Proposal: Allocate Capital to Maple Finance

Governance Discussions Nexus Mutual Protocol Improvement Proposals
1

Introduction

Currently, we have excess ETH in the capital pool which is not earning yield. Right now, roughly 20% of the capital is staked in Lido while the remaining 80% is held in ‘cash’.

As we work to cover ETH 2.0 slashing risk, it makes sense for us to look for avenues to earn yield on our ETH outside of staking. One of the few options is to lend ETH, but the yield is typically incredibly low because demand is relatively small given the ease of borrowing in stables, and the supply of ETH lending is high.

Rationale and Understanding Maple

The exception to this rule is large ticket size, undercollateralized lending to institutions, funds and DAO’s. Currently the best platform for this is Maple Finance which has issued more than $1bln in loans since its launch 10 months ago. As the collateral decreases, rates will naturally increase to compensate for the additional risk, which is why the vast majority of loans of Maple are not only undercollateralized, in fact they have zero collateral.

For all intents and purposes, it’s best to think of these loans as zero collateral, as this is the most likely scenario. Instead of relying on collateral, Maple loans are enforced by legal contracts (Master Loan Agreement), governed by NY State law. Ultimately, the bet is that the borrower will not default on the loan, and if they do, that recovery will to some extent be possible. Therefore, the most important role is in assessing the creditworthiness of the borrowers. On top of that, the pool is protected by the so-called pool cover (aka junior tranche) that serves as a first loss capital in case of a default.

Maven 11 Pool

To assess the risk of a borrower and establish terms of the loan, Maple introduced a concept of Pool Delegates with Maven 11 running two permissionless pools (USDC and wETH). Besides being an investor in Maple, Maven 11 opened a brand new lending division within their firm to manage the pools and perform the credit risk analysis on the borrowers. For this purpose, they dedicated a total of 4 FTE including one quantitative analyst with a risk management background, external advisor who has over 10 years of experience in high-risk loan origination and credit risk management who currently manages a EUR 2B credit book in his TradFi job but equally has been a DeFi participant for years and shares their passion and conviction to the space. After the successful launch of the USDC pool, they also added a junior with credit background and a director spearheading their institutional sales with over 15 years experience in capital markets including managerial roles at Kempen & Co, KBC Securities and NIBC.

Risk and Reward

Much of the risk associated with no collateral loans to these funds and institutions is mitigated by the size of the loans relative to the balance sheets of these funds. Most of these funds act as market makers, where continuous streaks of profitability into the 18 months+ are not uncommon (e.g. Alameda, Wintermute). These are reliable businesses that deploy market-neutral trading strategies i.e. don’t face directional exposure to the underlying assets. Even under relative black swan events these strategies have proven to be resilient and this helps to explain why it’s possible to lend to them, without collateral, and expect a positive return over the long-term.

Currently, the returns on ETH lending with Maple and Maven 11 are ~ 9%. The ETH lending generates ~ 4%, while MPL rewards generate ~ 5% (subject to the $MPL price fluctuation). Obviously, the rewards currently make up the majority of the yield and at some unknown point in the future, these rewards will have to reduce, or stop completely. This is true across all of DeFi, not just Maple. This is a risk to keep in mind, as the 4% return solely from the lending would be unacceptable in my opinion, as the risk is greater than staking and the return is lower.

But as a combined yield, currently, the return is very competitive. As ETH staking yields increase post-merge, as many believe will happen, the yield on lending ETH should increase as well, as the supply of ETH lending will alter, and demand for borrowing will increase with ETH staking rate becoming a benchmark borrow rate.

How does lending work in detail?

This wETH pool by Maven11 is very new, and we’ve been involved behind the scenes for many months working towards getting comfortable with this strategy. The pool itself allows depositors to deposit wETH, which can then be lent out by Maven 11 when terms are agreed to with their group of borrowers.

Each borrower will get their own terms, for individual loans, depending on their credit worthiness, the term, the size, the collateral, the interest rate and the alternative demand for borrowing.

