Streetwise Professor

November 28, 2022

I Remain DeFiant: DeFi Is Not the Answer (to Price Discovery) in Crypto

The meltdown of FTX continues to spark controversy and commentary. A recent theme in this commentary is that the FTX disaster represents a failure of centralization that decentralized finance–DeFI–could correct. Examples include contributions by the very smart and knowledgeable Campbell Harvey of Duke, and an OpEd in today’s WSJ.

I agree that the failure of FTX demonstrates that the crypto business as it is, as opposed to how it is often portrayed, is highly centralized. But the FTX implosion does not demonstrate that centralization of crypto trading per se is fundamentally flawed: FTX is an example of centralization done the worst way, without any of the institutional and regulatory safeguards employed by exchanges like CME, Eurex, and ICE.

Indeed, for reasons I have laid out going back to 2018 at the latest, the crypto market was centralized for fundamental economic reasons, and it makes sense that centralization done right will prevail in crypto going forward.

The competitor for centralization advocated by Harvey and the WSJ OpEd and many others is “DeFi”–decentralized finance. This utilizes the nature of blockchain technology and smart contracts to facilitate crypto trading without centralized intermediaries like exchanges.

One of the exemplars of the DeFi argument is “automated market making” (“AMM”) of crypto. This article provides details, but the basic contours are easily described. Market participants contribute crypto to pools consisting of pairs of assets. For example, a pool may consist of Ether (ETH) and the stablecoin Tether (USDT). The relative price of the assets in the pool is determined by a formula, e.g., XETH*XUSDT=K, where K is a constant, XETH is the amount of ETH in the pool and XUSDT is the amount of Tether. If I contribute 1 unit of ETH to the pool, I am given K units of USDT, so the relative price of ETH (in terms of Tether) is K: the price of Tether (in terms of Ether) is 1/K.

Fine. But does this mechanism provide price discovery? Not directly, and not in the same way a centralized exchange like CME does for something like corn futures. DeFi/AMM essentially relies on an arbitrage mechanism to keep prices aligned across exchanges (like, FTX once up an time and Binance now) and other DeFi AMMs. If the price of Ether on one platform is K but the price on another is say .95K, I buy ETH on the latter platform and sell Ether on the former platform. (Just like Sam and Caroline supposedly did on Almeda!) This tends to drive prices across platforms towards equality.

But where does the price discovery take place? To what price do all the platforms converge? This mechanism equalizes prices across platforms, but in traditional financial markets (TradFi, for the consagneti!) price discovery tends to be a natural monopoly, or at least has strong natural monopoly tendencies. For example, it the days prior to RegNMS, virtually all price discovery in NYSE stocks occurred on the NYSE, even though it accounted only for about 75-80 percent of volume. Satellite markets used NYSE prices to set their own prices. (In the RegNMS market, the interconnected exchanges are the locus of price discovery.)

Why is this?: the centripetal forces of trading with private information. Something that Admati-Pfleiderer analyzed 30+ years ago, and I have shown in my research. Basically, informed traders profit most by trading where most uninformed traders trade, and the uninformed mitigate their losses to the informed by trading in the same place. These factors reinforce one another, leading to a consolidation of informed trading in a single market, and the consolidation of uninformed trading on the same market except to the extent that the uninformed can segment themselves by trading on platforms with mechanisms that make it costly for the informed to exploit their information, such as trade-at-settlement, dark pools, and block trading. (What constitutes “informed” in crypto is a whole other subject for another time.)

It is likely that the same mechanism is at work in crypto. Although trading consolidation is not as pronounced there as it is in other asset classes, crypto has become very concentrated, with Binance capturing around 75-80 percent of trading even before the FTX bankruptcy.

So theory and some evidence suggests that price discovery takes place on exchanges, and that DeFi platforms are satellite markets that rely on arbitrage directly or indirectly with exchanges to determine price. (This raises the question of whether the AMM mechanism is sufficiently costly for informed traders to insure that their users are effectively noise traders.)

The implication of this is that DeFi is not a close substitute for centralized trading of crypto. (I note that DeFi trading of stocks and currencies is essentially parasitical on price discovery performed elsewhere.) So just because SBF centralized crypto trading in the worst way doesn’t mean that decentralization is the answer–or will prevail in equilibrium as anything more than an ancillary trading mechanism suited for a specific clientele, and not be the primary locus of price discovery.

The future of crypto will therefore almost certainly involve a high degree of centralization–performed by adults, operating in a rigorous legal environment, unlike SBF/FTX. That’s where price discovery will occur. In my opinion, DeFi will play an ancillary role, just as off-exchange venues do today in equities, and did prior to RegNMS.

