[RFC]: Proposal - Investing the Capital Pool in Uniswap v3 ETH/DAI

:wave: oSaaT here, posting this as member of the Investment team.
Looking forward to hearing your thoughts on the proposal

Introduction

So far, the investments of the capital pool have been ETH-denominated.

The topic of asset/liability mismatch (ALM) pops up regularly (and rightly so) in the forum discussions.
See the latest forum post by Rei here: [RFC]: Matching Currency of Assets and Exposure

TLDR (on ALM)

  • As the mutual has a majority of claims being bought in DAI, it would make sense to increase the portion of DAI-denominated investments at some point in time.
  • However, as the gearing/leverage of the capital pool is still very low, there is no need to rush.
  • For the past DAI-denominated claims, ETH was swapped for DAI in the days before claims payment as needed.

This proposal addresses this situation, with the main benefit to earn yields in the process.

Rationale

Objectives:

  1. Generate fees
  2. Increase DAI portion of the capital pool (to be able to meet liabilities in DAI, no matter how much ETH/DAI fluctuates)

To be clear, I’m giving slightly more weight to the first objective (‘fees’).

But slowly increasing the DAI allocation is a welcome benefit; also it allows us not to focus too much on the impermanent loss aspect of this AMM-based strategy.

Of the available universe of returns for the mutual to obtain, a Uniswap ETH/DAI position is appealing for a number of reasons.

  • The yields are relatively high
  • The position would accumulate DAI with rising ETH prices (and remain ETH-neutral in a down market)

As a side-note: this strategy allows the mutual to sell ETH for DAI programmatically, removing emotions from the decision (when to pull the trigger); it does so in a market with rising prices, which feels much better than when we were discussing it with ETH reaching new lows on a weekly basis.

Description of strategy

Given the objectives above, I suggest the capital pool should become a liquidity provider in the ETH/DAI pool(s), and start supplying ETH as a ‘single-sided’ LP.

LPing on Uniswap v3 could be an attractive solution for our capital pool, in general.

And especially as it solves the objective of DAI-denominated assets for the mutual, having yields high enough to compensate the much-dreaded ‘impermanent loss’ (IL) of an LP in Uniswap becomes less important a factor.

It is conceptually similar to entering a covered call position, but with no strike date, agreeing to sell our ETH at a DAI price in the middle of the defined range.

Resources:

https://defi-lab.xyz/uniswapv3simulator

https://app.apy.vision/uniswap-v3-positions (filter for WETH+DAI, add min nbr of days and min amount)

Investment strategy

=> Providing liquidity single-side with 100% ETH starting 1 tick above current price

A. Choosing a range:

  1. I suggest designing the position to become 100% DAI at ETH prices above $4,000.
  • Why? close to all-time-highs, as the ETH price was above this level only for about 2 months at the end of 2021. This represents a justifiable medium-term target (150% increase off today’s prices at $1,600).
  • Also, at prices above $4k, the ETH-denominated portion of the portfolio will have more than doubled in value, so this Uniswap investment would represent a smaller portion of the capital pool.
  1. Depending on the weight that the members choose to give to the diversification in DAI vs fees, we could also consider lowering the upper limit of the range
  • eg half-way towards $4k
  • there is always the possibility to split this “Uniswap v3 strategy” in several pools. Let’s say 50% with the $4k limit, and 50% with the lower limit…

B. Reinvestment:

  1. We should claim fees frequently (weekly or fortnightly)
  2. The DAI portion of fees should be sent to the capital pool (for future claims payment)
  3. The ETH portion of fees could be:
  • either swapped for more DAI (aiming for $-denominated rewards)
  • supplied as single-sided liquidity in the same pool 1 tick below current price (range tbd)
  • also sent back to the capital pool

Proposal

Amount and duration of investment

Amount:
The proposal is for Nexus Mutual to invest 10% of the mutual’s current assets.
This represents investing about 16,000 ETH from the Capital Pool into Uniswap v3 ETH/DAI pool(s).

At the present time, this seems like an appropriate level where the risks to the overall soundness of the Mutual are limited (see Risks section), while obtaining a meaningful reward for our investment (see Rewards section).

Duration:
There is no set duration for the investment, with the caveat that the appropriateness of providing liquidity in these Uniswap v3 pools remains.
The longer we stay in the position (“in range”), the more fees we get.

