wanted to buy back Binance’s share of FTX, originally purchased for $80m. Binance demanded $2bn. This was considerably higher than FTX’s lifetime revenue of $1bn, let alone profit, and Alameda hadn’t made enough profit to fill the gap either. Note FTX also lost several hundred million dollars in the MobileCoin incident earlier that year.
So Alameda and FTX were not close to being afford this with their own funds. Caroline came to the same conclusion, per her trial testimony:
At the time, Binance’s equity in FTX was worth approximately $2 billion, and FTX’s revenue was only half that amount. According to her trial testimony, Ellison told Bankman-Fried, “We don’t really have the money for this, we’ll have to borrow from FTX to do it.” (Tr. 668). Bankman-Fried responded, “That’s okay. I think this is really important, we have to get it done.” (Id.). Bankman-Fried then repurchased the FTX shares owned by Binance using a mix of FTX funds and customer deposits. (GX-1024). The transfer came from Alameda’s main account on FTX, which was able to make the transfer, notwithstanding the fact that it had insufficient funds, because of the allow negative feature that had been enabled on its account.
In late 2021, despite the hole created by MobileCoin and the Binance buyout, SBF wanted to and later did invest billions in less-liquid venture investments. Again, Caroline claims she pushed back but ultimately allowed herself to be overruled:
In the fall of 2021, Ellison told Bankman-Fried that Alameda had more loans than assets, and that without certain FTX-related cryptocurrencies such as FTT, Solana, and Serum, Alameda’s net asset value was negative $2.7 billion. (Tr. 710). Her analysis also showed that in the event of a market downturn and the recall of Alameda’s loans, the only way to repay third-party loans would be to borrow billions of dollars of FTX customer funds. (GX-36). Ellison told Bankman-Fried that she thought that new venture investments were a bad idea and too risky. (Tr. 710-11). Notwithstanding the risks, BankmanFried told her “he wanted to go ahead with billions of dollars of venture investments.” (Tr. 729).
I want to call out that negative $2.7 billion in the above. ‘Net Asset Value’ is finance-speak for ‘what is this worth?’. So what Caroline calculated is that if FTX-related cryptocurrencies became worthless, Alameda would be insolvent to the tune of $2.7bn. It is still unclear to me how Alameda lost so much money during a couple of years when crypto went up a ton, though I assume the MobileCoin and the Binance buyout were a large part of it. Put another way, $2.7bn is how much Alameda would need to raise from rapid sales of assets such as FTT if everybody—lenders and customers—asked for their money back at once, as they had the right to do at any time. And Caroline was clearly aware that the FTX-related coins becoming ~worthless was a real worry in that scenario, otherwise why calculate this number? At any rate, $2.7bn is already a very tall order, and per (1) it got worse from there.
Much earlier in time, shortly after FTX launched, co-founder Gary Wang heard a trader ask about Alameda’s negative balance on FTX. A negative balance in an individual coin might be Alameda making use of the margin trading feature, as described in the Appendix. But this appears to be a use of their special-to-Alameda ‘allow negative’ feature; an overall negative balance is essentially an overdraft, and represents Alameda borrowing from FTX.
SBF responded that “it was okay as long as it was lower than FTX’s total trading revenue”.
This is actually kind of true? It’s not how I would set things up, but as detailed in the Appendix exchanges do need some mechanism to take out the customer money that is now theirs after fees, and I guess withdrawing it to your CEO’s hedge fund is one such way. Not legal advice!
But according to Gary, rather than stopping once 100% of revenue had been withdrawn, the number kept going up. Some background context here is that Alameda didn’t have outside investors. It was always a bit of a mystery how they acquired their starting capital base—the money they were trading with—but it being crypto people assumed they’d made some very good (or very lucky) investments. In practice, a big chunk of it was this ever growing line of credit:
Around 2019 or 2020, Wang checked the database and discovered that Alameda was negative by about $200 million, which was more than the $150 million FTX made in revenue. That had to mean Alameda was taking customer money. It surprised him and he says he discussed it with Bankman-Fried, who told him that he just needed to take into account the value of FTT. Wang said he trusted Bankman-Fried’s judgment and didn’t pursue the issue further. Later, though, Alameda’s balance was more negative than FTT and the trading revenue, and Wang had more conversations with Bankman-Fried about it. Wang said he knew at the time that Alameda was using FTX’s customers’ funds and that he knew it was wrong....The line of credit didn’t start that big — Wang testified that Bankman-Fried had asked him to increase it a few times because Alameda kept running into its limits. First it was “a few hundred million,” then a billion, then even that wasn’t enough.
Lying to lenders
If you prepare a balance sheet for a lender and your boss says “why don’t we present this information in a different way,” you probably need a lawyer. If you prepare SEVEN BALANCE SHEETS and your boss is like “let’s go with Alternative 7” then one of you is going to prison absolutely forever. Caroline Ellison smartly decided it wasn’t going to be her.
-Matt Levine, FTX Had Many Bad Spreadsheets
The next three crimes are mostly in service of the big crime. If you’re committing fraud against your customers, you can’t say so, so you will be forced to lie to people who ask you questions.