Loans are typically for 90-180 days, with an average interest rate of 5% (the lower APY above accounts of time when funds are stagnant). So far, seven loans have been made since early March. The details are here:

image
image1418×680 68.1 KB

wETH is deposited into the pool, and an LP token is received. That LP token can then be deposited into a second pool to earn MPL rewards.

Once deposited, funds are locked-up for 90-days before they can be removed. After 90 days, if the lender wants to leave the pool, they start a so-called cooldown period that lasts 10 days. Cooldown period allows the pool delegate to prepare the required amount for the lender ahead of time. When the 10 day period passes, the lender has 48h to withdraw the funds. If the lender misses the withdrawal, they need to trigger the 10 day cooldown period one more time.

The pool delegate (Maven11) earns a 10% fee from the pool of interest earned, and an additional 10% goes to those providing ‘cover’ for the loans. On top of that, pool delegate receives ⅓ of the 1% annualized establishment fee that is charged from the principal loan amount. The remaining ⅔ of the 1% annualized establishment fee is flowing into Maple treasury which will be used to buy back MPL on the open market and distribute to xMPL stakers (new tokenomics).

Cover refers to a pool of capital which is staked to act as a first-loss tranche in the event of a loan default. This cover is fully burned before any of the lender’s capital is impacted. Currently, there is 763,657 USD in cover, which is provided from Maven 11 themselves (Pool Delegate and venture fund). The first figure represents 1.77% of the principal at risk. Therefore, lenders would only be hurt when a default results in the loss of more than 2.37% of the principal. Currently, the only way to provide pool cover for the pool is to add liquidity to the wETH:MPL pair on Balancer which faces IL, in the upcoming weeks single sided staking will be enabled which is expected to bring more capital to the pool cover as the issue of IL will be mitigated and stakers will be rewarded with 10% of all interest payments in the pool (pure wETH yield). Maven 11 is also committed to increase their allocation towards the pool cover once single sided staking will be implemented.

To summarize, depositing wETH in the Maven 11 liquidity pool on Maple Finance will generate two different types of yield: yield in wETH, and yield in the MPL governance token.

  • wETH yield, as reported, has a target of ~ 4% APY. This yield is released on an ongoing basis as the interest payments are paid.
  • MPL reward token yield, boosting expected total APY to ~8-10%. These MPL rewards are released on a block by block basis.

The APY itself is dependent on 3 factors:

  • Interest rates - when the interest rates among the borrowers in the pool are increasing, the wETH part of yield is increasing as well (and vice versa).
  • Amount of liquidity in the pool - when the pool increases in size, the rewards in MPL are decreasing proportionally as the number of MPL tokens needs to be distributed among a larger number of participants (and vice versa).
  • MPL price - when the price of MPL is increasing, the rewards in MPL are higher in $ terms, effectively increasing the APY (and vice versa).

Proposal

At scale, this is one of the few strategies that allows us to deploy 8-figures ($ terms) and earn a competitive yield for the foreseeable future. I believe that this strategy has a risk profile that’s acceptable, given the mutual’s liabilities, for up to 10% of the capital pool. This % might change as the borrowers change, both in number and quality, as the level of collateral alters, and as alternative yield sources change. On top of that, to allow for maximum utilization of the capital provided to the Maven 11 pool, the 10% allocation will be deposited in tranches. This way only the capital demanded by the borrowers will be deployed to earn yield which will prevent dilution of the pool. We aim to deploy the first tranche of $20-25 million immediately.

My proposal is to allocate 10% of the capital pool value to lending wETH with the Maven11 pool on Maple Finance.

It would be also beneficial to discuss what can be done with the MPL rewards. There are 3 options:

  1. Sell continuously to lock in the rate
  2. Stake MPL to xMPL to receive MPL rewards
  3. Option 2 + use the xMPL to provide the pool cover and earn extra 10% of interest payments in wETH

My suggestion is that we pursue option 3 with half of the MPL rewards, and continuously sell the other half to lock in the rewards. Given the size of our investment, rewards will be substantial and excessive selling is likely to lead to large slippage. I also have a long-term positive view on Maple, and think there is a benefit to accruing additional MPL tokens.

