If you abstract away the financial details there’s also a question of like, what your utility function is. Is it infinitely good to do double-or-nothing coin flips forever? Well, sort of, because your upside is unbounded and your downside is bounded at your entire net worth. But most people don’t do this, because their utility is more like a function of their log wealth or something and they really don’t want to lose all of their money. (Of course those people are lame and not EAs; this blog endorses double-or-nothing coin flips and high leverage.)
Alameda CEO Caroline Ellison, writing on her tumblr, February 2021, my emphases
One of the minor mysteries of the FTX saga for me, is how they could possibly have this attitude. She writes as if something about EA actually justifies this notorious “martingale” betting strategy. But the standard opinion about martingales, as far as I know, is that it is justified only if you have infinite funds. Did they assume that they could find new investors forever? Was there a particular level of wealth they were aiming for, after which they would start to behave more cautiously?
This it not the martingale betting strategy. In martingale, you double your bet after every loss so you could theoretically at most recover your entire stake (“throwing good money after bad”); whereas here in the St. Petersburg Paradox, you’re offered a betting situation where you can repeatedly bet and either double your winnings or lose them all. Here you could start with $1 and end up with a $1 googol, if you won often enough and were principled enough to stop at that point. It still suffers from a similar problem as martingale, but it’s not the same situation.
Regarding that Twitter thread, I only skimmed it, but it emphasizes the Kelly criterium which we IIRC discussed several times here on the forum, and IIRC the conclusion was that it depends on some assumptions (log utility of money—Wikipedia: “The Kelly bet size is found by maximizing the expected value of the logarithm of wealth”) that don’t necessarily apply in all real-life situations (?). That said, I probably remember this incorrectly; you might want to search LW for Kelly to find those essays.
This is not to dispute that if you have infinite appetite for risk, you’ll eventually lose it all.
I don’t think it’s quite a mystery as you think… if you do the same overleveraged high stakes bet and you come out winning ten out of ten times, it’s likely you will continue doing it (this is the nature of a ponzi scheme.) I think they could have easily find investors if they could have sustained the bank run and prevent the balance sheet leak.
The latter, I think, is the real mystery.
SBF probably knew at all times the dollar value size of a bank-run he could sustain until having to forcedly lock withdrawals, plus which whales could dump FTT at any given time (one of this being the CEO of Binance.)
While most people are thinking that Binance’s CEO Changpeng Zhao made a huge move dumping all of FTT, I think it’s terrible for him—it highlights the fragility of the entire enterprise and shifts all eyes on the remaining centralized exchanges (I imagine SBF’s line of thinking going something like this, he didn’t think that Zhao would martyr himself, but he did and thus we’re here)
Alameda CEO Caroline Ellison, writing on her tumblr, February 2021, my emphases
One of the minor mysteries of the FTX saga for me, is how they could possibly have this attitude. She writes as if something about EA actually justifies this notorious “martingale” betting strategy. But the standard opinion about martingales, as far as I know, is that it is justified only if you have infinite funds. Did they assume that they could find new investors forever? Was there a particular level of wealth they were aiming for, after which they would start to behave more cautiously?
edit: Some insight here.
Two points:
This it not the martingale betting strategy. In martingale, you double your bet after every loss so you could theoretically at most recover your entire stake (“throwing good money after bad”); whereas here in the St. Petersburg Paradox, you’re offered a betting situation where you can repeatedly bet and either double your winnings or lose them all. Here you could start with $1 and end up with a $1 googol, if you won often enough and were principled enough to stop at that point. It still suffers from a similar problem as martingale, but it’s not the same situation.
Regarding that Twitter thread, I only skimmed it, but it emphasizes the Kelly criterium which we IIRC discussed several times here on the forum, and IIRC the conclusion was that it depends on some assumptions (log utility of money—Wikipedia: “The Kelly bet size is found by maximizing the expected value of the logarithm of wealth”) that don’t necessarily apply in all real-life situations (?). That said, I probably remember this incorrectly; you might want to search LW for Kelly to find those essays.
This is not to dispute that if you have infinite appetite for risk, you’ll eventually lose it all.
I don’t think it’s quite a mystery as you think… if you do the same overleveraged high stakes bet and you come out winning ten out of ten times, it’s likely you will continue doing it (this is the nature of a ponzi scheme.) I think they could have easily find investors if they could have sustained the bank run and prevent the balance sheet leak.
The latter, I think, is the real mystery.
SBF probably knew at all times the dollar value size of a bank-run he could sustain until having to forcedly lock withdrawals, plus which whales could dump FTT at any given time (one of this being the CEO of Binance.)
While most people are thinking that Binance’s CEO Changpeng Zhao made a huge move dumping all of FTT, I think it’s terrible for him—it highlights the fragility of the entire enterprise and shifts all eyes on the remaining centralized exchanges (I imagine SBF’s line of thinking going something like this, he didn’t think that Zhao would martyr himself, but he did and thus we’re here)