As a result I won’t go into as much detail on them, though they are significant crimes in their own right and I do think they provide interesting insight into what the inner circle thought of their situation at various points in time. If you think everything you’re doing is above board, you don’t need to do things like this.
Summary
Alameda’s other source of funding was the crypto lending space. As part of doing due diligence on the loans, the lenders asked for things like balance sheets. An accurate balance sheet would of course have highlighted the loans from FTX, so Alameda falsified them.
The clearest example of this comes from June 2022, when Alameda’s lenders requested updated financial information including a balance sheet. Caroline made a ‘Main’ sheet showing Alameda’s true financial condition. Like me above, she quickly assessed that Alameda’s condition was terrible, so terrible they couldn’t show it to a lender. So she made 7 alternatives and showed them to SBF, who picked one that deleted the related party loans—money ‘loaned’ to the inner circle—and the FTX borrows.
The CEO of one of Alameda’s lenders, BlockFI, testified in court that the information they received also edited these out.
“We always relied on the information that we were given by counterparties as being truthful and accurate,” said Prince...The statements provided by Alameda didn’t disclose the true extent of the firm’s liabilities, nor the existence of substantial loans from Alameda to FTX and company executives. All of those things would have led to concerns on BlockFi’s end, said Prince, depending on the amounts and the details of the transactions involved. Prince testified that if the true extent of liabilities and the outstanding loans had been disclosed, “we probably wouldn’t have lent to them at all” due to concerns that the firm might be insolvent.
Lying to FTX’s investors
Summary
FTX’s investors were of course intensely suspicious of the relationship between FTX and Alameda, and asked many questions about whether Alameda received special treatment. They were told no. This was false.
Gary Wang’s testimony is the one that goes into most detail on Alameda’s privileges. Even after it all I collapsed I was surprised when he testified Alameda was even allowed to place orders ahead of other users, which was a specific question that FTX was asked over and over by investors, other market makers trading on the platform, and half of crypto Twitter.
Today, the prosecution grilled Wang on a single, damning column in FTX’s databases. Called “allow_negative,” it allowed Alameda to have a negative balance. Alameda could withdraw money even when its accounts didn’t have any, and it had an enormous line of credit. Wang added that it could place orders faster than other users...
Lying to Banks
Summary
Apart from the abuse of the ‘allow negative’ flag, Alameda’s other path to customer deposits was the ‘fiat@’ account. Customers sent money to FTX by sending them to bank accounts controlled by Alameda, often in the name of ‘North Dimension’. Alameda proceeded to spend those funds as if they were Alameda’s own funds. Bank regulation is supposed to prevent this kind of thing, and the bank in question did ask questions. Alameda lied to them.
Detail
It is hotly disputed when various people became aware of Alameda failing to hold this money to one side. Not so disputed, as best I can tell, is that the banks offering the accounts were not told that this is what they were being used for, even when they directly asked about it.
Alameda never told Silvergate Bank that it was using its accounts to process FTX customer deposits and withdrawals...For example, on May 14, 2020, an employee at Signature emailed FTX.com and Alameda personnel asking about a transfer which was received by a bank account held in the name of Alameda but which said it was related to FTX. Instead of telling the truth, at Bankman-Fried’s direction and with his knowledge, an Alameda employee falsely responded that the transfer was actually related to Alameda.
Yeah, don’t do that. In a chaotic fast-changing company it may occasionally happen that you lose track of which account should be used for what. That’s not really any excuse for lying when the bank asks the question.
Perhaps spooked by these questions, SBF ramped the opacity up a notch. From same link:
Silvergate Bank made clear, however, that it would not open an account for customer deposits and withdrawals absent evidence that FTX was licensed and registered...
Notwithstanding those warnings from Silvergate Bank, in August 2020, to further obscure the relationship between FTX and Alameda, Bankman-Fried directed the incorporation of a new U.S.-based entity, North Dimension...Silvergate Bank was also given a completed North Dimension due diligence questionnaire—which Bankman-Fried signed—that falsely stated that North Dimension “trades on multiple cryptocurrency exchanges worldwide for its own account” and that North Dimension “also participates in direct peer-to-peer, OTC purchases and sales with certain third parties for its own account.”...
In April 2021, Silvergate approved the opening of the North Dimension account, based on those misrepresentations, without enhanced due diligence or review by Silvergate Bank’s executive committee, as would have been required had the true purposes of North Dimension’s account been disclosed to Silvergate Bank. Once the North Dimension bank account was opened, FTX directed customer dollar deposits to the North Dimension account...
Bank regulation exists for a reason. The reason FTX customer deposits aren’t suppsoed to be sent to Alameda accounts is because those deposits should live inside FTX accounts. To any bank, regulator, or auditor, having those deposits sit in accounts controlled by Alameda would represent an unacceptable risk that they would be commingled with Alameda’s other funds and ultimately spent. Which is exactly what happened.
Bribing a Chinese official
In early 2021, the Public Security Bureau in Changge, China froze two cryptocurrency trading accounts held by Alameda on two of China’s largest cryptocurrency exchanges.
