Streetwise Professor

A week ago Friday (10/10/25) there was a crypto crash. Not huge, but not small. Bitcoin:

Ether was also hit.

The proximate cause was the most recent Trump tariff spasm. But that was just the spark. The initial down move triggered classical self-reinforcing crash dynamics. And the crypto market structure is arguably uniquely vulnerable to such events.

As I’ve written for years and years, (a) tightly coupled systems are vulnerable to catastrophic collapse, and (b) a wide variety of factors produce tight coupling in these markets. The most important of these factors is marking to market and variation margin.

And this dynamic is particular fraught in crypto. Positions are marked to market at a very high frequency. Moreover, the resulting margin calls occur at the same frequency. That frequency is so high that margin breaches result in liquidations of positions (“auto liquidations”) because there is no time to replenish collateral. These liquidations exacerbate price movements, creating a positive feedback mechanism–and positive feedback is almost always a bad thing in financial markets.

But it gets better! Risky positions are sometimes collateralized with . . . risky collateral! Specifically so-called “stablecoins.” I say “so-called” because stability is aspirational rather than real, especially for a new variety of “stablecoins” implicated in the 10/10 event, like USDe.

Whereas traditional stablecoins like Tether are supposedly backed by high quality, low-risk dollar assets (and the recent “Genius Act” will reduce any uncertainty about that), USDe is “backed” by both dollar assets and–wait for it–crypto. So what is the lithium that keeps USDe “stable”? Hedging using perpetual futures.

No. Seriously. Yeah, because hedges are always perfect, right? Here’s the only perfect hedge I’m aware of:

So in reality the “stablecoin” has basis risk, the basis being the difference between the price of the crypto collateral and the perp futures.

Arbitrage is supposed to keep the basis small with little volatility. However, if you look at data, e.g., the spread between BTC and BTC perp futures on Coinbase, you will see that there are frequent and large basis movements, often as big as 1 percent.

Moreover, USDe is traded on multiple exchanges, and the prices can diverge. Again, arbitrage is supposed to keep those divergences small and of short duration, but the market moves faster than arbitrage: given the frequent revaluing of positions and collateral, even a very brief price divergence can trigger a self-reinforcing spiral before arbitrage can close the gap.

On 10/10, Coinbase used the price of USDe on its platform to market collateral to market. But that exchange is not the primary USDe platform, and is therefore relatively illiquid. So forced liquidations of positions on Coinbase due to the price move and marking to market led to liquidations of USDe collateral which pushed downward pressure on the Coinbase USDe price which led to more positions being undercollateralized which led to those positions being liquidated. And this all happened faster than the speed of arb. Rinse, wash, repeat.

Can you say wrong-way risk? I knew you could.

The industry cope is to say “well ackshually if Coinbase had used the USDe price on more liquid exchanges as its USDe ‘oracle’ this wouldn’t have happened.”

Maybe. But irrelevant. What happened on 10/10 was a “normal accident.” Normal accidents result from a sequence of events in a tightly coupled. complex system. Accidents are “normal” in such systems because there are many, many, many possible sequences that can trigger accidents. Meaning that even if this particular sequence is eliminated by referencing prices from the most liquid exchange, many, many, many other such sequences will remain.

The underlying “cause” of such events is the very nature of the system itself. The reason crypto is particularly susceptible is the tight coupling due to frequent marking to market and high leverage. Those factors remain.

The crypto system is acutely vulnerable due to another structural factor–the fragmentation of trading. As I’ve frequently written, crypto market structure is puzzling. The centripetal force of liquidity typically results in the concentration of trading on a single platform. Crypto, in contrast, is traded on multiple exchanges, with several have a fairly large market share.

This fragmentation increases the number of connections and therefore the number of possible paths by which a destabilizing shock can spread. And referencing prices from other exchanges to value positions or collateral on a given exchange creates more such interconnections.

In brief, a shock–a price shock, or an operational shock–on one exchange can impact other exchanges. These shocks can trigger the destabilizing auto liquidations.

“Too big to fail” is not the right description of the crypto market. Instead, “too interconnected and tightly coupled and complex to succeed” is more accurate.

It must also be noted that the factors that make the crypto market systemically fragile make it unduly susceptible to manipulation which can in turn trigger systemic crises.

