[Discussion] ETH-B Vault

Hey all,

We talked about a lower liquidation ratio ETH-B vault on governance call about two weeks ago and I want to start a discussion here before we proceed to implementing it.

Our proposal to governance was to implement a lower liquidation ratio ETH-B Vault with initially low debt ceiling due to risks related to OSM attacks (minting unbacked DAI due to old ETH price system uses because of OSM configuration). OSM attacks can become severe if low LR Vaults would have a high amount of unutilized debt ceiling.

Fortunately or hopefully we should be able to expect quite some demand for usage of such vault types due to farm related DAI minting from current ETH Vaults. This means debt ceilings might get utilized quickly if we set them lower. It also means such vaults might be less risky from a liquidation risks standpoint, as most of the users might be farmers who are long DAI rather than ETH leverage traders that need to perform trades when unwinding their debt position. Note that the ultimate goal of ETH-B is to attract new capital and generate additional DAI.

Our indicative results for 130% LR for ETH-B Vault shows that risk premium would be twice as high or around 8%, compared to 4% calculated for ETH-A base case scenario. This calculation is based on higher ETH Vault debt exposures, which means that risk premium is likely much lower if we start with only 10m debt of debt ceiling for ETH-B. We also think 130% LR might be conservative enough that hedges from OSM attacks.

Assuming 8% risk premium, currently this represents SF of 8% if Base Rate stays at 0%. This rate would be though higher than the one on Aave (+6%) or Compound (+4% or even negative including COMP rewards).

Proposal from our side would be to have a sandbox approach, initially start with lower DC in the range of 10m-30m for a 130% LR ETH-B Vault type, test demand with higher SF in range between 4% to 8% and see what kind of Vault users we attract. As said, having long DAI Vaults is less risky for portfolio exposure. I also think this will be important test, to finally give us some feedback how sensitive ETH Vault users are on rates.

If the general strategy is sound we should soon proceed to Signal request and have a on-chain vote after next week.


Thanks @Primoz. Those numbers are within the range I was expecting and sound very reasonable. I’ll be happy to implement a signal request around this range after a bit more discussion.


Overcharging borrowers who might bring DAI to the peg in a decentralized way is not a good policy IMO. The risk evaluated by the risk team means nothing as we see in the USDC case (125%->101%).

We want to ease the liquidations policy (pausing them would be ideal) for ETH vaults and set the LR as low as possible with 0% SF.

We want or you want?

I´m against pausing liquidations.

Say ETH increases in value, no liquidations are enabled, people mint dai, eth drops in value, we have unbacked dai. Because we have unbacked dai we should mint mkr to remove the dai of circulation, maker´s price takes another dump or we can gamble it away on the presumption that eth will increase in value (I don´t know which one is worst, in any case none of them are good)

Agree on reducing the LR while increasing the SF to mitigate risks and evaluate user demand of the product at different interest rates. Perhaps 8% is over current market rates however there should be a premium for having instant liquidity to your collateral but let´s see if the market thinks similarly.


Of course you are. The majority (the MakerDAO governance) is responsible for the situation with the peg that we have for the last 6 months. They will say that they rescued DAI from being even more off the peg and that they grew DAI to 850M MCap.

We have to change the definition of unbacked DAI in order to fix the peg. DAI backed by ETH is better than DAI backed by USD in a bank (USDC). ETH will eventually rise in value, but USD in a bank will probably be eaten by negative interest rates, frozen by governments, embezzled…

I don’t know whether the “unbacked” DAI is the best solution, but I’m pretty sure it’s better than the ones we have now,

1 Like

It´s a trade off between having the peg 1% above or losing credibility with integrators and new collateral onboardings. Vote for the first one.

Ref your second point, if you don´t enable liquidations your dai is not backed by eth because you don´t have the elements to enforce it. It´s one way or the other or just gamble it away to future eth growth in price


I am wondering is there is a way to set up a vault with a fix liquidation price instead of percentage?

I am not too sure if that makes sense but I can see a market here.

For example 300 dollars for 1 eth, you got liquidated if it goes under 300 dollars.

Then the vault can’t be used until ETH goes over 300 again? This is getting a little off-topic.

having the peg 1-3% above = losing credibility with integrators

No, first frozen and then if it goes above 600 USD it may be held further or liquidated depending on the peg.

1 Like

either that or we take the eth as collateral.

  • This is getting a little off-topic.
    Well yes and no, because you can set it up with a LR at 100% but limited at 300 dollars by eth.
    Which is the aim of the topic decrease the LR but with a limited risk.

I mean liquidated but not sold. the eth is just lost for the user.
It is like an insurance for him with a fees attached and for us a way to increase our eth collateral or any others collateral.

I really do not like the idea of pausing liquidations. I think we have a number of controversial tools we could use that are better than this one. Any of: negative interest rates, lower LRs (all the way down to 100% even), minting unbacked dai and having the protocol buy other assets on the market, TRFM. Any of these ideas are significantly better than pausing liquidations.

I do like the idea of an ETH-B vault with a lower LR and higher RP than the ETH-A vault. I would probably set the RP to 6%, or even 5.5% or 5.75% to outcompete Aave. This will be a good test before liquidations v2. Once the auction system is improved I could even see us lowering these RPs to just below the competition (3.5% or so, I do not think we should compete with yield farming) and moving toward heavier use of the ETH-B vault.

Having lower LRs should improve the peg due to greater capital efficiency in dai minting.


We might need to think about raising the box from 15m. Currently we have 439M DAI minted from ETH-A and WBTC-A.

Assuming we’re limited by only 15m of capital willing to bid on auctions seems low.

1 Like

I am with @Aaron_Bartsch here the numbers look about right and if we want to expand somewhat by lowering the LR to 130-133 and a DC limit I think this is a good way to possibly bring more ETH over. At a minimum it would be an interesting experiment on multi-tiered vaults with different LR, RP, etc. to gauge market sentiment on these.

I think the parameters risk came up with look reasonable and as far as I am concerned we can start everything now at 10M because this is approaching 1% of outstanding and 2% of MKR cap which isn’t unreasonable.


Before implementing something like this, I’d love to see some proposals to solve the OSM delay risk. I.e. The risk of a vault being <100% collateralized during the OSM delay and minting Dai up to the debt ceiling for that asset (as you outline above). IMO a 23% drop in one hour in these volatile assets is just too likely to occur to have something like this without a super tight debt ceiling.

Are there any potential solutions to the OSM delay that could only prevent new minting and not impact liquidations? Maybe @NikKunkel has some input?


I support a small-scale trial of ETH-B, but the liquidation must be strictly enforced. I have always insisted that ETH is more risky than stablecoins, so it is impossible to suspend liquidation like stablecoins. We must have a good clearing system to ensure the clearing of ETH-B.


@g_dip does raise a relevant point ref OSM.

@Primoz the risk premium of 8% factors in the probability of having undercollateralized debt due to > 30% drops within an hour?


Unfortunately not, we would need to adapt the model for that and it is close to impossible, because another input needed would be utilization fo debt ceiling through time which is hard to predict. This is why lower DC is proposed. Keep in mind 20%+ drops for ETH within an hour are very very rare (I was able to find only one event on random hourly historical returns), but of course you never know. This is why 130% still seems conservative enough.


Seems like instant access modules could fix this problem since voting in new debt ceilings is not necessary and we could then be able to keep a tighter debt ceiling.

1 Like

130% vaults can be risky especially with Liquidations V1.0. 1 hour OSM, 6 hour liquidation delay, chance of the box getting filled.

I wonder if the Liquidation Fee should be higher.