As far as I know it’s unclear what prompted this. It could have been a shakedown, could have been someone getting concerned about possible fraud, or just a general distrust of crypto. The reason given to Alameda was apparently money laundering by one of Alameda’s counterparties, but I don’t think that means much; at least in the West tipping off is a serious offense that almost always stops you telling the person you’re investigating why or even whether you’re investigating them.
At any rate, Alameda lost effective access to $1bn, not a small sum. They responded with a huge bribe, immortalised in Caroline’s list of major losses as ‘the thing’.
In November 2021, after several months of failed attempts to unfreeze the accounts, Bankman-Fried directed one Chinese FTX employee to make a multi-million-dollar bribe to government officials to have the accounts unfrozen...On or about November 16, 2021, the first bribe payment of cryptocurrency then worth approximately $40 million was transferred from Alameda’s main trading account to a private cryptocurrency wallet. Shortly thereafter, the accounts were unfrozen. After confirmation that the accounts were unfrozen, Bankman-Fried authorized the transfer of an additional approximately $100 million of dollars in cryptocurrency to complete the bribe.
This is a clear breach of anti-bribery legislation, e.g. the Foreign Corrupt Practices Act (FCPA). This rough contours of such legislation are part of basic compliance training for many finance employees; it is thoroughly implausible that SBF was entirely unaware of it, and Caroline was clearly sufficiently aware that there was a problem to rename it.
Falsified revenue
Summary
Nishad testified that FTX’s 2021 revenue, widely reported as $1bn, was actually just short of $1bn. This displeased SBF.
I find this story interesting as an insight to what crimes SBF was willing to commit even when not ‘under the gun’; while hitting $1bn is a nice morale-boosting round number, $950m instead was hardly going to make or break the business the way some of the other decisions did. In spite of these seemingly far lower stakes, SBF asked Nishad to make it happen.
For example, in late 2021, when Bankman-Fried realized that FTX was $50 million short of his goal of earning one billion dollars in annual revenue, he instructed Singh to transfer funds from another entity that he controlled, and to falsely characterize the $50 million as revenue that FTX earned throughout 2021. Singh then backdated a series of fraudulent transfers, and later lied to auditors about the transfers and created false documentation to support those lies.
Falsified Insurance Fund
Another lower-stakes crime that gives insight to how readily SBF committed crimes is FTX’s insurance fund. Most crypto exchanges have insurance funds, and they are supposed to provide a layer of protection in the event that a sharp market move causes an account to go negative too fast for the exchange to sell the collateral; the insurance fund covers the loss first. If there really had been some Alameda margin blowup due to plummeting FTT or similar, it could have come in handy!
Generally, these are funded by the profits exchanges make on liquidation during more ordinary market conditions. At FTX though, we got this from Gary’s testimony:
Code snippets shown to the jury demonstrated how Nishad Singh wrote some code that would update the insurance fund amount by adding to it the daily trading volume, multiplied by a randomish number around 7,500, and dividing it by a billion, thus making it appear as though the website was referencing a real account balance that was fluctuating as the exchange added funds or withdrew from it to cover losses. In reality, it was all made up.
FAQ
Didn’t SBF only know about the hole in June 2022?
In a word, no. June 2022 comes up a lot for a few reasons:
It is when the prosecutors felt comfortable proving beyond reasonable doubt that SBF knew about the hole. By that time they had a spreadsheet detailing it which he had left digital fingerprints on, and Caroline / Nishad / Gary all testified to conversations around that time.
Link is long, so giving exact quote:
So they, customers, had deposited 13 billion and only 3 billion was available at FTX. The other 10 billion, that minus 10, that’s what they borrowed on FTX. So this spreadsheet, Government Exhibit 50, is very important and from it the defendant knows the following: 1. Alameda has a total negative balance of $11 billion; 2. It has borrowed over $13 billion in customer money; 3. Around 3 billion of that is by going negative on FTX in its main account. And there is another 10 or 11 by taking fiat deposits. Finally, 4, as little as a quarter of the customer money is actually on FTX. This shows there was a gigantic hole. This spreadsheet is a key piece of evidence because it basically shows you everything you need to know about the defendant’s conspiracy, and what’s so important about it is that he and his coconspirators all work on it together. It’s absolute proof that he knew in June 2022.
The hole got bigger around this time, due to repaying Alameda’s other lenders with customer funds.
One of the mechanisms that Alameda used to borrow from FTX—the customer deposits into Alameda bank accounts and associated ‘fiat@’ account—was most obviously known that time, since the liability there is included in the spreadsheet and in some of the balance sheets prepared for lenders.
But as detailed above, Caroline testified that she told SBF that they could not afford to buy out Binance in mid-2021; they’d have to borrow from FTX. Gary testified to noticing and discussing Alameda’s too-negative balance even earlier, in 2019 or 2020. While I concede it is possible they are lying, it seems much more likely to me that they are telling the truth.
The focus on fiat@ in some places is a little odd. Alameda had an unlimited line of credit from the FTX exchange itself with no collateral requirement[5]; it could withdraw any number of dollars it needed to via that pathway, limited only by how many dollars customers had put in. Customers sending deposits direct to Alameda accounts just provided a marginally more convenient way of misappropriating the funds.
In terms of figuring out when the fraud really started, it’s worth noting that this is chiefly challenging because:
SBF instructed employees to write as little as possible, and if things must be written to use Signal and its auto-delete feature.