One of the oldest strategies in the book is “gunning the stops.” The play being that if you force prices to hit stops order triggers, that will induce trades that move prices further in the same direction.

In crypto, the game is “gunning the liquidations.” The idea is the same. Trade like a whale to push the price in a particular direction. If the move is big enough, that will trigger liquidations that move the price further in that direction. And since some collateral is risky, there is another way to manipulate–bash the collateral.

Given the high leverage and high frequency marking to market, the necessary push isn’t that big–maybe a dolphin can do it. And as shown on 10/10, and as discussed above, once the price moves begin they can trigger a crash.

It is also disturbing that crypto exchanges can profit from liquidations. Conceivably this could provide an incentive for the exchanges to manipulate themselves. But even if that’s not the case, the fact that they can profit from destabilizing manipulations, and instability generally, means that they have a weaker incentive, and perhaps a positive disincentive, to prevent/deter manipulations or big price moves.

In sum, crypto market structure is crapto. It has all of the inherent vulnerabilities of markets, but on meth. Supposed “innovations”–like near continuous marking to market and unstablecoin collateral–exacerbate these vulnerabilities. As does the fragmentation.

All meaning that it is inevitable that you will see this movie again.

In my recent post on Don Wilson and GPU trading, I expressed skepticism at the prospects for such trading based primarily on market structure issues. Specifically, a commodity futures market can thrive if the underlying product has a highly intermediated value chain, but that vertical integration or long-term contractual relationships between producers and ultimate buyers is inimical to their success. Further, I surmised that transactions costs considerations would lead to long-term contracts and/or vertical integration dominating the GPU sector.

And indeed, that outcome is becoming clearer by the day. OpenAI (it of ChatGPT) has agreed to a massive chip purchase from AMD and has also bought a 10 percent stake in the chipmaker. OpenAI has also entered into a five-year deal with Oracle to buy “computing power.”

There is also increasing verticality in power supply arrangements. (Apparently the new industry buzzword is “circularity” but we are really talking about vertical relationships here, so I’ll stick with traditional IO speak). Example: AI firms building gas-fired plants in Texas. Musk’s xAI building a gigawatt scale plant in Mississippi to power a data center outside Memphis.

Microsoft has entered a 20 year deal with Constellation Energy to restart–believe it or not–a unit at Three Mile Island. And on and on.

It’s somewhat amusing to me. Years ago Paul Joskow wrote a seminal article about coal mine mouth power plants in which he showed that this co-location led to long term contracts or vertical integration. We are now seeing data center mouth power plants.

The economics are basically the same. It makes sense to locate something that consumes as much (or more) power than a generator can produce to economize on transmission costs and enhance reliability. However, this creates bilateral monopoly driven transactions costs issues that lead to vertical integration and long term contracting.

Going back to GPUs, there is no reason for colocation, but economies of scale on both sides of the market–huge consumers operating data centers, and a small number of GPU producers–create similar bilateral monopoly opportunism risks, which in turn are leading to the implementation of traditional responses to these risks–tight vertical connections between producers and consumers.

This market structure is inimical to the development of a thriving derivatives market. So I wouldn’t be surprised to see the launching of GPU futures contracts, but I fully expect them to sink faster than a Nork destroyer after launch. But unlike the Nork ship, GPU futures would not be refloated.

Odd Lots was in Chicago last week, and after interviewing Don Wilson they interviewed CME CEO Terry Duffy. The interview was posted today.

CME's Terry Duffy touts its partnership with FanDuel. What CME will offer is basically what bucket shops offered back in the day. The irony is that back then bucket shops were the betes noire of exchanges. https://t.co/Z34P89KDJZ

— streetwiseprof (@streetwiseprof) October 2, 2025

I’ll get to the FanDuel tie-up in a bit, but first I want to discuss something else that came up–perpetual futures, which are a hot topic now (even though hot topic is not punk rock). Duffy cast doubt on the feasibility of perpetual futures in most CME products–he mentioned ags specifically. I agree with his conclusion but not his reasoning.

Duffy argues that perpfuts can’t work for commodities because they don’t fit the legal definition of a futures contract, which involves pricing something for delivery at a later date. He also argues that it can’t work for deliverables.