There is a limit to how much pool cover can be provided in a capital efficient manner, so once the pool cover is at 10% of the pool size, we should refrain from adding additional cover. Instead, excess returns beyond this point should be sold as efficiently as possible to maximize the achieved APY.

Technical Integration

In terms of bringing this to life, the easiest way is for us to use Enzyme. For those that don’t know, Enzyme is an asset management protocol that allows you to create a pool, interact with smart contracts and hold assets.

Given that every asset that we hold in the capital pool increases the gas costs of operating Nexus Mutual, and running this Maple strategy would require a price oracle, doing it inside of Enzyme simplifies this. By doing so we can get this live in a couple of weeks, rather than months.

The cost of doing so will be:

  • 25bps in fixed fees on the assets in the pool
  • 50bps in performance fees on the profit in the pool

Presuming an 8% APY (on the lower end of our predicted range):

40m * 25bps = $100k per year in protocol fee for usage of the platform (standard fee for every vault, not specific to us)

40m * 8% * 50bps = $16k per year in “management fees”

40m * 8% = $3.2m in APY per year.

Net of fees, we’re looking at $3,084,000 on a $40m base, or 7.71% APY after fees.

Useful links:

Lido proposal: https://research.lido.fi/t/yield-generation-on-maple-finance/1620

Rook proposal: https://forum.rook.fi/t/kip-draft-yield-generation-on-maple-finance-weth-pool/267

FAQ Lenders: https://docs.google.com/document/d/1cIp1ldtfNN3w5rr1bs7H9gEYRI_1jDwWkIxyGE_BgI4/edit

Token Terminal: https://tokenterminal.com/terminal/projects/maple-finance

6 Likes
2

Thanks @Dopeee - this is a super high quality post. Appreciate the efforts you and the rest of the investment committee have been putting into this.

I agree that this needs to be generating yields materially above staked ETH otherwise the risk/reward isn’t worth it, and on balance I think this is a good opportunity for the higher risk end of our investment allocation. I support this on the basis that we keep it to 10% of the capital pool max and review on a semi-regular basis given the yields will fluctuate with MPL.

On the technical side, using Enzyme has further benefits in the future as it gives us options to manage some smaller asset allocations without impacting gas costs on regular Nexus users (due to increased oracle usage). So I’m very supportive of this implementation direction.

5 Likes
3

Great proposal. As a pool delegate on Maple, can attest to the team’s and the platform’s strength, as well as the rigorous underwriting process required of pool delegates.

The outsized yield, combined with strong underwriting and high-quality borrowers makes this a very attractive risk-adjusted opportunity. If long-term bullish on MPL, the expected yield is really a floor given the potential right-tail upside.

2 Likes
4

Excited to see the mutual invest Capital Pool assets! A few questions, what is the time period for calling back capital in the case of claims payouts? Are there concerns about MPL liquidity when selling rewards we earn?

It’s worth noting that the mutual doesn’t cover Maple Finance so this doesn’t suffer from the “double dip” problem faced by other defi projects that have been explored as an investment vehicle for the Capital Pool. Overall, looking forward to seeing this implemented!

2 Likes
6

Quite good idea :+1:

1 Like
7

Hey Katie - There is a 90-day initial lockup and after that funds are callable at anytime thereafter (side note - the way pools currently work is subsequent deposits reset timer to weighted average lock up period). after 90 days pass, at any time we can trigger the 10-day “cool down” period. This 10 days is for them to arrange funds etc. Realistically, would likely be longer to pull out 100% of the funds IMO. I think it’s conservative for us to plan for 21-28 days to get 100% of the funds out.

I’m not overly concerned about this, given that it’s only 10% of the capital pool.

In terms of liquidity at say a 4% MPL reward APY rate on a $40M deposit, that’s approximately $1.6M of MPL rewards per year, ~$4,400 per day in MPL. MPL averages ~$6M per day in trading volume. Therefore our daily rewards amount would be 0.07% of the total traded MPL volume making liquidity not much of a concern. On top of that, we can sell rewards in larger batches via Wintermute if slippage ends up being significant.