SBF was CEO of Alameda and FTX until 2021, and presumably felt little need to share details of any fraud he was committing with his underlings, so there’s a lack of witness testimony from before this time. It’s perhaps unsurprising that what testimony we do have from this time came from Gary, who coded much of the system and was able to run database queries independently.
So ultimately you have to guess whether the SBF who took billions upon billions of customer funds in 2021 and 2022 acted in a sensible and restrained manner in 2019 and 2020, despite no technical or governance barrier to taking funds at any time and Gary’s testimony to the contrary. The same SBF whose chosen management team universally decided he needed to go in 2018, in part because of resistance to informing investors about misplaced funds. I know what I think.
Didn’t customers know their funds might be lent out?
FTX did have a margin program, where customers could lend to other customers. You had to sign up for it. There were only a few billion dollars in there. You could also borrow from it, and see the rules and formulas which described under what circumstances you would be able to do so; in short the borrower would have to hold other assets and borrow against them, maintaining overall positive equity, similar to what I describe in the Appendix. The lower the quality of those assets, the less you could borrow. The larger your position, the less you could borrow as a % of your assets. There was also an insurance fund that should pay out in the event of a sudden market move that caused a borrower’s assets to be insufficient to cover the loan.
In practice, all customer funds were available, not just those of customers who signed up.
In practice, Alameda did not have to post collateral on the exchange, and so ended up with the curious collection of assets described above.
And of course the insurance fund was a pure fabrication.
‘Customers knew their funds might be lent out under some circumstances’ is only a defense to the extent that the actual circumstances were reasonably close, and SBF in fact discussed this possible line of defense with Can Sun, FTX’s general counsel. Here’s how he responded, bold added:
And in brief, explosive testimony, former FTX lawyer Can Sun demonstrated what Bankman-Fried’s tactic was....Bankman-Fried came to Sun to ask him to come up with justifications for the missing funds.
That “basically confirmed my suspicions that had been rising all day” that Bankman-Fried had purloined the funds, Sun said. He ran Bankman-Fried through possible explanations, including a margin loan program, and explained that none of them could fit the actual situation — for instance, Alameda had borrowed more from FTX than had ever been in the margin loan program. He said Bankman-Fried acknowledged their conversation with a “yup, yup.” “I was expecting a bigger response, but it was very muted,” Sun said.
After that, Singh spilled the beans: Alameda had withdrawn assets that included the customer funds, Sun testified. He quit the next day.
Weren’t FTX and Alameda total messes? Intent matters; isn’t it possible that this all got lost in the chaos?
I’m not going to dispute that these companies were a mess. In the now-infamous words of John Ray, who previously dealt with the Enron mess:
Never in my career have I seen such a complete failure of corporate controls and such a complete absence of trustworthy financial information as occurred here.
I do want to point out the obvious fact that if you’re going to commit multi-billion-dollar fraud, keeping your companies complete messes and firing people who seem like they ask too many questions is an excellent way to go about it. So is not hiring a CFO. Incidentally, Sam’s response to VC’s trying to push him to hire one, quoted in Going Infinite, was one of the most hilarious moments of the book for me:
“There’s a functional religion around the CFO,” said Sam. “I’ll ask them, ‘Why do I need one?’ Some people cannot articulate a single thing the CFO is supposed to do. They’ll say ‘keep track of the money,’ or ‘make projections.’ I’m like, What the fuck do you think I do all day? You think I don’t know how much money we have?”
It is just about possible that the complacency demonstrated by that quote, and the weak governance created to cover up other crimes, blinded SBF to the true magnitude of the misused funds until it was too late. So he may have been genuinely shocked to discover that he was only able to return around $6bn of $15bn in customer funds in November before declaring bankruptcy. But this is not a defense if that state of affairs is deliberate!
To be honest, if the total amount of misused money were smaller; <25% FTX’s total revenue, or <10% of customer deposits, or <25% of SBF’s spending spree, I would put more weight on the idea that insiders didn’t realise they were in a hole billions of dollars deep. This wouldn’t do much for criminal liability, but perhaps worth keeping in mind on some level. But the fact is that the amount was a large multiple of FTX revenue, perhaps half of pre-panic customer deposits, and most of SBF’s spending. And of course Caroline’s actual testimony is that SBF was very aware in 2021. They calculated out Alameda’s financial position and discussed what it would take to make $3bn of venture investments, and he just decided to ignore her recommendation not to do so.
Isn’t FTX going to make all Customers whole? So there wasn’t ever a hole?
A thief who takes his loot to Las Vegas and successfully bets the stolen money is not entitled to a discount on the sentence by using his Las Vegas winnings to pay back all or part of what he stole if and when he gets caught.
-Judge Kaplan, sentencing SBF to 25 years in prison
Technically yes to the first, no to the second. There are a few things going on here:
FTX cashed out customers at bankruptcy prices
If you held pure dollars on the exchange at time of bankruptcy, you’ll likely get the vast majority of those dollars back. Eventually. But if you held 1 BTC at time of bankruptcy, your claim has been ‘dollarised’ at a price of approximately $17k and you are now in a similar position to someone who held $17k. Even though BTC is now $66k, the bankruptcy court will consider you to have been made whole if you receive your $17k back. In total customer claims amount to roughly $9bn calculated this way, but the assets people lost would be worth considerably more than $9bn now.