Maybe perfuts check the legal box, maybe they don’t. Twirling my Oliver Wendell Holmes-esque mustache, I could come up with a clever legal argument that based on the way perpfuts work, they are pricing a commodity at the time of the next marking-to-market, say an hour from now. Or a minute from now. Which would be in the future, either way.

And that marking-to-market is the real reason commodity perpfuts won’t work. Take the case of Bitcoin. Bitcoin perpetuals involve periodic (usually hourly) “funding payments” based on the difference between the futures price and the reference spot price taken from an exchange like Coinbase or Bitstamp. Basically cash settlement on an hourly frequency. But the contract remains live after each settlement.

So basically, a perpfut it is a perpetually (automatically) rolling cash-settled one hour futures contract. The advantages of the future over trading the spot are (a) embedded leverage, often 100-to-1, and (b) the ability to short without borrowing the underlying. Perpfuts are like trading spot, but with leverage.

This works for things for which there is a liquid reference spot market. Bitcoin and some other cryptos have that.

You could do this with say the S&P 500 because there is a large, liquid, continuously traded, transparent cash/spot market. You could do it with currencies, and via its FX Spot+ (“Rolling spot”) contract, the CME (via its EBS subsidiary) is doing just that.

You can’t do it with most commodities because there are no liquid, continuously traded, transparent cash/spot markets. Not in grains. Not in oil or other energy products (with the possible exceptions of natural gas and power). Not likely in precious metals–there is RTC trading in gold, but it is bilateral and there is no transparent pricing. Maybe you could do it with LME non-ferrous metals, because of their unique “cash” pricing mechanism.

So it’s not legal obstacles or deliverability that preclude perpetual futures in most (if not all) physical commodities. It’s the lack of a reliable spot pricing mechanism that can tie the futures and cash/spot prices together as perps are intended to do.

This is basically a special case of why I have ridiculed proposals for cash settlement in commodities for over 30 years. It’s only that the problem is even more pronounced for perpfuts because you need to observe spot prices with metronomic regularity every day, not just around contract expirations.

Turning to CME and FanDuel, my X post basically says it all. It is hilarious to me that CME is offering “event contracts” that are indistinguishable from the kinds of bets one could make on bucket shops back in the bad old days. Duffy describes contracts like “will the price of gold be above X an hour from now?” (NB: if that’s a futures contract, as Duffy claims it is, then perpfuts are no different: see above). The contracts would be “digital” (i.e, have a fixed positive payoff if the event happens, zero payoff if it doesn’t).

Well, that’s exactly the kind of thing bucket shops traded back in the day. (0dte options, which are also a thing now on many exchanges, are also very similar to old-school bucket shot bets). Exactly like them.

That’s an 1892 painting of a bucketshop. Note the use of the high tech chalkboard app.

So why hilarious? Because exchanges waged war on bucket shops for decades in the late-19th and early-20th centuries. A journal article provides some color:

On the morning of August 29, 1887, Abner Wright, president of the Chicago Board of Trade, forcibly removed the instruments of the Postal Telegraph and the Baltimore and Ohio Telegraph companies from the floor of the exchange, literally throwing their equipment out of the building. A few months later, on the night of December 15, Wright discovered some mysterious electrical cables leading out of the basement of the exchange building. Thinking that they were telegraph lines, he ordered them cut with an axe. Instead, they were cables connecting the building to the police and fire departments.¹ His desire to sever the Board of Trade’s telegraph connections might seem surprising, since the telegraph network was indispensable to the operations of the major stock and commodity exchanges.

That’s why the CME’s current move is hilarious. If you can’t beat ’em, join ’em.

I’m not censorious of bucket shops per se. People gonna gamble. The CBOT’s (and NYSE’s and other exchanges’) ire reflected not a moral judgment, but a dislike of competition. (Yes you can make arguments like Mulherin, Netter and Overdahl that the shops free rode on the price discovery of exchanges, and that impaired liquidity. But as I’ve shown in my academic work, this “free riding” can actually enhance efficiency if exchanges have market power, which they almost certainly do).

Exchanges today almost certainly realize that they can’t drive bucket shops, excuse me, betting apps, out of business: Terry Duffy can’t forcibly remove their equipment, a la his distant predecessor Abner Wright. So the best alternative is to get in the game with them.

Perfectly reasonable commercial strategy, but a deeply ironic one.