One last remark. One thing that many in the financial markets deplore is the fragmentation of trading in equities. It is allegedly highly inefficient. Dark pools, etc., have been heavily criticized.

Fragmentation and decentralization is also a criticism leveled against OTC derivatives markets–here it has been fingered as a source of systemic risk, and this criticism resulted in things like OTC clearing mandates and swap execution facility mandates.

It’s fair to say, therefore, that in financial market conventional wisdom, decentralization=bad.

But now, a failure of a particular centralized entity is leading people to tout the virtues of decentralization. Talk about strange new respect!

All of these criticisms are largely misguided. As I’ve written extensively in the past, fragmentation in TradFi is a way of accommodating the diverse needs of diverse market participants. And just because some hopped up pervs found that running a centralized “exchange” was actually a great way to steal money from those blinded by their BS doesn’t mean that centralization is inherently unfitted for crypto because decentralized mechanisms also exist.

If crypto trading is to survive, well-operated centralized platforms will play an outsized role, supplemented by decentralized ones. Crypto is not so unique that the economic forces that have shaped market structure in stocks and derivatives will not operate there.

So don’t overgeneralize from a likely (and hopefully!) extreme case driven by the madness of woke crowds.

Print Friendly, PDF & Email


  1. Thanks for another interesting article. It got me thinking there might be situations when satellite markets are desirable. If so, in what context do they provide value?

    Could you use a similar approach to the above to define the relative price of related commodities like nickel and ferro-nickel? This could then enable participants to fix the rate of exchange of Ni and FeNi (at a given place and time) and increase the utility (and hopefully liquidity) of the underlying contract (LME in this case) and enable the grade diff risk to be transferred without reducing the liquidity of the main contract.

    Comment by lundenwic — November 29, 2022 @ 8:51 am

  2. Wow, prof, that’s quite a dive into deep theory. The discussion at the end helped. For the record do you regard price discovery in stock trading on US markets as having proprietary, natural monopoly aspects? I guess not, but would like to hear the answer anyway.

    Also is “consagneti” a typo or a real word?

    Comment by Ty Kelly — November 29, 2022 @ 11:07 am

  3. The crypto world, to this outsider, seems to be following David Chapman’s modern classic about “scenes” that turn into “subcultures” that then get monetized and overrun by sociopaths.

    But… there is still a hard core that believes that Bitcoin is an incipient currency, and is into a) buy and hold while waiting for inflation to eventually erode the value of dollars and b) promoting the use of Bitcoin as a medium of exchange via the Lightning protocol and whatever gets built securely on top of that. This crowd is reveling in the collapse of FTX and the comeuppance of the latecomer wannabes who treat crypto as a casino for speculation.

    As for DeFi, exchange trading hardly seems the killer app. Investment pools and other such cooperation-by-strangers arrangements seem the area where something non-redundant is perhaps possible, though fraught with hazards.

    Comment by SRP — November 29, 2022 @ 12:14 pm

  4. Think he meant “cognoscenti” and “once upon a time” earlier. Maybe weird apple auto-correct. who knows. nice thoughts though

    Comment by PCH — November 29, 2022 @ 12:44 pm

  5. Centralized markets provide the best price…for those who can participate in centralized markets. The whole push for decentralization is to ensure access to markets (even if imperfect or inefficient) for the 80% of the world’s population who cannot participate in these markets. The argument is moral, not technical: the coders behind these projects seek decentralization (really permissionlessness via decentralization) as a human good, not a performance improvement.

    So yes, decentralization comes at a price. Whether that is worth it depends on one’s circumstances. For some, the choice is between imperfect DeFi markets, and even worse monopoly rent extraction in centralized markets, or no market at all.

    There are ways to discover price in a decentralized way. The Bisq network, for example, simply keeps an order book; if your ask meets another bid then the trade executes at the midpoint price. Not exactly a speed demon (a typical trade takes hours or even days) but it works. Some systems set up a Schelling game among stakeholders where block rewards go to the median guesser.

    Comment by M. Rad. — November 30, 2022 @ 11:12 pm

  6. “What constitutes “informed” in crypto is a whole other subject for another time.”

    Yep. The best life lesson I ever got was playing poker with a couple of kids from Toronto on a Jewish teenager trip to Israel*.
    They relieved me of 10 quid’s worth of shekels in as many minutes!
    If you can’t tell how the game is being fixed, it means you’re already marked down as the “mark”. Hahaha

    I can’t quite recollect where the game was played but it might well have been on the kibbutz where we were sent for ten days to work – unpaid. They did feed us – rubbish. Ugh! A second great life lesson for a 16-year-old. Even ostensibly nice people (kibbutzim) will exploit the naive and the gullible given half a chance!

    Comment by Simple Simon — December 4, 2022 @ 11:32 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress