[Action Required] State of the Peg

LINK has actually been quite uncorrelated to the rest of the crypto market if you look at the long time frame chart of both LINK/BTC and LINK/ETH (though it also crashed on black thursday like all other cryptos). I’m a bit biased here as a LINK holder, but what I can say that there are many other LINK holders who would be equally as willing as me to mint DAI against their LINK to purchase more LINK.

I also think mTokens (mDAI, mUSDC) from DMM DAO would be another good collateral type to add as those tokens are overcollateralized by a basket of interest-generating real world assets which currently includes $8.5M in automobiles from the United States, but are expanding to more asset classes like aviation, construction, and real estate. These would be much more stable/uncorrelated to crypto as whole and is already ERC20 compliant with an in-built exchange rate, collateralization/valuation determined by chainlink oracles, and metatransaction support.

Interestingly, mDAI most of the time technically wouldn’t even need to be liquidated through an auction as there is a reserve of DAI held in the contract (in addition to the real world asset backing) so mDAI can be exchanged for the corresponding amount of DAI (same for mUSDC but then that USDC would need to be auctioned for DAI). This method would be subject to the liquidity of the reserve ratio set currently at 50% but liquidity is actually sitting at above 100% right now to attract more liquidity. If this isn’t the right path or there’s not enough in-built liquidity, then auctions can be performed as usual. Just as a disclaimer for transparency, I do work at DMM, but I do think there is real potential here and wanted to bring it up to the community.


On a personal note what has held us, as a medium-sized fund, back from minting DAI with ETH is the volatility of ETH. Black Thursday showed that you absolutely cannot get liquidated.

So what price floor do we see with full confidence for ETH in the short term? Maybe $70, but with some cascading liquidations currently lurking at $79, perhaps $50? Ok so set the liquidation price at $45 and the whole exercise is not worth our time anymore.

Long story short: we need uncorrelated, price-stable collateral.

I’m not too informed about PAXG, but it would at least be uncorrelated and price-stable.


Why is PAXG more complicated to add? (Serious question)

FYI, there are like 1000 messages in https://chat.makerdao.com/channel/governance-and-risk?msg=yp9n5ufWEHgNX7kn3 today discussing this forum thread.


I keep my liquidation < $60 for the most part except in rare cases and only for short times. Right now I am at like $45 with a collateralization of 450%.

I also believe - and will comment when I have time that we can do quite a bit of things to improve this system for both vault owners and DAI owners. But each time I toss out something it usually is a protocol change - and unless we have a DAI emergency seems to move in glacial time, vs. instant time when others see a change as critical (i.e. the auction halt mechanism add), we need a DAI minting halt, as well as collateral removal halt - which if we have all three that can happen instantly and only can be removed after GSM delay is a better way to halt the system than an ES imo.

My goal is to create a system where the ES for the most part is off the table and instant halts can be on the table. Then lets talk about liquidation insurance additional SF rate. Most of these people if they could have had 24hr grace here with a fee for the service would NOT have been liquidated - or at least would have had a guaranteed 24hrs to get their vaults in order (i.e. liquidate themselves avoiding the 13% liquidation fees).

I already have DSR-L not released as a full auction l iquidity backstop mechanism that can literally catch the system while it falls. But we literally have to halt the system when market prices against OSM are massively disjointed >20%. I have a growing list of ways to almost bullet proof this system and backstop it with OPD (other people’s DAI) and give people options but they all require contract changes that people tell me will take ‘years’ - well hell crypto years are like dog years - projects literally live and die in year time frames. I know there is a lot of the plate here of Maker - mostly because foundation wants to move aside - fast. Personally I think we need to think about our clients and to improve robustness of the system first before we grow the thing.

I was told once by a business person I valued. To have a successful business focus on doing your business better than anyone else and everything else will follow. There is one caveat before this - make bloody sure the market(s) you want to enter are large. There is no point in having 10 businesses to make $1M when 1 can make you $10M if you focus and do it right.



Hi all! Tom from Dragonfly Capital here. I strongly agree that Dai’s recent drift from the peg is a problem, not just for Dai users, but for the Maker ecosystem overall. The truth is that Maker does not exist in a vacuum, and people will shop Maker against other options for getting leverage on underlying collateral (mostly ETH, for now) and will shop Dai against other options for stablecoins, and I worry about competitiveness on both fronts right now. For Dai specifically, we’re (anecdotally) speaking with a few teams that are considering switching to USDC away from Dai, and we see this switch already happening through on-chain data. dYdX posted its first day with more USDC-WETH volume than DAI-WETH volume last week. Currencies, even fully-backed ones like Dai, still rely on network effects and strong narratives, which can often unwind just as quickly as they grew.