The assets those customers’ money had bought did well
Looking back at SBF’s balance sheet, one of the largest holdings at time of bankruptcy was $1bn of SOL. At the time SOL traded around $14. It now trades around $182; the same stash would be worth perhaps $13bn. This alone meets the $9bn of liabilities mentioned above, though the estate has presumably sold much of its SOL by now so actual profit is unknown. Similar, though less extreme, dynamics have played out with the Anthropic investment and GBTC holdings.
Had FTX stayed afloat, things would be very different
It’s also important to flag that many of the assets that have been recovered were only recovered because FTX went bankrupt. Charities have returned money, business and executives’ real estate is being sold, SBF’s Robinhood shares were confiscated and sold for $600m, Modulo Capital returned $465m, and so on.
But perhaps more importantly, SBF was stopped from spending money like water.
We’ll never know what would have happened if Alameda’s balance sheet hadn’t been leaked [6]. Maybe crypto would have recovered, Alameda would have cashed out some assets to replace the missing customer funds, and the enterprise would have continued with nobody the wiser.
But I rather doubt it. Nothing in SBF’s behaviour during the previous years suggests such a disposition to take his customers’ chips off the poker table. Everything suggests someone who was so convinced of their invincibility, so sure of the need to keep flipping the slightly-favourable coin, and so in awe of their newfound power that they would have felt virtually obligated to continue as they were in 2022.
Thanks goodness he was not able to.
*******************************************************************
Appendix
How crypto exchanges work
Spot Trading
Crypto exchanges are a little different to regular exchanges, because they also take roles that in traditional finance would be met by broker-dealers and clearing houses. Nonetheless, the basic business model is not that complicated:
Customer A sends the exchange $70000
Customer B sends the exchange 1 BTC
The exchange now holds $70000 and 1 BTC, and owes A $70000 / B 1 BTC.
A then buys and B then sells 1 BTC in exchange for USD, at a BTC/USD price of $70000.
Now A has 1 BTC, and B has $70000...except they don’t quite, because the exchange charges fees. If fees are 0.05% on each side, A has 0.9995 BTC and B has $69965.
The exchange still holds 1 BTC and $70000, but customer liabilities are now 0.9995 BTC and $69965; the 0.0005 BTC and $35 are the exchange’s revenue.
What the exchange does from there is mostly up to them, but one sensible approach would be to periodically ‘sweep’ the revenue into separate accounts so that it’s segregated from customer funds, and do things like pay salaries only out of those accounts the revenue is swept into. This transfer also creates a record of what you thought your revenue was at that point, which can then be verified by your own staff and eventually by auditors.
Margin Trading
One way the above can get complicated is if the exchange allows margin trading, as FTX did. Regular brokers allow margin trading as well! Suppose it’s the end of 2021 and you have $1m of META (Facebook) stock held at SuperBroker[7]. You need $200k for a house deposit.
You could sell some of your META stock, but you don’t want to, perhaps because you think it’ll go up even more. So instead you borrow $200k from SuperBroker, and your position on SuperBroker now looks like:
$1m META
-$200k cash
There are two interesting questions here:
Where did the $200k come from?
Varies. Could be SuperBroker’s other customers, or from business funds, or from a third-party lender funding their margin program.
What happens if you can’t pay it back?
Well, SuperBroker is holding $1m of META. They will no longer allow you to transfer that META to another broker. Also, if META falls 70%, as it did in 2022, you’ll very likely get a call from SuperBroker requesting that you return at least some of the cash. Should you refuse or ignore the calls, they will sell your META stock and replace the cash themselves; this is called liquidation.
This works fine as long as:
The value of your META stock according to SuperBroker is at least reasonably close to what they can sell it for. In particular, if they liquidate you when the face value of your META stock hits $300k, they better be able to get at least 2/3rds of that, or they might not get their $200k back.
They do actually have the nerve to liquidate you when the time comes, even though this will risk your ire if META then recovers, as it did in 2023.
If SuperBroker can’t value your META shares accurately, or if they know they won’t have the nerve to liquidate you, they shouldn’t offer you the margin in the first place.
- ^
Though the email used makes me readily identifiable to EA Forum staff, and I wouldn’t be surprised if the style is familiar to people who know me well.
- ^
The most likely way for it to be illegal is if this fact was not represented accurately to other owners of Alameda / FTX, or to the banks providing the accounts. We’re getting to that.
- ^
The tweets were later deleted a day later, but honestly I don’t think that matters much. To tweet that you don’t invest Customer funds at all, in an attempt to forestall a bank run that will wipe you out, when you definitely knew by then that this was false, might be sufficient for prison time all on its own.
- ^
Binance then passed after doing some due diligence and seeing the balance sheet deconstructed above, resulting in FTX’s bankruptcy.
- ^
Ok fine, technically it was $65bn.
- ^
The sheet was dated June 30th 2022, but this actually appears to be one of the doctored balance sheets; no mention of related party loans or borrows from FTX exchange.
