[AIP-20] Diversifying treasury's ALCX holdings without selling

Alunara

tl;dr
AIP-20-A: put 15k alcx on abracadabra, mint mim at 18.25% liquidation price of current price, buy cvx, lock cvx and vote for alusd gauge and put cvxcrv rewards from locking back in convex for flywheel effect.

AIP-20-B: put 15k alcx on abracadabra, mint mim at 18.25% liquidation price of current price, swap mim for 3crv, put 3crv on alusd farm, farm crv + cvx + alcx, put crv in flywheel, lock cvx, use alcx for paying off original debt.

Proposal
In this post I propose two ways of diversifying the ALCX currently in the Alchemix treasury without directly selling that ALCX in the open market. These two votes can be voted on separately under the monikers of AIP-20-A and AIP-20-B. Both proposals are not mutually exclusive and can be executed both if approved. They are very similar in nature due to their reliance on Abracadabra and therefore will be discussed in this single forum thread.

The first proposal's goal is to use ALCX's value in order to gain Curve gauge voting power through the means of purchasing CVX.

The second proposal's goal is to use ALCX's value in order to diversify into stablecoins and putting those stablecoins to work (and enhance AIP-17).

Current situation
At the moment of writing, the Alchemix treasury holds around 100 million dollars worth of ALCX and around 2.2 million dollars of stablecoins that are currently in the 3pool Convex farm as per AIP-17.

Abracadabra
Abracadabra is a defi protocol that allows users to put up interest-bearing tokens as collateral in order to mint MIM; a USD pegged stable coin similar to DAI and USDC. Abracadabra has an ALCX vault which implies users can deposit ALCX in order to mint MIM. Like MakerDAO, there is a chance of liquidation, but in both proposals I will suggest a conservative and low-risk liquidation price.

You can read more about it in their Medium post.

AIP-20-A: Minting MIM to purchase CVX
As many may have noticed, Convex is taking defi by storm. Convex is a protocol developed on top of Curve that boosts rewards for CRV stakers and liquidity providers alike. As it stands right now, Convex has locked up 93.4m CRV and controls 34.51% of Curve's gauge voting power. That voting power is given to CVX holders that lock up their CVX tokens for roughly 16 weeks.

The first CVX gauge vote will be on the 16th of September, so there's still time for people to lock up their CVX if they want to vote. As it stands 1 CVX holds the voting power of 16.57 CRV (source). This will lower as we get closer and closer to the 16th of September, but it wouldn't be surprising if 1 CVX ends up holding the voting power between 6 and 10 CRV. For the sake of argument, let's assume 1 CVX will hold the voting power of 7 CRV.

We currently have $2.2m in stablecoins farming CVX. Those stablecoins have farmed around 500 CVX, which if we extrapolate linearly (bad assumption for multiple reasons but the most simple to reason with) means we will farm about 15k CVX per year. A rough estimate would tell us that it would take 6-9 years of farming in order to accumulate around 85k CVX.

In this proposal I suggest we take 5% of the ALCX treasury (15000 ALCX) and put it on Abracadabra in order to mint around ~$780k MIM at a liquidation price of $70 at the moment of writing (18.75% of current ALCX price). This is a custom liquidation price that's even lower than what is considered safe ($116), and there are 0 loops involved.

We can then sell that ~$780k MIM MIM for roughly 58k CVX assuming the price remains around $13.25. This way we can accumulate more CVX way faster than the 6-9 years of farming while the token is not yet relatively scarce. Assuming 7 CRV per locked CVX, this would give us around 0.4m of veCRV voting power.

For reference, our 400 ALCX bribes have shifted 2.4m veCRV votes to alUSD, doubling the gauge rewards. This is however just for a single gauge vote. Buying CVX directly is preferential over bribing CVX holders as this will give us permanent voting power instead of temporary voting power and will most likely pay off in the end. Bribing CVX holders may become more lucrative as CVX gets more scarce and pricier in the future.