Rewards & performance

Parameters

(as of January 23rd):

There are 2 main pools:

  • ETH/DAI 0.3% pool with a TVL of ~$21m (liquidity in ETH per tick is about 75 ETH at current prices)
  • the ETH/DAI 0.05% pool with a TVL of ~8m (liquidity in ETH per tick is around 7 ETH at current prices)
  • the uniswap TVL data doesn’t include just-in-time liquidity. Depending on pools, there are ‘guesstimates’ it makes between 15% to more than 50% of the liquidity on Uniswap v3…
  • volumes are quite volatile month-on-month, but do seem to be regularly at $10m per day between both pools, which means daily fees of more than $10k
  • The cheaper pool has been getting a larger share of the volume, and this trend may accelerate.

Position size: 16,000 ETH (at ETH/DAI ~1600 => $25.6m)
Nexus Mutual would represent close to 50% of the liquidity.

I suggest we should split the investment across the 2 pools in the same ratio as current TVL (73%/27%).
It appears it would make more sense to put funds in the higher-fee pool, to avoid driving more volume to the lower-fee pool.
However, even at the current levels of liquidity, the 0.05% pool gets about 4X the daily volumes. This shouldn’t be ignored.
As investments of the mutual are more of the conservative and passive types, I suggest we follow the market signals and allocate based on the same ratio.
This could be reviewed on a monthly or quarterly basis (cf. Monitoring)

Rewards (in $ terms) are expected to be in the double-digit %; but these yields will be highly volatile, as they depend on several parameters:

  • if the position is “in range”
  • the trading intensity for this pair,
  • competition from other DEXes
  • and the behavior of additional LPs in these pools, especially Just-in-time liquidity providers.

To give some color - if all things stay the same:

In the ETH/DAI 0.3% pool:

  • Volumes have averaged about $2m per day over the past month → daily fees of $6k
  • Liquidity at the current tick is about $130k → 4.6% yield per day
  • You need to dilute this figure by your own assumption of JIT liquidity (50%, more?..)

APY.vision (https://app.apy.vision/uniswap-v3-positions) has an interesting tool that allows us to see existing positions in all ETH/DAI pools, with filters on duration, amount. Most have an APY above 20%, but all with very different ranges.

See the links above to the other dashboard services which are quite well referenced; especially the one by Defi-Lab allows to compare strategies with different ranges and vs HODL (50/50 or 100% in one or the other token)

Monitoring

The position will be monitored on the basis of these parameters:

  • total TVL
  • Nexus share of TVL
  • level of fees
  • split of TVL across pools
  • combined exposure with the amount of covers sold on Enzyme v4 and Uniswap v3

Outcomes

Scenario Description Outcome for our ETH investment Split ETH / DAI of Uniswap investment
Bear ETH price remains below range and keeps dropping 0 % yield remains 100 / 0
Moon ETH breaks above $4k Get swap fees as LP xx% 0 / 100 (+ ETH & DAI fees)
Crab ETH remains between $1.6k and below $4k Get swap fees as LP xx% a / 100-a (+ ETH & DAI fees)

Risks

Nexus cover of protocol(s)

In order to avoid the compounding of risk within the mutual, Nexus should be minimising exposure so that total risk is capped to 20% of the mutual’s assets.

This is valid both for Enzyme v4 (active Protocol covers) and Uniswap v3.

Risks and Considerations for Nexus Mutual

This section discusses the risks for Nexus Mutual.

Based on the list of risks laid out in the investment philosophy, this strategy qualifies for the “lower risk” bucket:

Risk Comments (cf. Investment Philosophy) Uniswap v3 ETH/DAI pools
Illiquidity Risk Locking up capital for periods of time presents risk to the balance sheet should the mutual need those funds to pay claims. Note that the risk buckets here refer to liquidity in non-stressed scenarios. Stressed scenario liquidity is also a consideration but difficult to quantify. Lower risk as the liquidity can be withdrawn immediately* (*Stressed scenario liquidity is also a consideration but difficult to quantify)
Basis Risk The balance sheet is largely ETH denominated. ETH and DAI are effectively ‘cash.’ Investing in other tokens introduces basis risk. Lower risk as fully ETH/DAI denominated
Protocol Risk (DeFi Safety Score) Putting funds in a vault or liquidity pool to earn a yield opens up our funds to risk of loss from smart contract hacks Lower risk as DeFiSafety score = 96%
Liquidation Risk In the case of lending, Nexus collateral could be at risk of liquidation Lower risk as no liquidation risk in strategy
Leverage Refers to leverage of the balance sheet as a whole. Leverage may either be an explicit component of a particular strategy, or embedded leverage (Options, Futures, etc.) Lower risk as no leverage in strategy
Counterparty Risk Other additional qualitative & quantitative measures of counterparty risk may be used to assess investments & managers Lower risk as exposures to single counterparty is below 20%
Economic risk Some investments may result in losses on a short term basis, e.g. impermanent loss. Lower risk as negligible possibility of loss (in this case impermanent loss would be a feature)

Uniswap v3 is one of the highest-ranked protocol by the service DeFiSafety (96%)

This attests to the robustness of the contracts.These have been live for almost 2 years now and have a high bug bounty of $500k to encourage further reviews.

There are a few additional risks worth addressing:

  • Impermanent loss

This risk is intrinsic to an AMM, especially in Uniswap v3 with concentrated liquidity
This risk is addressed above in the proposal with the rationale why - in most situations - it won’t be relevant for the mutual

  • Liquidity / share of TVL

As the position would represent close to 50% of the liquidity in the ETH/DAI pools, this is worth considering.
DAI being an established ERC-20 and a top stablecoin listed on many exchanges, we do not expect these Uniswap pools to have a large impact on prices.
Increasing liquidity in these pools could also attract more trading volumes.

  • DAI peg risk

This is a risk accepted by the mutual as DAI has been chosen as the stablecoin of reference for covers and claims.
The mechanisms in place at MakerDAO have withstood stress tests in the past.

Implementation:

Technical

The capital pool doesn’t have a direct integration with Uniswap v3, and claiming and reinvesting fees would pose a burden to the management of the capital pool.

For these reasons, we suggest using the Enzyme vault which was implemented for the Maple investment (link to proposal: Proposal: Allocate Capital to Maple Finance ).

Oracle

We need to know the size of the mutual’s position in these Uniswap v3 pools.
It is assumed that it would be handled by the Enzyme/Chainlink integration (to be confirmed).

Governance

The proposal will be put forward through the Nexus Mutual governance process after a period of review and feedback by the community.

If the governance process approves this investment, the implementation steps as outlined in the Technical section will be put in place by the Nexus Mutual team.

1 Like

Hi oSaaT,

Thanks a lot for your proposal, I have a few comments:

The amount of impermanent loss should not be dismissed so quickly. In the moon scenario, at 4000 DAI/ETH, we would be looking at the following:

ETH in the UniV3 position has exhausted. It would be equivalent as if it had all been sold at the middle price between the edges of the position, i.e. 2800DAI. So with a position size of 16kETH, we are talking about an impermanent loss of $19.2m ((4000-2800)*16000) This is not so straightforward but it is a good back of the envelope calculation.

Assuming your estimations that fees are ~10k/day, and that Nexus would own ~50% of that pool, this would mean ~$5000USD/day. Or in other words, it would be needed ~3840 days (~10.5 years) of ETH trading in that price range to just recover the impermanent loss, that would have accumulated if ETH reaches $4000.

And the impermanent loss kicks in from the beginning, which means that a price of 2800 DAI/ETH, the impermanent loss would already be $4.8m.

If the price of ETH goes beyond $4000 and never comes back, anything above it will be missed capital by the mutual on top of the $19.2m, because the mutual will not have those 16 kETH when ETH actually moons and gets to, for example, 10k$. If this proposal gets implemented, ETH at 10k$ would mean $115.2m potential less capital for the mutual due to impermanent loss. It would be basically impossible to recover this with fees.

I think my calculations are ok now, but I may be wrong, so please correct me if I am.

Furthermore, you wrote that this proposal aims to address the situation with Asset/Liability Mismatch, and that currently ETH has been sold for DAI when it has been needed to pay a claim bought in DAI. However, you seem to have mostly considered the case in which the price in the future of ETH is higher than today’s. And if that is the case, it would be better for the mutual to just keep on selling ETH when it is required to pay DAI claims. For example, if ETH gets to 4000USD, in order to pay a claim bought in DAI you should only sell 40% of the ETH you should sell today to pay the same claim with ETH at $1600. And you would not have lost $19.2m to impermanent loss. ALM would be less of an issue than today, so ALM can’t be a reason to start this proposed UniV3 position with this design.

