Stablecoins are doomed to fail, Part II: MakerDAO’s “DAI” stablecoin is breaking, as predicted


This is a follow-up from an earlier post, “Stablecoins are doomed to fail,” from December 10th. I’ll keep this one short.

Long have I been a critic of the “stablecoin” concept – the techno-magical idea that a cryptocurrency can tell the market what its price should be, rather than the market determining what a cryptocurrency’s price should be (the usual way these things work).

My first encounter with stablecoins was with the Bitshares/BitUSD scheme in 2014. I pilloried it. Following which it fell on its face 100 hours after launch.

In October, I moved on to point out how harebrained a new scheme called Basecoin was. This scheme is similar to Bitshares, only with different names and slightly different functionality for the various moving parts. Incredibly, this profoundly ill-considered project attracted investment from A16Z, a smattering of new cryptocurrency-only VCs and Bain Capital Ventures.

Then, last month, after a developer from the “MakerDAO” team told a reporter that the fact I hadn’t commented on their “Dai” stablecoin project “spoke volumes” about its legitimacy and quality, I wrote a brief post in the space of 20 minutes pointing out that I thought “Dai”, too, was a bunch of twaddle.

The nice thing about predicting the demise of a stablecoin is that, in any free (i.e. unmanipulated) market, you are 100% guaranteed to be right. You just need to wait for your day to arrive.

Today is that day.

1. What I said about MakerDAO

Four weeks ago, I wrote:

Having taken fifteen minutes to review the MakerDAO paper, the Dai system is at its core a very simple cryptocurrency-collateralised derivative contract, with a lot of intermediate steps to confuse its buyers of the facts that (a) that contract is massively overcollateralized in the underlying cryptocurrency (which is Ethereum by default) and (b) in the event of an Ethereum black swan event the value of the underlying collateral, and therefore the value of the stablecoin, will also be wiped out.


Speaking generally, the system requires someone who wishes to obtain $100 worth of Dai to post, say, $150 Ethers’ worth of collateral. This, of course, is insane, because it would be easier for the user to simply go to Coinbase and sell his Ether for actual dollars, and he’d have $50 worth of Eth left over to go spend on other things.


The system also assumes that overcollateralising will protect the value of the Dai. Not so; it simply increases a Dai holder’s exposure to the price of the underlying Ether. If Ether gets wiped out, the Dai collateral will be worthless, so the user will have lost $150 in an effort to create $100.

2. What MakerDAO said about me

Cryptocurrency projects don’t like being criticized.

Well, nobody likes being criticized. But cryptocurrency projects in particular, with their communities of thousands of financially invested scheme participants eager for the value of their investment to rise, are known to dislike it a great deal. A sample:

The problem with Preston’s brash, hyperbolic, stance is that it favors showmanship over facts. While it may be great for increasing clicks he has backed himself into a corner of never being able to admit he groks it even when he does.


Saying something will eventually fail is the safe position of never having to be wrong with the chance of maybe being right some day.

Or this:

 if you didn’t spend enough time to understand this point, you end up saying incorrect things like “[issuing DAI] is insane, because it would be easier for the user to simply go to Coinbase and sell his Ether for actual dollars, and he’d have $50 worth of Eth left over” – selling ether directly for fiat or DAI, and opening a CDP, expose you to completely different “risk profiles” (margin long vs closing a regular long position).


It’s not true that the system “only works if the price of Ether goes up”. The whitepaper specifically discusses what happens if the collateral starts losing value, with different “lines of defence” that need to be used depending roughly on how fast it loses value – liquidating CDPs, then minting MKR, then the foundation.

“Read the white paper, bro.” Cute.

3. What happens when reality strikes

This was the dream:

Screen Shot 2018-01-12 at 11.54.10 AM.png

“Always maintain.” Strong language there.

The language sets the standard to which this stablecoin project must be held. For Dai to live up to its promise, a person seeking to redeem 1 Dai for money or money’s worth should reasonably expect to be able to convert it for $1, every single time, without exception.

Even if the marketing were more nuanced, we would do well to remember that near-perfection is also the standard in the grown-up world of finance which many of these crypto projects aspire to supplant.

If a money market fund’s investment income is less than its expenses, it breaks the buck (something that has only happened once in the United States in the last 47 years). If a bank can’t satisfy depositor withdrawals, it collapses. If an SPV can’t redeem bonds that have matured, it defaults. If a company cannot satisfy its debts as they fall due, it is insolvent.

If a stablecoin can’t hold its peg, all the time, every time, it’s simply not a stable coin. It is a mere aspiration of stability, clothed in technobabble, that has broken the buck, and a stablecoin’s users – if stability is indeed what they seek – would be better off just withdrawing cash, depositing money in a FDIC-insured bank or making an investment in a money market fund.

