Lessons learnt from FTX

Posted by

The story that is emerging about the collapse of Sam Bankman-Fried’s crypto exchange, FTX, and its affiliated trading firm, Alameda Research, is that FTX took customer money for crypto trades, and Alameda used its own money for crypto trades, and nobody kept track of the money, so Alameda kept using FTX customer money for trades, and FTX didn’t notice, and Alameda kept losing that money, and it also didn’t notice? And then one day someone noticed and the whole thing vanished in a puff of smoke? I don’t know! That story makes no sense to me at all! And I don’t think that it is a good story or anything; in many ways an orderly fraud would be better than a vast money-sucking whirlwind of chaos. But “nobody kept track of the money” does seem to be the story that everyone is going with.

Here is a Wall Street Journal story about the early days of Alameda:

Mr. Bankman-Fried placed huge bets on crypto assets but paid little heed to the risk of those bets, brushing off the staffers’ concerns, according to people familiar with the matter. The firm commingled trading capital with operating cash and had poor record-keeping that left its profits and losses unclear, they said.

“He didn’t want to feel constrained,” said Naia Bouscal, a former software engineer at Alameda who left with [Alameda co-founder Tara] Mac Aulay and the others. “But as a result we ended up not really knowing how much money we even had.”

They left in 2018. “Mr. Bankm

an-Fried told The Wall Street Journal … that Alameda addressed the accounting, risk and other issues they raised,” which makes it sound like Alameda did know how much money it had for the last four years, but in fact Bankman-Fried has been saying on Twitter and in interviews that Alameda and FTX did not keep particularly careful track of their money, so who knows. The Journal goes on:

Mr. Bankman-Fried had worked at quantitative trading giant Jane Street Capital LLC, a firm that has extensive risk controls. But when colleagues at Alameda proposed setting up rigorous structures and systems for risk, compliance and accounting, Mr. Bankman-Fried was dismissive of the idea, according to the people.

He said such extensive controls could crimp Alameda’s activity and limit how fast the firm could move to place trades, the people said, reducing potential profit.

Alameda built a trading algorithm to make a large number of automated, rapid-fire trades, but some at the firm feared it couldn’t keep track of all the activity, the people said. The staffers said Alameda needed to do a better job tracking its trading to properly account for its gains and losses, but Mr. Bankman-Fried rejected the idea, according to the people.

One thing about working at Jane Street is that, when you show up for work at Jane Street as a trader, your job is not, like, “rebuild Jane Street using only materials found in nature.” Your job at Jane Street is more like “we have built a good technology stack and risk-management and accounting and performance attribution systems, now go find some profitable trades.” You find some trades that you think are profitable, you do them, and if you’re right then Jane Street’s existing systems will notice and reward you. If you’re wrong, the risk systems will stop you. You are just in charge of finding the good trades.

If, later, you leave to start your own firm, you will take with you your skill at finding good trades. (Which seems like a perishable thing.) But it’s not obvious that you will take with you Jane Street’s risk-management culture, and you certainly won’t take its risk-management systems. And then when your colleagues say to you “hey, let’s set up some risk systems and hire an accountant, like we had at Jane Street,” you can reply “nah, I’d prefer to ignore that stuff, like I did at Jane Street.” You’re both right! Someone else at Jane Street set up those systems so you didn’t have to worry about them. But now you’re in charge, and if you don’t worry about them no one will, and then you end up not knowing how much money you have, or how much you’re making or losing, or how much you owe your customers.

Meanwhile here is a Financial Times story about the lavish later days of FTX:

A lack of internal controls that are typical of large financial companies meant FTX’s spending went largely unchecked, according to former employees and filings in the group’s Delaware bankruptcy case.

“[It was] kids leading kids,” said one former employee. “The entire operation was idiotically inefficient, but equally mesmerising,” they added. “I had never witnessed so much money in my life. I don’t think anybody had, including SBF.” …

The perks enjoyed by employees of the now-collapsed exchange included round-the-clock catering in the Bahamas office, “in addition to the free groceries, barbershop pop-up, and bi-weekly massages”, according to one employee.

FTX also provided Bahamas staff with a “full suite of cars and gas covered for all employees [and] unlimited, full expense covered trips to any office globally”, the employee added. Staff at FTX US, its separate arm for the American market, were allowed $200 a day in DoorDash food delivery credits.

Alameda Research, a crypto hedge fund founded by Bankman-Fried, also owes $55,319 to the Margaritaville Beach Resort in Nassau, which was founded by US musician Jimmy Buffett, according to bankruptcy filings this week. A “Who’s To Blame” margarita at one of the resort’s bars costs $13.

I think that this is best understood as an accounting problem. A month ago, a common perception of FTX was that it had “one of the highest revenue/profit/valuation per employee as any company in the world.” As far as I can tell, FTX believed this. If in fact your employees are incredibly good at making money, then lavishing them with perks is simply optimal. Allowing $200 a day in DoorDash to prevent your insanely profitable employees from ever having to think about their meal budget is a good business move. Your employees’ time and attention is so valuable that it is foolish to make them waste time filling out expense reports.

But a big part of the reason that FTX believed it was so profitable is that nobody made any effort to keep track of the money. If you lose $4 billion trading and borrow $8 billion of customer money to cover the gap and don’t keep track of that borrowing, then it looks like you made $4 billion trading, and you will say things like “we are incredibly profitable per employee, $200 DoorDash for everyone!” FTX spent like the company it thought it was, but it didn’t bother to find out what kind of company it actually was.