Bonus: Locking the CVX up in order to gain voting power yields cvxCRV rewards, which is around ₁₂% APR atm. I would suggest to stake these up together with the rewards of AIP-17 to increase the flywheel effects. The 3crv we earn with staking cvxCRV can also be used to pay off the debt on Abracadabra.

AIP-20-B: Minting MIM to diversify into stables
As you might imply from the current situation, stablecoins are only 2% of Alchemix's portfolio, the other 98% is ALCX (ignoring the other scraps). It's been talked about before, and I too think it's wise to diversify a little bit into stablecoins to reduce risk during market down turns (although we've already had a pretty rough time looking at ALCX price history). The goal of this proposal is to diversify, not to maximize CVX farming output.

Instead of selling this ~$780k MIM and taking an active trading position in some other token, I suggest we put it in a Convex alUSD farm which yields between 20% and 25% APR at the moment of writing. In the end we'll have a bigger rainy day stablecoin fund that is being productive by farming more CRV, CVX and ALCX for us. The CRV and CVX can again be staked into Convex for the flywheel effect, similar to AIP-17 and AIP-20-A if it passes. The ALCX can be used to pay back debt or added as extra collateral. This would be almost like selling ALCX for stable coins, except we're not selling any! Another benefit is that in case of an ALCX downturn we can buy additional ALCX using the alUSD to secure the liquidation price of our debt if necessary.

Risk management
Whenever we get close to 50% of the liquidation price because of a falling ALCX price, we should either add collateral or pay back debt such that the liquidation price becomes the original 18.75% that we started with. I think the best course of action is to pay back the debt instead of adding collateral, since adding collateral increases the potential loss instead of decreasing it.

Disclaimer
I personally own CVX and SPELL tokens (13% and 1.5% respectively of my portfolio) and I assume that this will "be good for my bags". I hope I can objectively convince you that these proposals are indeed good for Alchemix.

Amendements:
https://forum.alchemix.fi/public/d/208-aip-20-diversifying-treasurys-alcx-holdings-without-selling/7

Underscore

"Bags don't dictate my bias, my bias dictates my bags". IMO it's better that you're holding CVX & SPELL while proposing this since it shows you believe in the protocol enough to get behind it with your own money. I'm all about this, Abra has been synergistic with Alchemix so far so I'd like to see more cooperation between the protocols.

ov3rkoalafied

My only worry is that a bear market comes, $91 ALCX doesn't seem crazy. That's only like a 70-80% drop from here. I would add a "multisig repayment price", ie if alchemix drops to a certain level, we repay the debt up to a new safe level.

So for ex - If ALCX drops to our preset value of $182 (91x2), we liquidate some of the farms to pay until we are back up to the same approximate leverage we started with (₂₀% I believe). That way we are never really risking paying the liquidation fee.

This way we're always ahead of the liquidations with a predetermined plan.

ButlerAndTheThirdStringers

I like the idea, but I think it's too aggressive and risky as currently proposed. Under the current proposal 10% of the entire treasury ALCX would be at risk of liquidation. That's a significant allocation. $91 might seem safe, but we've seen violent moves downwards before and there's no guarantee we won't approach that range at some point in the future. At that point the devs would be scrambling to pay back the loan or lower the liquidation levels, and given the fact that the CVX could be locked for voting purposes, this could put us in a position where we need to spend more treasury stables or collateralize additional ALCX for safety. A slippery slope. Additionally, AIP-20-B would expose us to depegging risk of an asset we have no control over in MIM. I like the project a lot, but I don't fully trust their ability to maintain that peg in perpetuity, and I don't think it's been battle tested enough to consider it safe.