For Dai, then, this is reality:


The markets are crashing and, as this marmot predicted, Dai is trading at $0.80 on the dollar. It dipped to as low as $0.72.

The coin launched on December 30th. It lasted a grand total of 12 days before breaking. This is better than Bitshares’ BitUSD, whose freshman attempt lasted just over four days. TBD: whether Basecoin can set a new record by making it two whole weeks.

Rather speaks for itself, doesn’t it.

Why this happened, we cannot say. The coin, my friend Rick Dudley points out, remained over-collateralized despite the fact that the value of the collateral dropped 12.5% overnight. This lends itself to the suggestion that not only are stablecoins super inefficient from a cost of capital perspective, but also that stablecoins are not “stable” at all.

Furthermore, this episode suggests that, like markets for every other security in human history, stablecoins are not closed systems and are therefore subject to the vagaries of supply and demand and contagion from the wider markets beyond. In this instance, one market circumstance which might have caused the seemingly irrational selloff could have been a Dai holder getting margin called with respect to some other, unrelated, exposure and needing to liquidate his or her Dai at a loss then and there in order to cover.

If we were talking about any other security traded by grown-ups, mind you, we would not need to have this discussion, as the point would be taken as read by everyone, including the interns. Yet this point does not appear to be something that the existing stablecoin projects, the VCs investing in them, or the media covering them have considered at all.

Mic drop

It may be that this “stablecoin” will recover when the market recovers, or when those with an interest in seeing the project succeed deploy Ether to ensure its price reaches the “correct level.” A return to dollar “parity” for Dai should not surprise us. Stablecoins work as long as the price of the underlying collateral rises, or as long as traders with a financial interest in the coins’ success have sufficient firepower to paint the tape or otherwise subsidize their differences of opinion with the market.

It is not reasonable to assume Ether’s price can  or will rise forever, or that the MakerDAO people have infinite financial resources. Eternal inflation works in cosmology. It does not work in economics.

The teachable moment here is the same as the teachable moment from Bitshares back in 2014. When you make a “coin” which is in form and substance a repackaged exposure to another underlying cryptocurrency, as Dai is simply repackaged Ether, and Basecoin is simply an abstraction of demand for “Base bonds,” and peg that exposure to some meatspace asset like an ounce of gold or a U.S. dollar, a sudden move against the underlying collateral – in this case, 12% – can trigger a sell-off that breaks that peg, and breaks it hard.

Which means that despite all the marketing budget and smooth pitchmen dedicated to this folly, if we’re being responsible and/or honest with ourselves, maybe we should think twice before we use “stable” and “coin” in the same document, let alone the same word.

At this juncture, all I’ve left to say is:


UPDATE: DAI has a bot problem

Remember what I wrote above?

It may be that this “stablecoin” will recover when the market recovers, or when those with an interest in seeing the project succeed deploy Ether to ensure its price reaches the “correct level.”

Welp, it turns out that was a pretty good guess: the only reason the DAI “stablecoin” was stable is because a bot was wash trading most of the volume in the markets back and forth around the $1.00 mark.


In case you’re wondering how much volume this exchange was pushing through at the time, on 10 January 2018 (when the bot went down and the peg was briefly lost) Bibox’s Eth/Dai market traded 1981 Eth at $1,200/Eth.


That’s a market doing $2.6 million of daily volume, but lose one bot – one – and volume collapses to $300.

Draw your own conclusions.


22 thoughts on “Stablecoins are doomed to fail, Part II: MakerDAO’s “DAI” stablecoin is breaking, as predicted”

  1. What is says on CMC entirely depends on the Bibox order book This one exchange has all the volume so if someone buys even a small amount of ETH at an inflated price on their low volume book, CMC suddenly says DAI is worth $0.9. For this exact reason, yesterday CMC said DAI was worth $0.77 for about 10 minutes.

    I wouldn’t trust the CMC value until more exchanges list DAI trading pairs.

    1. I see no reason why introducing more variables should make this system any more stable.

      Additionally, the price disruption occurred across a 12-hour period. That’s more than a “fat finger” error at work.

      In any event, the peg is also broken on *all four* exchanges where Dai is currently listed.

      1. Hi Preston, nice piece.

        I have two questions:
        Wouldn’t having multiple forms of collateral (not just PETH) promote stability?

        What about a mechanism where a short position was taken with some of the pooled assets to bet against volatility. If this was implemented, couldn’t this further improve things?

        The way I see it the value of the reclaimable asset should be locked to the value it had when it was offered as collateral.

        A fee could then be taken where the collateral appreciates, and used to pay relevent fees for shorting against volatility.

        Hope this makes sense!

      2. No, it would simply add more variables. If you want to limit volatility vis a vis, say, a dollar, the smart thing to do is to go out and buy a dollar rather than overcollateralising a cryptocurrency-denominated smart contract.