With this in mind, it’s important to think about the problem itself, which is that ~6-10MM more Dai needs to be minted in order to re-peg. So, where is this going to come from? Problematically, overall interest in ETH leverage is down substantially since a month ago, almost 50% lower by Skew estimates. Screen Shot 2020-04-18 at 2.59.09 PM Thus, trying to win the ETH leverage market is already going to be a less fruitful way to get this Dai minted than it would have been in March. Not to mention that with the SF at 0.5%, Maker is already the cheapest way to borrow Dai and get ETH leverage (looking purely at interest rates) from all the decentralized lending protocols. Because the ETH leverage market has shrunk and Maker is already winning this market on pricing, it’s unclear to me that dropping SF even more will have a meaningful effect.

We’re open to proposals on new collateral – I think many of the candidates (LINK, PAXG, etc.) seem reasonable and I appreciate the Risk team’s work to vet them – but it’s also essential to keep our goal of minting 6-10MM more Dai in the next week top of mind. In this light, assets that are too small to generate meaningful Dai demand and assets that are too complicated to add in such a short time frame seem like they should be held for future discussion.


LINK is promising but can bring some troubles, it s surely uncorrelated to the main cryptos, but it s owned by a very small group of people… and it can be wildly volatile. It is a risky solution…but we need a solution quickly.


Thanks for sharing this tom. I completely agree that if we are going to do something it has to be of substance and reasonably quickly.

I see only option here with least questions is USDC which I discuss in above post. I’d like to see the DAI come out in 5M tranches but not against starting with the 10M DC limit and SF drop to .5%. I DO NOT want to drop the LR as I am sure enough USDC will come to mint DAI and want to keep this LR sane against turning on OSM and liqudations.


While I support a lower LR for USDC in the short run, I think it’s important to stress this point. No matter what LR is chosen, there has to be decent room to drop it further once liquidations for USDC are turned on - otherwise it will immediately cause mass liquidations when it gets turned on for those that have generated the maximum amount


Per one of the Devs, with the small Time frame, it would take some engineering work for a new type of medianizer. I think the same goes for uniswap liquidity tokens. Also Ilan Gitter of Paxos stopped by Rocket Chat recently and he mention something about a fee with PAXG?.. You might know more about that

That’s a good point–I saw this piece written by ChainLinkGod which sort of tries to give you an idea of some of the owners:

"During this monopoly period, Binance steadily held between 110M to 125M LINK tokens (31% to 35% of the circulating supply), but sharply decreased to 100M LINK (28% of circulating supply) in June 2019 with the Coinbase listing. Afterwards, Binance’s holdings of LINK continued to drain outwards due to user withdraws over the next ten months down to where it is today at 70M LINK (20% of the circulating supply).

Coinbase’s holdings on the other hand have continued to grow from the initial 20M to now 70M LINK tokens (5% to 20% of the circulating supply). Binance and Coinbase together now account for over 40% of the circulating supply of LINK. LINK’s distribution across the two exchanges have just recently come to parity with each other with both exchanges holding 70M LINK. It should be noted however Coinbase does not make their wallet addresses public, so this information was aggregated by tracking particular wallets that exhibited identical robotic accumulation behavior which started at the same exact time as the Coinbase listing."

Turning on USDC liquidations anytime soon would definitely be counterproductive.

On the other hand, I’m against a USDC-A CR lower than 110%. As long as the sf is low, there will be demand to mint DAI with USDC collateral.

I agree that 10m DAI may need to be minted, but note that this is also a moving target. Shorting DAI using USDC vaults also adds demand for DAI. And there will be naturally more demand for DAI as we get closer to the peg. Also maker holders may not have the commitment to continuously raise the USDC debt ceiling to whet the appetite for DAI.

Maker holders should be committed to raising the debt ceiling for stablecoins further if needed. And onboarding additional stablecoins is important for risk and optics.

Stablecoin vaults can’t be the only solution to bring us back to the peg. The DAI supply shortfall must be met by minting from other sources. Adding new collateral that can be easily onboarded is key.

1 Like

Seems like Link holders have a special appreciation for Link that might make it attract more Dai minting at least in the medium term, which is all we need. I admit this is more anecdotal than data driven.

Michael from Curve here.
Probably the best emergency action here is, indeed, lowering stability fee. But not sure if it will be enough.

Here is some idea for a more long-term working solution:

  • Maker contract allows to mint unlimited uncollareralized DAI to a special trading contract T;
  • T can only dump that DAI for other stablecoins when DAI price is > 1. Or to deposit that DAI on Curve (which shifts DAI price down);
  • If DAI price is < 1, the only thing that contract is allowed to do is to buy that cheap DAI and burn;

The whole operation (since DAI is only allowed to be sold when expensive and bought when cheap) is profitable, and so doesn’t create the actual DAI being uncollateralized (although it may seem so on paper).

That said, contract T having unlimited (or time-limited) DAI minting powers is quite a serious thing for the system, so that couldn’t work as an emergency action.


I think the plan is to implement essentially the solution you propose except using the regular elements of the Maker protocol (i.e. vaults) and allow participants to profit from the arbitrage instead of the protocol.

Ah good. This’d be an “arbitrage in time” though (not immediate)

scott from DEX.AG

there is a very large long term problem for MakerDAO regarding the peg, CDP owners are more sophisticated than DAI holders.