Here are the revisements I would like to see entertained personally:

Change AIP-20-B to use a safer Convex farm. I would advocate for the typical 3Pool (USDT, USDC, DAI). Yield earned would be less, but at least we'd easily be able to pay off the MIM if needed and we would not be exposed to additional peg risk. At the very least we should be splitting liquidity between the two.
A reduction in the treasury allocation devoted to AIP-20-A. We have to assume that at any one time the CVX we purchase will be locked for an extended period and not be able to be used to pay back the Abra loan if needed. To be honest I'm not sure what number makes the most sense here, but 5% seems aggressive exposes us to the aforementioned risks of a larger treasury allocation being needed for fortification in the future. Maybe 2.5% is right? Maybe less? Suggestions are certainly welcome.

0xFelix

So we already discussed this quite extensively over on Discord, but here's my two cents:

AIP-20-A
I would support this and this is pretty much the idea as discussed initially. I think this is a lasting way to buy influence for any gauge voting which we will need for current alTokens pools and for future alTokens pools. It's also more sustainable than participating in the bribing game. On top, as you mentioned, it provides a nice income stream for the treasury as well.

Also, I agree with ov3rkoalafied in that we should add an addendum to this proposal that if we the price of ALCX were to drop by x%, we would authorize the treasury to make a payment to avoid liquidation. This could either be by deploying stablecoins or by withdrawing more ALCX and adding this to the collateral position.

AIP-20-B
This to me is more risky and I have reservations about deploying treasury assets in this way. It's true that MIM Convex farming is currently very lucrative but that's mostly because MIM has won the bribing game for now - I don't expect these yields to last, and I don't think the treasury should be speculating in the yield farming game.

I do think that we need to look into diversifying the treasury assets but then I would say lets look at something like UMA, who have an entire product build around DAO treasury diversification.

Dropping 5% of the treasury assets into this strategy exposes us to a lot of risks and to me personally I don't think the rewards are worth it.

Overall
Finally let me just say that with both these proposal we will (again) be moving treasury assets from the timelocked treasury multisig to the dev multisig which is inherently more risky. Don't get me wrong, I trust the devs to manage these assets properly, but I also think this is an inelegant solution and I would rather see a system of treasury proxy contracts (as mentioned earlier here) or some other solution.

In summary, I support AIP-20-A if we add an addendum for emergency situations, but I have several reservations against AIP-20-B.

Gravity

I like the idea.
I don't find this aggressive (imo) but I hear others concerns.
Prioritising debt repayments via 3crv rewards would alleviate these concerns. Note, the first cycle of collateral > debt > debt repayment, is always the most concerning as no real data exists prior to extrapolate from. Post the first cycle, it gets very exciting and very interesting. Am for this proposal.

Only additional comment - Alchemix should put out their CVX voting address as some may choose to delegate.
Further to this point, bribing would be far more efficient at the CVX delegation level than gauge voting on Curve due to the multiplier Convex offers.

Alunara

I've made a few amendments:

Updated numbers in general to reflect the state of the market. Sadly, CVX price has gone up relative to ALCX, so AIP-20-A would now only yield us 58k CVX instead of 85k CVX.
Lowered liquidation price from 25% of the current price to 18.75% of the current price. This is halfway 12.5% and 25%. I feel going all the way to 12.5% is too conservative and we wouldn't get the most out of this proposal. Even 25% was already below what is considered low risk, as that would require a 75% dump which would bring the ALCX price around 50% below it's ATL
Changed the farming strategy for AIP-20-B from MIM to alUSD. This has several advantages:
a. We do not have to rely on MIM holding peg. Several community members didn't feel safe yet relying on a new and unproven stablecoin.
b. We will be partly providing liquidity to the pool of our own synthetic stablecoin, which is good for the protocol in itself.
c. We will not have to decide what to do with the SPELL rewards.
d. We will be farming back ALCX tokens which can be used to pay back our debt.
e. In case of an ALCX downturn we can buy ALCX using the alUSD to secure the liquidation price of our debt if necessary.
Added a risk management trigger that states we should restore our 18.75% liquidation price target in case we reach 50% of the liquidation price.