The ALM could get a bigger problem if the price of ETH declines vs DAI. However, your proposal would not be providing any yield or any solution in this situation, because you defined the lower limit as one tick above existing price.

In case that the price of ETH is expected to decrease, and it is wished to cover for ALM, the optimal solution would just be to sell some ETH today for DAI, and there would be also no need for initiating a position in UniV3 in that case.

Also, the most significant asset of the mutual is the high exposure to ETH and its potential, with a large capital pool. Anything that reduces the ETH exposure of the mutual should be extremely well justified, and I don’t think that this proposal justifies it.

Finally, v2 tokenomics should also affect the capital pool size, and so, it would be convenient to wait to see the remaining capital after dust has settled before committing any additional investments. A 10%-of-the-capital-pool investment now, may be a 30% investment later after redemptions. Is 30% a level we’d be comfortable to expose to impermanent loss?

I also don’t think that this proposal can be improved by modifying parameters of how funds would be distributed in different pools or positions based on ETH price levels. Based on all of this, I strongly think that this proposal should definitely not be approved in any form, and if it gets to a vote I will be voting against it.

2 Likes

Thank you for taking the time to review the proposal.

You made some good comments.

Let me summarize these in my own words:

  1. In the moon scenario (ETH >$4k), Impermanent loss (IL) would be very high
  2. Still in the moon scenario, the position is unlikely to generate enough fees to compensate for the IL
  3. In a scenario with ETH at $10k, the outcome for this position looks very bad (IL>$100m)
  4. When the ETH price rises, ALM can be solved with selling ‘less ETH’ for DAI when needed
  5. When the ETH price falls, this strategy would not solve for ALM in any way as it is then out of range and does not accumulate DAI
  6. When the ETH price falls, ALM can be solved optimally by selling DAI today
  7. The high exposure to ETH of the capital pool is very important to members
  8. The future version of tokenomics is likely to reduce the ETH in the capital pool, this should be taken into consideration for position sizing
  9. Changes in the split of funds between various Uniswap v3 pools or in the ranges selected for the strategy would not improve the merits of this investment

I’ll start with the bottom of the list - items 7 and 8:

  1. ETH exposure:

Indeed, I’m aware of the key importance of the ETH exposure to the members (as @Dcc reminded us a few days ago on the forum).

→ Even if this strategy is implemented, the total exposure to ETH would remain higher than 90% no matter the outcome in terms of ETH price.

  1. Tokenomics 2.0:

The impact of the evolution of tokenomics is a very good point.
It should be taken into consideration for the position sizing.

→ If for example we can reasonably expect the capital pool to shrink by 50% (or more), then yes, I’d suggest limiting the position size to 5% of the pool (or less).

Items 1-6
Next, I’ll address items 1 to 6 together by taking a step back and looking at different scenarios.

(A quick note on your calculations: these seem correct regarding impermanent loss. However - all things staying the same - I expect the daily rewards to trend higher following the increase in the ETH price)

On Outcomes (items 1-3):
Indeed, there is no scenario which gives us a perfect outcome across all criteria (IL, fees, ALM)

→ This strategy represents a balancing act between fee generation and addressing ALM (for the position AND the capital pool), as stated in the objectives.

Unfortunately, there is no way of knowing where the ETH price will be 1 year from now.
(although I do remember quite well the exuberance across the board when ETH reached $4k (during times with interest rates much closer to 0), this period does feel quite remote)

→ Each member should make his own estimate of probabilities for each of these scenarios (and others) to evaluate the merits of the proposal.

On ALM (items 4-6):
One advantage stated in the description is that the strategy programmatically sells ETH for DAI, instead of waiting for claims and hoping the ETH price is right at the time.

