[AIP-36] v2 Initial Guarded Launch and RFC for launch params

scoopy

Alchemix V2 Launch

Over the past year, the Alchemix Protocol has grown in many ways. Obviously, our TVL is sizable, with over $1B give or take sustained thus far. Our core team is larger, our community has blossomed, and subdao formation is in progress.

Alchemix DAO has had its governance tokens listed on several exchanges, engaged in numerous strategic actions to secure its long term future, and become increasingly integrated into DeFi. We’ve recovered from an unfortunate error with alETH and came back from it anyway, resulting in over $500m TVL of alETH at its height. It’s been a wild first year.

On the eve of our first birthday, we are now ready to deploy our version 2 of Alchemix. While the Alchemix Core team has worked relentlessly on v2, we are committed to becoming progressively more decentralized. In this context, it means ALCX governance must approve the v2 deployment, and the parameters within it.

Overview

There are 3 brand new components in the AlchemixV2 system:
AlchemistV2
TransmuterV2
TransmuterBuffer

The AlchemistV2 and TransmuterBuffer each have their own set of configuration parameters that need to be determined by governance. Similar to the launch of V1, the Alchemix core team has come up with initial configurations that are meant to be “guarded launch” configurations targeting security and reliability in the initial week(s) that the system is active. The intention is to loosen these parameters over time (raise caps, etc) to achieve more scalability and a better overall UX.

AlchemistV2

The new version of the Alchemist contract accepts deposits in the form of yield-tokens (eg. yDAI) and underlying tokens (eg. DAI). A shares-based accounting system (per yield-token) ensures that each user always owns a percentage of the total amount of any given yield-token held by the alchemist. This is similar to the way yearn vaults work in that each depositor maintains a percentage of the total collateral held in the vault.

For the alUSD system, the 3 proposed underlying-tokens are DAI, USDC, and USDT. These assets share a strong correlation across the greater DeFi landscape, and will provide a resilient, deeper, and more varied backing for alUSD.

The Alchemix core team proposes launching with the following values for the governance- defined, underlying-token parameters:

The 3 proposed yield-tokens are yDAI, yUSDC, and yUSDT. Alchemix has always been a fan of yearn vaults, and they make the most sense as the first yield sources for the V2 system.

The Alchemix core team proposes launching with the following values for the governance- defined, underlying-token parameters:

The new Alchemist also has some global parameters that must be set for the debt-token. The Alchemix core team proposes launching with the following values for the governance-defined, debt-token parameters.

TransmuterBuffer

The TransmuterBuffer is a new component in the Alchemix system. It’s main purpose is to decouple the Transmuter from the Alchemist, enabling governance to finely tune and better control the debt-token peg. As yield gets harvested and debts get repaid and liquidated, underlying-tokens get sent to the TransmuterBuffer and are exchanged for debt-tokens in the Transmuters. In order to prevent inefficient arbitrage opportunities like we have seen in the Alchemix V1 system, there is a flow-rate for each underlying-token in the buffer. This provides a cap on the rate at which funds can be exchanged into the transmuter.

Additionally, (unexchanged) funds can either be deposited back into the Alchemist to earn more yield, or be redirected from the Transmuter buffer to other integrations like an AMO a’la Frax in order to proactively keep the peg. When rehypothecating buffered collateral back into the Alchemist, the TransmuterBuffer uses a weighting system to determine where underlying-tokens are placed.

The Alchemix core team proposes launching with the following values for the governance-defined, TransmuterBuffer parameters.

Post Guarded-Launch Phase

In the above section, the proposed initial parameters are purposefully conservative in order to ensure the safety of the protocol and the first adopters of our v2. However, following the guarded-launch phase, we invite the community to come together to decide the parameters for Alchemix v2 going forward. While the above proposal is extremely granular in all of the parameterization, the key parameters presented below are generalized for the purpose of clearly communicating the meaningful decisions the ALCX community can make. Please read the subsequent notes section if any of these parameters need clarification.

Collateral Types for alUSD:
Limited to single side yearn vaults for stablecoins (not crv lp vault tokens)
Eligible stablecoins currently: DAI, USDC, USDT, TUSD and LUSD
(no choice available for alETH as it will only support yvWETH at launch)

Deposit Cap:
Alchemix v2 views a combination of a collateral and strategy to be a yield token (e.g., yvDAI, aDAI, cDAI are all different yield tokens)
We can set deposit caps for each yield token
Helps us to limit risk, suggesting riskier yield tokens getting lower caps, and vice versa
Must consider liquidity, strength of peg, execution risk (for both centralized and decentralized stables)

Debt Cap:
Global debt cap total for the system across all collaterals.
For reference, Alchemix v1 has a 150m alUSD debt cap, and a 6k alETH debt cap

Convertible Cap:
In v1, users can repay their loans using alUSD or DAI, with the DAI collateral also being able to be self-liquidated to repay their loan.
V2 has the same feature, but to guard the system against a sudden depeg, there is a total convertible cap for how much someone can repay their loan in any of the collateral currencies over the course of a day.
If a token depegs severely, then they can arbitrage it via alchemix, and too much of this could over-balance our composite deposit of stables towards this now weakened stablecoin. Thus, having a convertible cap high enough for good UX, but low enough to mitigate the damage from any collaterals depegging is important.

