Almost All Stablecoins Are Stupid

I have been meaning to write down these thoughts I have conveyed to many colleagues over the years on the topic of cryptocurrency stablecoins.

The most recent event that has sparked my interest is the death of Basis Coin, which many intelligent observers did predict pretty easily (like

I won’t bother going into the history of stablecoins or taxonomy, BitMEX’s always excellent Research Desk has done a great piece that does this just fine:

Instead I would like to make a more concise point: almost all stablecoins are stupid. They vary from stupid, kind of stupid to “not stupid but illegal or unreliable”.

Below I will walk through why the above statement is true, and will for ease of use define “stability” as represented by the currency of the United States empire “US Dollar (USD).” Taxation of the world’s biggest economy can only be paid in this currency, so its demand is important and their banking system and government invests heavily in weapons to defend this value.

Without furthera do, why almost all stablecoins are stupid:

Bank Balance-backed Tokens: Stupid

Gemini Dollar, Tether, Paxos, Circle Coin all have variations of the same model: investing your USD into deposits varying in returns (a simple 300 basis points over at the Fed discount window, or even more if you have a higher risk appetite like Circle with their fixed income securities they invest your USD in

The way these models are best described by their proprietors is “PayPal on the Blockchain”.

Okay, nice, you take my USD, give me 0-interest tokens, and make a little margin on it while I have utility in using this token to acquire other crypto? Why don’t I just send my USD directly to the exchange to acquire crypto? Compliant stablecoins will have to KYC everyone who holds USD tokens of any sort.

Since all the projects listed above are compliant and regulated in some form (FinCEN or whatever) — eventually they will reach a point where the regulators understand what is going on and they will force the projects to only allow token transfers to whitelisted addresses.

Currently there is utility in these tokens because they can be used to move USD around no-KYC. Sure, you have to KYC when you create tokens directly with Gemini, Circle, TrustUSD. But once I have those tokens I can go send them to ISIS, North Korea, whatever I want. And do the banks and regulators really understand that? Of course not.

But, in practice, these tokens all have the ability to RETROACTIVELY censor transactions by blocking money. Tether has done this multiple times, and all the other Tether-clones have the same ability to just blacklist after the fact.

However, that does not change the fact that all of these tokens can be moved around on the blockchain to anyone, and no regulator or the underlying bank has a single clue who the recipients are. KYC on creation and redemption, along with the ability to censor transactions ex post facto is not remotely close to the compliance requirements in practice for USD custodians.

Once people realise there is no point to these things if they require KYC whitelisted transfers only they will ask themselves: why not use some 24x7 settlement bank pseudo prime broker like Silvergate, Noble (RIP), Signature, etc.?

Sure, you can create a walled garden as an exchange and KYC all incoming GUSD deposits, and KYC all outgoing GUSD withdrawals — that would be sufficient for an exchange to cover their ass. But the minute you are out of the garden, those GUSD are just like any ERC20 token that can be used on dark markets or gambling or funding America’s enemies.

Once regulators and banks realise this, they will cripple these stablecoins and people will go to more efficient centralised systems (if you are KYCing-doxing yourself why the hell not just go fully centralised anyway?)

Noncollateralised Algorithmic Shitcoins: Stupid

Think about it: investing in a project to produce a coin that has a stable value. Amazing idea right? What?! Any financial engineer can throw together some goofy model that encourages people to pay into a system that just tracks some measure for value stability.

Whether it is Basis or Seignorage or whatever, making a whole project out of creating an asset that is simply stable in fiat terms is flat out fucking dumb.

If you want stability in form of crypto then just collateralise the model instead of creating complex risk systems that are fun maths projects but not much more than that.

Gangsta USD Stablecoins: Stupid, Illegal, and Risky

A hypothetical unregulated gangsta USD stablecoin would very useful: bring USD banking to unbankable entities. Imagine that you can no-KYC hold USD and move it around in a censorship resistant manner on some major chain.

But guess what: no bank will allow that. That is why the unbanked entity is unbanked, why do you think you can just bank them with your tokens that themselves have a bank behind them?

So you get the worst of both worlds because you will end up with depeg risk from liquidity or solvency and centralised state/bank system risk. These just go the way of Liberty Reserve: fun and useful for a while, but inevitably rekt.

Well-collateralised Algo Coins: Not stupid, but inefficient and risky

DAI is a project worth giving credit to here (as well as bitShares bitUSD and others). They have basically leaned on a major asset (ETH) that is liquid and has reliable liquidity and designed an inefficient derivatives style contract that allows DAI holders to hold USD value in the form of ETH.

So if ETH-USD goes down in price, your DAI maintains USD value in ETH.

Sound familiar?

Yea, it is like holding a short position in ETH-USD in an inveres futures contract using ETH as collateral. So, if you want DAI for the purpose of stability in USD terms holding ETH, then why not go hit a 24x7 futures orderbook market and go short with ETH collateral and just make more ETH as price drops?

Furthermore, DAI creates a system that is susceptible to shocks. There is a risk that there is not enough collateral to pay out for those who are on the other end of DAI on a big move. This means the system could theoretically depeg and blow up which makes DAI not elegant in a mathematical sense to ensure true stability (although its performance has been quite well tested in the past year, so kudos).

However, it is quite useful, as long as ETH survives and maintains a proper USD quoted market that has reliable data sources to value the collateral. The DAI are ERC20 tokens themselves so that opens up opportunities for having USD value stored in a legit censorship resistant manner (to the extent you see Ethereum’s chain as truly immutable).

So this is probably the only stablecoin approach that is not stupid because it:

  • Legitimately maintains the censorship resistant form of stability
  • Does not rely on the banking system or state being involved in any way: it just relies on crypto (ETH in case of DAI) to thrive enough to have liquidity and value

Even though it is useful, it is less efficient than being in futures and prone to instability when not collateralised adequately.


In summary: crypto has value because it is censorship resistant. It allows for value transfer and running of decentralised applications in a structure that is robust to state and well-funded actors attacking.

As soon as you try to jam “stability” into this, you are going to be compromising on these underlying value drivers. Once you do that, you may as well be centralised. Run a centralised exchange where you hold USD deposits.

PayPal used to be no-KYC, you could send USD to people by e-mail address! That did not last long. And once banks and regulators understand what GeminiUSD, USDC, Paxos, TUSD, etc. are doing, they will be forced to KYC every recipient of token transfers that are carrying the US empire’s almighty stamp “USD” in their names.