$ALCX: Alchemix Cannot Turn Lead to Gold

There is a brand new DeFi protocol that lets you immediately take out a loan for half your collateral and the protocol will automatically pay back your loan. Financial magic, involving transmuting one coin, into another, while depositing one in a pool, magically you can get a loan and have no risk of liquidations (not really but that’s what people are claiming [1, 2, 3, 4, 5, 6, and the coup de grace their website (archive)]). I think this protocol is flawed. You can find the whitepaper here. (Archive) (My copy)

So what is Alchemix? It’s a ‘Future Yield Tokenization Protocol’. What does that mean? Basically, it’s a way to increase your leverage, while taking advantage of yield opportunities in the market.

This protocol works by having you deposit your Dai into a smart contract. After this has been deposited you are able to withdraw a new stablecoin called alUSD. You can withdraw up to half of the amount of Dai that was deposited. The Dai that was deposited is then used to earn yield in a yield aggregator, specifically the Yearn yvDAI vault.

90% of this yield is then placed into a special pool they call the ‘transmuter’. This is a pool where people who hold the alUSD token can stake their tokens. Effectively to withdraw your proportion of that yield then you have to burn your alUSD token and you withdraw the Dai. The tokens in this transmuter are often placed again into yield farms in order to increase the total yield of the system (this might be the Boosted Alchemix APY that I cannot find in the documentation or whitepaper).

The remaining 10% of all yield is sent to the DAO treasury, currently controlled by a few multi-sig key holders.

Approximately 40% 20% of the tokens will go to the developers and creators of this protocol. This does not include the treasury which will be controlled by these same multisig holders.

This protocol is currently un-audited.

I cannot seem to access the page linked to under governance on their website, this is likely user error. I believe I found it here. It does not seem to be very active.

Most of their documentation has not been updated in months.

The contract for their alUSD (archive) token seems to include several features (setBlacklist, pauseAlchemist) which might make people hesitant including the ability for the admins to blacklist and pause various users, which to me seems to directly limit the ability for this protocol to maintain meaningful censorship resistance.

Update: I was mistaken about the above. The setBlacklist function and setWhitelist functions are working as intended to make sure only the Alchemix contract can mint.

There is a Boosted Alchemix APY that is mentioned on the site, but it is not mentioned in the whitepaper or in the documentation.

When an emergency exit is declared there are functions (recallFunds, recallAll) that seem to allow funds to be recalled from the individual vaults and deposited in the main alchemist contract. (Archive) It appears that this state can be triggered by the Sentinel Address or the Governance Address (also not mentioned in the whitepaper or documentation that I can find).

These problems suggest that at the very least this protocol is underdeveloped. However, we should also look at the incentives and the design to determine if there is any potential in the idea.

So Dai you deposit is sent out to yEarn to earn yield. Specifically it is deposited in this vault (per founder). This vault pays 2% of deposited funds and 10% of earned yield back to the yEarn treasury. 10% of the earned yield goes to the creators of the strategies that are being used in the vault. The specific strategies being used are:

  1. StrategyLenderYieldOptimiser
  2. StrategyIdleidleDAIYield
  3. StrategyAH2EarncyDai
  4. StrategyGenericLevCompFarm
  5. IBLevComp
  6. SingleSideCrvDAI

These strategies largely work by finding places where this Dai can be loaned out in order to maximize returns. Whatever yield they get is regularly sold for Dai. This Dai is then regularly harvested by the Alchemist contract. A portion determined by the Harvest Fee is sent to pay down the debt in the system, and the remainder is sent to the transmuter.

Fundamentally, in order for Alchemix to continue working there are several things that need to happen:

  1. Dai has to remain valuable, a prospect I have been worried about.
  2. There needs to be no exploits in any of the smart contracts deployed for Alchemix.
  3. There needs to be no exploits in:
    1. MakerDAO
    2. yEarn
    3. The protocols yEarn depends on to earn yield
      1. This risk is partially mitigated by the fact that there are multiple yield opportunities, and a failure of one does not necessarily impact the others.
  4. The yEarn vault cannot be compromised, in the past this vault has been attacked.
  5. The admin keys cannot be compromised.
  6. The admin themselves cannot be compromised.
  7. The yield payed out by the yDai vault needs to exceed the interest paid.
    1. Interest is currently zero, which can be accomplished because it is a stablecoin to stablecoin loan
    2. Interest seems unlikely to be zero in the future when the protocol integrates WBTC and WETH

Fundamentally, what is happening here is shockingly simple. The protocol allows for interest rate arbitrage. Since you are over-collateralizing your loan, and exchanging one stable asset for another ‘stable’ asset the loan can be done effectively at zero interest. Then the collateral is deployed to earn interest, which is then applied to the outstanding loan paying it down. Part of the reason this can work for now is that there is significant demand for loans of Dai and as such people are willing to pay high amounts of interest to get that, which they are willing to do because there are opportunities to make quick gains in cryptocurrency trading. However, in situations where the market rapidly moves down you start to see all of this leverage start to break down. We have seen Dai significantly break peg in the past, which in term is likely to hurt the value of alUSD. Since alUSD is effectively pegged to Dai it has all of the risk factors of Dai compounded with the risk factors inherent in their own protocol. Besides that downturns in the market are likely to rapidly affect the yield available, increasing the time until your loan can be paid off. Overall, I am skeptical of the long term resiliency of the Alchemix protocol. In much of DeFi I am increasingly worried about the tendency to try to find ways to increase the total leverage of the system as a whole.

Consider for example a user with ether who wants to retain upside exposure to that Ether and find ways to double down on that position. It is possible to deposit that ether into MakerDAO and then receive Dai in return. This Dai could then be deposited into Alchemix where you could receive alUSD in return. The Dai you deposited is then loaned out to other people who can then buy Ether. Meanwhile you still have your alUSD which you can then use potentially to purchase even more Ether. This Ether can be deposited into MakerDAO… as you can see it is possible to pretty quickly build up across the DeFi system as a whole a significant levered Ether position, and I worry that rapidly falling Ether prices will see many of these individual components.

Please enter your email to subscribe to my newsletter:

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.