  2. Cherry-picking a single data point doesn’t make you point strong. Instead, it exposes you as a moron.
    You first need to research fundamentals: what is price and what stablecoins claim.
    BTW BitUSD worked flawlessly for 3 years, but let’s pick that one day when it wasn’t, makes so much sense…

    1. BitUSD got unilaterally shut down by its developers and now trades on very thin volume. I have little to no confidence that system is not the subject of market manipulation/painting the tape, all the more so when we consider that BitUSD is not actually used for anything apart from being a vanity project to justify Bitshares’ existence (which is not used for anything else).

  3. I’m just waiting for StableGeniusCoin™ to revolutionize the industry.

    But seriously, shouldn’t some sort of “stablecoin” be possible if you are able to arbitrarily modify supply increases of your coin (like block rewards)? Basically increase inflation when you get above $1 and decrease inflation when below $1. That would be my amateur attempt to stablecoin at least. Interested what issues would come with that.

  4. The actual flaw in this and the previous post is not the concept of a stable/pegged crypto coin, but instead the concept of a fiat stablecoin, one that cannot be redeemed for whatever commodity backing it. Your analysis of Basecoin makes it very clear why that cannot be built.

    Meanwhile, it would be highly useful to have a crypto coin or token that is stable against USD (or EUR or gold or oil), and it is possible to build such a token. However, all the simple ways to do that require a trusted entity who promises to trade the crypto for the commodity upon demand, and having a centralized, trusted entity then breaks the key ideas underlying cryptocurrency. If for example a FedCoin was convertible to a USD, operated by a bank (or collection of banks) then that FedCoin might just as well just be a checking account with routing, account number, and use the ACH network or Visa debit card network for transactions. Little is gained from adding a blockchain transferring USD tokens.

    That said, just as we all failed to see how proof of work plus a distributed ledger would together solve the double spend problem, there very well may be a solution out there that is a trustless cryptocurrency and which can stay stable against a USD or other commodity without a single, trusted entity. Perhaps even without convertibility.

    I suspect it will look something like Basecoin, with multiple, interacting tokens, just not the three the Basecoin inventors came up with. As you so clearly pointed out, their shares are flawed, their bonds are odd, and in total their coin is more likely headed to zero than 1 USD.

    I’ve a few ideas if you or any other reader want to help solve this problem. If only for a stablecoin, we’d have a global online payment system that is usable, rather than a growing pile of speculators.

  5. Dai seems to be doing quite well these days. Perhaps you were a little premature with the criticism?

    In any case, I do agree with your analysis. Take a look at this statement from the Maker white paper.

    “When​​ CDPs become​​ undercollateralized​​ due​​ to ​​market​​ crashes,​​the​​ MKR ​​supply ​​is ​​automatically​​ diluted and ​​sold​​off​​ in ​​order​​ to​​ raise​ ​enough​​ funds​​ to​​ recapitalize​​ the​​ system”

    This worries me. When the market crashes MKR is also likely to crash and when the smart contract dumps MKR, it is going to enter a death spiral for MKR, which in itself is likely to get people to liquidate their CDPs …etc.

    1. If by “doing well” you mean “price set by trading bots in the absence of any genuine market for the thing,” sure. One of MakerDAO’s founders even admitted this (see tweet linked at bottom of the post). For that reason I don’t regard Dai as “doing quite well” – I regard it as “being quite manipulated.”

      Agree re: MKR being hit by a market crash, although it doesn’t even need to be undercollateralized to break. We just need some traders who are hungry enough to exit that they’re willing to get out of Dai, even at a loss, just to get some cash back.

      1. Hi, I also read their whitepaper and I think third-party bots that buy DAI when it falls below 1$ are EXACTLY how the price was supposed to stay stable. These automated buyers are incentivized not just by the trust that the price will move back (as is the case with Tether in its current state) but also by an interest rate on holding DAI that gets higher the farther the price moves away from 1$. That is not “manipulation” but people acting according to their incentives in line with how the system was designed, right?
        Btw, I have no stake in MKR whatsoever and are also not convinced that posting 150$ in ETH to create 100$ DAI makes any sense.

      2. The market is supposed to do that on its own; if you deploy your own bots who act not in their own interest but in order to burnish the appearance of the project’s effectiveness, that’s not “people acting according to their incentives.” It’s an artifice.

      3. That is the market! Bots are part of the market. They are incentivised to keep the price close to $1 as it is basically arbitrage. The DAI has been untroubled by recent ETH falls, as collateral holders are incentivised not to let their ETH/DAI ratio go below 150% to avoid punishments. So few people understand this properly it is a shame. This is a real decentralised stable coin which is set to get more robust. Keep watching.

      4. Bots are most if not all of the market, judging from what happened when the bots switched off. That’s not a market, that’s a Potemkin Village, a front.

Comments are closed.