- ^
Not a real broker!
Closely related to “misappropriation of funds” is “lying to customers.” That ends up being a lesser offense than misappropriation, but I’ll note it since it is more intrinsically immune to the “intent matters / chaos” line of defense. As one example (quoting the prosecution’s sentencing memo):
Once you go down the path of lying to customers to induce them to give you money (and to a regulator charged with customer protection), any losses that are causally related to those fraudulent statements are on you. The chaos theory just won’t gel with an honest belief that customer assets were “appropriately safeguarded and segregated” and “held in trust.” I do consider this a lesser offense than misappropriation because the specific intent required is obtaining the money through deceit rather than affirmatively stealing it. But it’s still a very serious offense.
Thanks for posting this—on a quick read it looks pretty accurate to me and I’ll be glad to have this as a resource to point people to when they seem not to understand exactly why what FTX did was so bad.
This is great!
Note that this list is not comprehensive. In-particular it doesn’t go into detail on any of the campaign finance violations and other policy-stuff that Sam was involved in, which I think never ended up investigated due to a kind of weird agreement with the Bahamas authorities. But during the trial we did hear some pretty clear evidence of at least campaign finance violations (and my guess is there would be a bunch more if one kept digging, as well as stuff that wouldn’t necessarily be crimes but stuff I think most people here would still consider highly unethical).
Not weird. If you get a foreign country to extradite someone on charge X, you can’t turn around and prosecute them for charge Y (if charge Y is different enough). That’s a basic principle of extradition law, and I agree with the Bahamas that the campaign-finance violations were different enough from the original counts.
https://apnews.com/article/bankmanfried-ftx-crypto-charges-ellison-60aa798f4aacb0ddd6a422b72e5a7614
They probably could have recharged after some legal tapdancing, but they decided to ask Judge Kaplan to consider this conduct at sentencing (which he did based on the Guidelines calc he did). They knew recharging was very unlikely to change the final sentence.
I… think I will continue describing this as “weird” though it makes sense that as a lawyer it’s not that weird to you.
It feels very off to have a bunch of crimes uninvestigated because someone fled the country and then was extradited, and I am pretty confused why the Bahamas cooperated with Sam here (my guess is that it’s a case of political corruption, though I can also imagine other reasons). It’s not like the Bahamas had anything obvious to gain from Sam not being convicted of the campaign finance violations.
That’s fair.
Broader concerns about sovereignty and the integrity of the extradition system likely played a role. Although it’s understandable why US prosecutors asked for provisional arrest and extradition despite not having obtained an indictment for campaign-finance violations yet, it’s also understandable to me why the Bahamas wants to signal its expectation that the US comply with the treaty as written—especially where the argument that the first extradition was good enough for these charges is legally weak.
Note that the US could have asked for permission to proceed on other charges—see Article 14(b) of the treaty—but I think that may have required going back through the Bahamas legal system again, and probably delaying the trial. Hence, the prosecution severed off those charges, and then decided not to proceed after getting its conviction on others. I don’t know Bahamas campaign or extradition law, but there would also be difficulties if the offense was deemed one of a political character (Article 3(1)(a)) or was one that was not a criminal offense punishable by more than a year in prison under Bahamas law (Article 2(1)).
Also, I don’t think “completely uninvestigated” is a correct characterization—they were investigated enough to be presented to a grand jury, which indicted SBF for campaign-finance violations. Federal prosecutors do not generally indict without a pretty good investigation first, especially in high-profile cases. I think we have a pretty decent idea of what he did (see pp. 18-22 of the prosecution’s sentencing memo). Moreover, Salame and Singh—who don’t have extradition-related issues—pled guilty to campaign-finance violations.
Oh, that is interesting and an update for me. I had interpreted the relevant section of the memo as more of a “we didn’t really investigate it, but here is what we have”, but you know this stuff much better than I do.
I’m confused—elsewhere you identify yourself as the author of this post but here you are commenting as if you have independently reviewed it?
Habryka identifies himself as the author of a different post which is linked to and being discussed in a different comment thread.
Oh ok, thanks! Sorry for my confusion.
Thanks. In addition to lots of general information about FTX, this helps answer some of my questions about FTX: it seems likely that FTX/Alameda were never massively profitable except for large bets on unsellable assets (anyone have better information on this?); even though they had large revenues maybe much of it was spent dubiously by SBF. And the various actions needed to maintain a web of lies indicate that Caroline Ellison and Nishad Singh, and very likely Gary Wang and Sam Trabucco (who dropped off the face of the earth at the time of the bankruptcy [1]) were definitely complicit in fraud severe and obvious enough that any moral person, (possibly even a hardcore utilitarian, if it was true that FTX was consistently losing money), should have quit or leaked evidence of said fraud.
Four or five people is very different from a single bad actor, and this almost confirms for me that FTX belongs on the list of ways EA and rationalist organizations can basically go insane in harmful ways, alongside Leverage, Zizians and possibly others. It is not clear that FTX experienced a specifically EA failure mode, rather than the very common one in which power corrupts.