3 Likes
8

We’re actively discussing icETH as an alternative for some of our Lido exposure. There are questions around risk, as well as whether the technical burden for making this happen is worth it given a presumably small allocation size

3 Likes
9

Hey Nexus Mutants! Great proposal. Maven 11 joined Maple as pool delegate back in July 2021, and now operate two pools on the platform. The pools grew off the back of diligent work by the Maven11 team to issue capital quickly to blue-chip borrowers.

Pool utilization rates have remained over 80% meaning that APYs remained stable, and were boosted through MPL rewards.

I remember meeting Hugh back in 2020 when we were just getting started, its been great to see Nexus grow! Looking forward to seeing this approved and watching this partnership expand overtime. :handshake:

6 Likes
13

I’m a proponent of Maple. I’ve known the team since early 2021, followed their progress, and recently became a MPL holder just bought on the open market. These guys are leaders in their category, originating over $1.5B in loans but only $10M default so far and it was Babel caught in the 3AC contagion.

To summarize, we are earning substantial yield over stETH. I’m not sure how much we risk here, 10% sounded a little high but here’s a few reasons why I’m favorable and a few to consider in terms of risk to our deposited ETH:

  • The yield is sourced from fixed lending rates for terms of 90 or 180 days so the yield is more predictable.
  • The Maple model is an on-chain lending desk run by Pool Delegates who earn 10% of all interest + 0.33% as an origination fee (also 0.66% origination fee goes to Maple DAO, of which 50% is used to buy back MPL).
  • Pool Delegates from crypto institutional funds like Maven 11 not only earn a share of the interest, but also have to put up collateral in the form of MPL-USDC, so they have major skin in the game, where up to 1/3 of the Cover backstopping against a borrower defaulting can be liquidated.
  • The Pool Cover for this WETH Pool is only 1.33% of the Active Debt, meaning there’s about $328,759 of value to protect against default on $24.6M in debt to be paid back in ETH, and keep in mind we can only liquidate 1/3 of the Pool Cover in case of a borrower defaulting so that means we can incur a default of up to ~$110k before we as lenders get hurt by a borrower failing to repay their loan. That assume less than a 0.5% default rate. And to be fair, only one borrower has defaulted with Maple despite the 3AC contagion–Babel who borrowed from the Orthogonal-run USDC pool.
  • As Dopeee mentioned, we can capitalize on the continued MPL rewards which enables us to earn more than the standard 4% ETH staking rate. I’m assuming based on the history of loans that Maven will continue to charge market-competitive rates for borrowing ETH at 4% or higher and then the MPL we earn, can be conveniently sold into 50% USDC + 50% xMPL and even better, soon you can stake xMPL as Pool Cover to earn an additional share of the 10% of interest Pool Cover providers earn, while also receiving 2.98% APY in MPL from the MPL buyback vested to xMPL holders.

In the end, we’re earning yield in 3 forms:

  • Borrower interest paid in ETH
  • MPL rewards (a liquidity mining reward) getting us to the estimated 7-7.7% APY
  • Compounding MPL rewards by keeping at least 50% of MPL as xMPL and earning 2.98% APY in MPL
  • Additional share of interest in ETH paid for depositing xMPL as Pool Cover
1 Like
14

I have a few concerns to add here:

  • AFAIK, yields have dropped since this was originally posted so double-checking if we’re still confident in earning anywhere north of 7% APY after fees?
  • Currently, when you signal withdrawing from Maple as a lender, you give up MPL rewards from that point onward, which makes sense. However, i’ve seen complaints of lenders trying to withdraw with a loan being paid back but apparently, some are running bots and so it’s a race to reclaim liquidity.

My concern is can we and will we need to retrieve liquidity and be challenged by this?

15

Even though you mention Maple/Maven’s fee earlier on in the post, I think you are not taking Maple’s fees into account when calculating what Nexus will earn.

Maple charges an establishment fee of 0.99%. Assuming you roll this capital and do two sets of 180d loans, that means you’d pay those establishment fees twice (40mm * 0.99% * (180/365) = 195,288 per every 40mm you originate. For simplicity in the screenshot below, I assume you do 1 set of 365 day loans (for a total of 396,000 in fees).