To get a CDP you need to learn quite a bit about how ethereum works and then learn about how Maker works. To use DAI you need to understand almost nothing besides how to use metamask for tokens.

More sophisticated users lose their keys less often. So DAI wallets become dead quicker than CDP wallets. This mismatch is made quite a bit worse by the fact thatkeepers can close the CDPs held by dead wallets. As i have stated privately to various associated folks, this mismatch is a long term problem that will inevitably create as nasty DAI short squeeze if no solution is found.

A short term solution should be something like:

  1. Add $USDC as collateral.
  2. Set DAI pool cap at $10M
  3. Set USDC margin requirement to 1%
  4. Set USDC Stability Fee at 0%
  5. Set DSR at 2%
  6. Set ETH Stability Fee at 2%

A medium term solution is something like selling uncollateralized DAI on the market for 1.005, with a bid to buy this uncolleteralized DAI back at .995, vs USDC (or other stable coins).

The long term solution is the DAI depegs from dollars and runs its owning monetary policy based on stability of purchasing power vs a global inflation metric. the collateral in CDPs would become more like a foreign reserve, with the floating DAI being uncollateralized.

The long term solution requires nation-state size scale, so its prolly not that important to think about it too hard, as it is so far away.


Looks like Vault holder 3931 just came back for $2MM.


Earlier this week I was thinking about a medium term solution while dealing with some insomnia. Since insomnia ideas are the opposite of shower thoughts, it might be terrible with some obvious failing. What’s more, this idea probably wouldn’t work all that well unless places like DEX.AG integrated it. I’m still wrestling with it, and the full version is probably its own post if not its own MIP, but here’s the short version.

We create a kind-of flywheel module to the Maker Protocol that is similar to the migration contract in that it has exclusive access to a set of collaterals with ideal parameters; a dark vault. Sane DC limits, 0% stability fees, 100% collateral ratio, and no liquidations. To the outside world this module would behave like a simple market trade. Show up with ETH, BAT, or USDC, and get ETH, BAT, or USDC amount of DAI out, where DAI is minted at $1. Same in reverse, show up with DAI and get back ETH, BAT, or USDC again where DAI is burned at $1. The former service would be used when above peg, the later service used when below peg. The one additional constraint is that the amount of liquidity is rate limited by seconds, risk, peg deviation, and some desired time-box where we return to the peg. I’m waving my hands at this formula for now.

At the time, this seemed like a great idea:

  • For these time slices where DAI could be minted against collateral, we are no longer dependent on the fear/greed of market participants.
  • The right formula could have us returning to peg within that timebox.
  • More potent the further off-peg we get, and has very little liquidity when close to peg.
  • Solves the peg problem efficiently when DAI is over peg and collateral markets are bullish, and when DAI is under peg and markets are bearish. Both rare cases.
  • Felt cleaner than a TRFM solution or PID controller on rates, as we can side-step the rate and market incentivization problem.
|   t1   |   t2   |  ...  |  tn   |
| 12k DAI| 10k DAI|  ...  | 2k DAI|
|-------------1 week--------------|

Then, sweet sweet sleep came, and in the morning I started seeing all the failings:

I was trying to fit the solution too much to our existing problem. That is, it’s a reasonably elegant solution for a set of market conditions that are rare. Without going into detail on all the cases that could happen, it should be obvious that the underlying collateral is volatile and could leave those dark Vaults with either a surplus or debt.

The surplus case isn’t so hard to deal with as there are any number of things one could use the surplus for, burn MKR, buy DAI on the market, or just keep it around for a rainy day.

But the debt problem defeated me. The debt problem meant that the system would have unbacked DAI, and that we would probably need to accrue our losses to system debt, which could eventually result in flop auctions and dilute MKR holders. This might mean that the available liquidity should probably be proportional to system surplus, and since we burned that on Black Thursday, it doesn’t seem that helpful of a solution now.

Anyway, this is where I gave up on the idea, or at least pushed it to a more long-term consideration. I leave it here in case someone sees more failings or better yet, a way to make it work in the medium term. At very least, perhaps it will inspire other solutions. I don’t know, maybe there’s a way to integrate your idea for this failure case.

1 Like

I was thinking about this today as well. In the long term, I think an escheatment/abandonment procedure will be necessary. Operationally it could work like this:

DAI held in a wallet that has been inactive for “x” number of years becomes inactive, transforming from DAI to DIE. The DIE becomes redeemable for a basket of collateral assets based on the OSM price of the collateral pool at the time it becomes inactive (basically a personalized emergency shutdown). Concurrently, the system mints and auctions/sells new DAI in order to accumulate the collateral assets required to pay the claims on new DIE.

This would have the impact of recycling DAI from abandoned addresses and ensuring that lost keys/dust don’t precipitate a liquidity crisis. But I’m not sure if this would be possible technologically, and it would definitely pose unique risks as well.

1 Like