I was confused by this until I read more carefully. This link’s hypothesis is about people just trying to fit in―but SBF seemed not to try to fit in to his peer group! He engaged in a series of reckless and fraudulent behaviors that none of his peers seemed to want. From Going Infinite:
Example from Matt Levine:
It sounds like SBF drove away everyone who couldn’t stand his methods until only people who tolerated him were left. That’s a pretty different way of making an organization go insane.
It doesn’t seem like this shouldn’t be an EA failure mode when the EA community is working well. Word should have gotten around about SBF’s shadiness and recklessness, leading to some kind of investigation before FTX reached the point of collapse. The first person I heard making the case against SBF post-collapse was an EA (Rob Wiblin?), but we were way too slow. Of course it has been pointed out that many people who worked with / invested in FTX were fooled as well, so what I wonder about is: why weren’t there any EA whistleblowers on the inside? Edit: was it that only four people plus SBF knew about FTX’s worst behaviors, and the chance of any given person whistle-blowing in a situation like that is under 25%ish? But certainly more people than that knew he was shady. Edit 2: I just saw important details on who knew what. P.S. I will never get used to the EA/Rat tendency to downvote earnest comments, without leaving comments of their own...
You know what, I was reading Zvi’s musings on Going Infinite...
And it occurred to me that all SBF had to do was find a few people who thought like Michael Lewis, and people like that don’t seem rare. I mean, don’t like 30% of Americans think that the election was stolen from Trump, or that the cases against Trump are a witch hunt, because Trump says so and my friends all agree he’s a good guy (and they seek out pep talks to support such thoughts)? Generally the EA community isn’t tricked this easily, but SBF was smarter than Trump and he only needed to find a handful of people willing to look the other way while trusting in his Brilliance and Goodness. And since he was smart (and overconfident) and did want to do good things, he needed no grand scheme to deceive people about that. He just needed people like Lewis who lacked a gag reflex at all the bad things he was doing.
Before FTX I would’ve simply assumed other EAs had a “moral gag reflex” already. Afterward, I think we need more preaching about that (and more “punchy” ways to hammer home the importance of things like virtues, rules, reputation and conscientiousness, even or especially in utilitarianism/consequentialism). Such preaching might not have affected SBF himself (since he cut so many corners in his thinking and listening), but someone in his orbit might have needed to hear it.
(Author of the post) My model is that Sam had some initial tendencies for reckless behavior and bullet-biting, and those were then greatly exacerbated via evaporative cooling dynamics at FTX.
Relatedly, this kind of evaporative cooling is exactly the dynamic I was trying to point to in my post. Quotes:
Sorry if I sounded redundant. I’d always thought of “evaporative cooling of group beliefs” like “we start with a group with similar values/goals/beliefs; the least extreme members gradually get disengaged and leave; which cascades into a more extreme average that leads to others leaving”―very analogous to evaporation. I might’ve misunderstood, but SBF seemed to break the analogy by consistently being the most extreme, and actively and personally pushing others away (if, at times, accidentally). Edit: So… arguably one can still apply the evaporative cooling concept to FTX, but I don’t see it as an explanation of SBF himself.
What do you mean by “(Author of the post)”
I am the author of the linked post that DPiepgrass was commenting on: https://www.lesswrong.com/posts/HCAyiuZe9wz8tG6EF/my-tentative-best-guess-on-how-eas-and-rationalists
He meant that he wrote the linked post on hypotheses for how EAs and rationalists sometimes go crazy.
I thought that Sam Trabucco was not EA, but rather someone that SBF knew from math camp and MIT.
This seems right, thanks. I don’t think we have positive evidence that Trabucco was not EA, though.
Wang pled guilty to serious crimes including wire fraud, conspiracy to commit securities fraud, and conspiracy to commit commodities fraud
(can’t link plea PDF on mobile) [Edit: https://fm.cnbc.com/applications/cnbc.com/resources/editorialfiles/2022/12/21/1671676058536-Gary_Wang_Plea_Agreement.pdf ]
On the Sam Harris podcast MacAskill and Harris seem to think it plausible that most of the 8bn$ of losses came in Summer 2022 - that Alameda was accepting funding on behalf of FTX, was meant to transfer it but didn’t. To me this seems too generous to FTX. Does anyone know?
The framing sounds too generous, since I do not think there was any plan to transfer it at any point. But I can see a grain of truth here. As covered in OP, SBF seemed to think Alameda’s balance sheet by October 2022 was very roughly:
$18bn of illiquid assets
$6bn of liquid assets
-$15bn of customer liabilities
Then the customers started to withdraw, $5bn in liquid assets was returned, then they more or less ran out and declared bankruptcy.
So what might this have looked like before returning money to crypto lenders in June 2022? I did not find any concrete figure for what the size of those transfers was, but I’ve generally assumed around $10bn. If so then the pre-return state was:
$18bn of illiquid assets
$16bn of liquid assets
-$15bn of customer liabilities
-$10bn of crypto lender liabilities
So on this hypothetical, at this point Alameda/FTX have liquid assets exceeding customer liabilities; in a crisis of confidence they can meet a full bank run from customers, though of course at cost of still going bankrupt and blowing up the lenders.