Loan to Value Ratio:
Current alUSD ratio is 50% (deposit 200 to borrow 100)
Current alETH ratio is 25% (deposit 400 to borrow 100)
Need to consider debt repayment times from yields, and peg strength
Too high of a LTV ratio can allow users to leverage their yield more by looping alUSD borrows into a larger principal deposit
Possible LTVs and their approximate max leverage:
50% LTV -> 2x max leverage
66.67% LTV > 3x max leverage
75% LTV -> 4x max leverage
80% LTV -> 5x max leverage

0xFelix

I think the parameters for the initial guarded launch were well addressed during the call - the only open point from my side would be the multisig question (and the timeline to transfer things back to the treasury multisig). But I understand this will now become an official snapshot vote.

After the guarded launch, the biggest question in my opinion relates to the LTV. I'm personally in favour of increasing the LTV significantly to make the vaults more attractive and practical for users.

An easy win is increasing the LTV on the alETH vault to 50%, to match it with the alUSD parameters. This can be done in two steps (e.g. 33%/50%) to make sure there are no unexpected issues with the vaults and the peg. This could be done immediately after the guarded launch.

After that, I believe we should probably have a focused discussion on protocol parameters and how we want to position Alchemix in the market - do we want the protocol to be seen as conservative (attracting a certain type of user), or do we want to attract a wider DeFi audience with more competitive LTV rates? Personally I would like to explore moving the LTV of the alUSD vault to 66%/75% LTV. So far the peg has been very tight and our liquidity should support a higher LTV.

But changing the LTV% beyond 50% is not that urgent in my opinion (especially if yields remain as low as they are today) and I would prioritize new chains and/or new assets first.

ButlerAndTheThirdStringers

0xFelix I agree with launching with the alETH vault at 50 LTV, but I think any discussions on greater than 50% LTV should be tabled until we have the full gamut of peg mechanisms including the AMO

Afroman726

As a smooth brain gotta say if Eth has LTV 50% I’m good! Agree with everything posted above 😎

ImYourHodlberry

Very good job on the part of the team to get to this point and to put time into laying out these parameters. It's a bit like a congressional bill, though, in that that the only choice is to vote yea/nay on the entire package. I'm sure future votes can be more focused where there is a discussion, for instance, around a single parameter, such as LTV. Having said this, I think this is the only way to present this vote at this time and I will certainly vote for it.

LogrisTheBard

ImYourHodlberry I think it's fair to say if people voted no we'd get some discussion on why and come out with a vote with more granularity.

LogrisTheBard

I'll repost this from Discord. In terms of initial parameters I think we should have a larger discussion as a DAO what the desired ratio between the transmuter pool vs the deposits should be and how the coming AMO will affect that desired ratio. Minted assets that aren't absorbed by Curve liquidity are absorbed by the transmuter pool funds. Once those funds are exhausted the peg is secured only by the income of the assets deposited.

Could we take alUSD collateralization ratio to 150%? Sure. Should we? What are the pros and cons? What's the balance of the alUSD Curve pool and could it absorb the sudden influx of alUSD that would cause? Should we have different caps on these assets or different collateralization ratios for each alUSD backing asset based on their yield? Can we establish some type of consensus on how the collateralization ratio should be adjusted as the yield of the underlying assets shift and as the protocol owned liquidity in the AMO and transmuter pool shift? I'd basically just like to establish a formula we can follow by default as a DAO. Then we can experiment with that formula over time and evolve it.

Regardless of what level we decide, if the collateralization ratio for ETH is lowered, I'll be maxing out whatever I can and using it to farm. So for me, it's more yield. I just want to have a non-selfish take on the matter.

CMSPeary

Excited to see AlchemistV2 and varying deposit caps for each yield token. I like the suggestion by @0xFelix to increase LTV on alETH vault in two batches, but want to remind people that increasing LTV without increasing debt caps probably just means less people use it, so we should also look at potential methods to increase the debt caps.

Overtime I would like to see a more quantitative approach to the LTV and debt caps parameters based on a rolling historical yield generated per vault and how much slippage we are willing to have in each pool (Curve A factor + liquidity). For instance, increasing LTV and debt caps will rocket various pools and we need to make sure pool does not become unbalanced. In addition, different tokens will have different yields, impacting time to repayment and LTV leverage risk. An example is ETH, which have a different yield than stables, especially when ETH goes to POS and the base yield of ETH turns into the staking rate. Setting all other things equal, we should be ok with a higher LTV given a higher base yield.