Hi again oSaaT,

Thanks for answering my comments, you did a good job summarizing them, I am not sure that the answers addresses the concerns though. To summarize:

  • Accrued fees wouldn’t compensate for IL, even if fees are accrued for 10 years time (not sure why you chose 2 years in your diagram). And that is for a Ethereum price of $4000. If ETH price goes up to $10k+, the cost of opportunity for the mutual would be catastrophic. You mentioned exposure of 90%, so it would remain high This is relative exposure: the share of the total capital invested in ETH. To calculate losses, you need to use absolute exposure: the total capital invested in ETH. Obviously absolute exposure would be lower if we convert the ETH into DAI. It is absolute exposure that should not go down unless fully justified, regardless the relative exposure. And it is true that it is anybody’s guess ETH future price, but we are all here because we believe that the adoption of crypto would only increase with time. Since The Merge in Sep’22, the total amount of ETH has barely increased, and it is expected to decrease with time. Under these conditions, ETH increasing in value with time above previous ATH has a significant probability of happening.
  • With regards to ALM, ‘no-action’ would be a better strategy for the mutual, compared with a UniV3 pool, in every scenario, except one in which price of ETH goes up, then down, below the average sale price for the ETH in the UniV3 position, AND in this moment, it is when a cover underwritten in DAI needs to be paid. Suffice to say, this is an unlikely scenario.

I have to say that unfortunately your answers did not convince me, I still strongly oppose this investment proposal, I believe it would be very bad for the mutual, and I will vote against it.

1 Like

I appreciate this proposal but no one can escape or outsmart impermanent loss. I have near zero faith in our ability to do so when there’s protocols dedicated to managing and beating IL and they’re all failing at it.

The idea here is on point, and I hate to shoot down the proposal so bluntly but no way in hell am I putting ETH into an ETH/DAI LP.

If we were to ever place ETH into a like-asset Curve LP (ie stETH/rETH) then that would be worth considering because we deposit ETH and are guaranteed near 0 IL.

I’m a NO on this proposal but thank you for the taking the time to suggest this.

5 Likes

Thanks @DeFi_Dad for taking the time to post a response.

I’m with you regarding an ETH/DAI LP position not being particularly exciting (or a “hell no” as an ETH investment) - especially when looked at in a vacuum.

A clarification for all readers:
→ This proposal is really meant to be estimated together with the topic of asset/liability management (as mentioned in intro), which is being discussed in parallel on the forum at this page.

Excerpt of @Dopeee’s reply in that discussion (emphasis mine):

ReeseWickham made his position very clear above.

Although implicit in @DeFi_Dad’s reply, I assume you’re also in the camp:
keep max. ETH exposure on the asset side as long as max. gearing is not in sight; sell ETH for DAI to pay claims when needed.

Again, thanks for sharing your views!

1 Like

This proposal is ultimately just an extension of whether or not we should be selling ETH and diversifying into DAI. If you disagree with selling ETH, you’re obviously going to disagree with this proposal.

But if you think that selling ETH to DAI is good but doesn’t need to be rapidly dumped, you’re probably going to be more open to this proposal.

I’d suggest that we do a Snapshot vote on whether or not we should be selling ETH. Then, likely we need another vote to decide how much we want to move towards a 1:1 with our exposure, as there’s no requirement that we sell ALL excess ETH. Then, we can vote on the appropriate mechanism to handle any sale.

Perhaps those first two votes can be combined, just want to avoid a situation where “no” is one option and there’s multiple other options, so “no” wins because of split votes.

3 Likes

In favour of this proposal. If one agrees there’s an asset/liability mismatch for the covers, then there’s a need to sell ETH to DAI at some point.

A Uniswap V2 pool will always have equal amounts of ETH and DAI in it at any price [0, inf); a V3 merely splices up the entire curve into buckets and allows you to withdraw the ranges not needed. I would propose we do not excessively concentrate the liquidity in V3, but maintain the V2 shape.

As with others, I’m also rather bullish on ETH in the long run, so would support at most 10% of the capital pool be deployed for future sale (e.g. single sided ETH, above current price, up to a end target price). Using the ETH from Maple makes sense.

IL is the necessary side-effect of having a balanced portfolio; we don’t call rebalancing a 60-40 equity/bond portfolio IL, do we, if the goal is to maintain the ratio?

The fees generated is quite significant as well, and a cherry on top of maintaining the portfolio in the right ratio.

Now that V2 has launched successfully, might be a good time to revisit this and snapshot perhaps!

3 Likes