To be clear, this is all very rough and speculative. But I can readily imagine that Alameda had sufficient liquid assets to cover customer liabilities before they repaid their other major funding source. Of course the problems of the commingled assets, investing of customer assets, lying to everyone, and various other crimes would remain.
There are some assumptions that go into what counts as “liquid”, and what valuation your assets have, that may be relevant here. One big thing that I think happened is that FTX / Alameda were holding a lot of FTT (and other similar assets), whose value was sharply correlated with perceived health of FTX, meaning that while assets may have appeared to exceed liabilities, in the event of an actual bank run, some large fraction of the assets just evaporate and you’re very predictably underwater. So just looking at naive dollar valuations isn’t sufficient here.
(Not confident how big of an issue this is or how much your numbers already took it into account)
I could have been clearer about what is being counted as what, but such FTX-related assets are all counted as illiquid in this categorisation / hypothetical. I agree that assets appearing to exceed liabilities in itself doesn’t necessarily mean much, was covered in OP in the first section.
All I’m counting as liquid here is:
Roughly $1bn of the final SBF balance sheet
Mostly looking at $200m ‘USD in ledger prime’, $500m ‘locked USDT’, and $500m of HOOD shares.
The $5bn returned to customers during the bank run
Since this was successfully returned, it’s almost liquid-by-definition.
I would assume this was overwhelmingly USD / stablecoins / BTC / ETH, since those collectively made up almost all of the final liabilities (SBF balance sheet over on top left)
The $???bn returned to the lenders in June 2022
I speculated $10bn in prior comment, but again this is very much just a guess.
Anyway, it’s hard to put much weight on any of this because so much is uncertain, including the accuracy of that balance sheet.
My sense is they weren’t able to track their finances. Would you agree with that? Is there evidence I can look at for that?
Caroline claims she was able to track their finances well enough to (a) establish that they couldn’t afford to buy out Binance and (b) calculate a -$2.7bn NAV-excluding SamCoins for Alameda and recommend against $3bn of venture investments, both in 2021. I gave some links for that in OP. Then they calculated out how to repay lenders in June 2022, creating the spreadsheet that was central to the eventual guilty findings. So I don’t think they were completely clueless when it came to 10 figure numbers or the big picture more generally.
I suppose I consider it almost a given that they did not have good financial controls on ‘mere’ 7-8 figure sums, because at this scale and complexity that would be a full time job for a professional, probably multiple professionals. Per Going Infinite, VCs knew this and tried to push Sam to hire a CFO; he mocked them as quoted in OP and refused. So in the end all four people permitted to know the true state of things had substantial other responsibilities and no accounting background.
What does it mean for “losses” to have “come” (or not) at a specific point in time in this context?
I can accept that depositor losses had not “come” if FTX would have been able to pay depositors the fiat or crypto in their accounts in a matter of a few days. I specify that time period because depositors should have understood that a few days’ delay might be necessary for non-fraud reasons like technical failure.
Beyond that, I’d probably say the loss had come but that there were prospects for mitigation. If a thief takes my car, I’d normally say the loss occurred when I was deprived of my right (here, of possession). That I might get the car back from the person to whom the thief sold, or cash compensation from the thief’s bank account, does not negate the timing of that loss.
I believe the theory is that Alamada had accepted money on behalf of FTX and FTX thought that they’d transferred it but they hadn’t. And in the summer Alameda lost it. Honestly even writing that it looks like fraud, since they should have transferred it immediately.
Even Alameda accepting money for FTX at all was probably bank fraud, even if they had transferred it immediately, because they told the banks that the accounts would not be used for that (there’s a section in the OP about this).
See also this AML / KYC explainer, which I admit I have not read all of but seems pretty good. In particular:
(I found out about this explainer because Matt Levine at Bloomberg linked to it; a lot of what I know about financial crime in the US I learned from his Money Stuff column)
Executive summary: Sam Bankman-Fried and his associates at FTX and Alameda committed multiple crimes, including misappropriating customer funds, lying to lenders and investors, and bribing a Chinese official.
Key points:
Alameda was allowed to borrow FTX customer funds without sufficient collateral, creating a multi-billion dollar hole.
Alameda falsified balance sheets provided to lenders to conceal the extent of their liabilities and loans from FTX.
FTX lied to investors about Alameda receiving special treatment and privileges on the exchange.
Alameda lied to banks about the purpose of accounts used to process FTX customer deposits and withdrawals.
SBF directed a $140 million bribe to unfreeze Alameda trading accounts in China, violating anti-bribery laws.
Evidence suggests SBF was aware of the misuse of customer funds long before June 2022, contrary to some claims.
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.
Thanks for writing this.
I am confused why people are defensive of @Sam Bankman-Fried. I am fond of him as person and he was gracious to me personally. I even checked up on him after the crash. But that doesn’t change the fact he did a massive crime.
It doesn’t seem hard to say that I want Sam to be well as a person (and Caroline, Nishad, Gary and anyone else close to them) whilst also saying this was a huge and deliberate fraud. And I don’t even think we need to have discussions about utilitarianism. Why trade so sloppily? Why hide it for such a long time from anyone who could have given better advice[1]. I don’t get the temptation to say ‘Sam was trying his best’.
unless many here are lying about not knowing, which I doubt