Then there are also “ongoing” (performance) fees charged by Maple and Maven. I am a bit confused on how fees work here. Per Maple’s “pool delegate income calculator”, it seems the default “ongoing” (performance) fee is 20%. Right now, Maven 11’s ETH pool page, only makes it clear that Maven is getting 10% of that performance fee, but does not clarify what the total performance fee is. You do say “The pool delegate (Maven11) earns a 10% fee from the pool of interest earned” so maybe the Maple UI is just confusing and in reality, the total “ongoing” (performance) fee is 10% here.

In the screenshot below, I assume the total ongoing (“performance”) fee is 10%, but maybe the Maple UI is confusing and actually it’s actually 20%, but Maven is only getting 10% of the 20%.

On top of that, there are also “pool cover” fees. I am not sure how those work or where those are paid from. Is it that pool cover == staked maple and this comes from the portion of the performance fee that is not paid to Maven 11? Or are these fees coming from two different places?

CleanShot 2022-08-17 at 11.38.22
CleanShot 2022-08-17 at 11.38.22801×554 28.5 KB

16

Yes, @DeFi_Dad please see my comment above. I could be wrong (definitely could be wrong), but I believe @Dopeee 's original post does not even take Maple’s fees into account. In addition to that, it’s not totally clear how the rate paid to lenders is calculated as right now the Maple UI says lenders are earning 4.11% (inclusive of MPL token emissions). Is it that the MPL token price has dropped a lot since this post was originally made and that is what is causing the big delta in the rate earned by lenders?

17

Hey,

All of the fees that you mentioned are being paid by the borrower not the lender (Nexus in that case).
The APY that you see is net. Currently, it is lower than the target because we still did not issue all of the loans with the fresh cash so that the capital that is not utilized at the moment is temporarily dragging the APY lower.

18

Thanks for the response!

Can you breakdown how exactly the “ongoing” (performance) fee and the cover fee work? I don’t understand this.

I understand how the establishment fee can be paid by the borrower, but not exactly how the ongoing fee is paid by a borrower.

And to be specific, let’s say Wintermute borrows 9,000 wETH (16,534,148) for 180 days at a 5.0% interest rate.

  • At the conclusion of the loan Wintermute will owe interest of (9000 wETH * (180/365) * 5%) = 221.91780822 ETH + principle.

  • Wintermute will also pay an establishment fee of (9000 wETH * (180/365) * 0.99%) = 43.93972603 ETH

  • And then Wintermute will pay an additional “ongoing” / performance fee on top of that? How this this part work and what are the exact portions paid out / to where?

thx again!

19

Its pretty simple but maybe the performance fee name is a bit confusing.
The pool has the following structure of interest payments:

  • 80% of interest goes to lenders
  • 10% goes to pool cover providers
  • 10% goes to the pool delegate

On top of that, borrower pays an extra 1% annualized establishment fee, 2/3 of that goes to Maple treasury to buyback MPL tokens on the open market and distribute to xMPL stakers. 1/3 of that goes to the PD.

When the webapp displays 5% loan to Wintermute = Wintermute pays 5% + 1% establishment fee.
Lenders receive 80% of 5% per annum = 4% net yield in wETH on that particular loan.

Hope that clarifies your question :slight_smile:

20

I see. This makes a lot more sense. Thank you for clarifying.

Okay, so in summary…

  • Establishment fee is paid by borrower, meaning on a 9,000 wETH, 5% 180d loan, Wintermute pays an upfront establishment fee of (0.99% * (180/365) * 9,000 wETH)

  • Ongoing/performance fee is “paid” by the lender in the sense that its interest that would otherwise be going to them, but now is not. Said a different way, this is not an additional fee the borrower is paying.

  • The ongoing/“Performance” fee is broken up into “Pool Delegate Ongoing Fee” and “Cover Ongoing Fee”. On the UI for your ETH pool, it looks like each of these is 10%, meaning 80% of yield goes to lenders, 10% to PD, 10% to cover (which is also Maven in this case, right?)

  • When @Dopeee says above

It seems he is accounting for the fact that Maven’s weighted average of loans will be 5% (before fees) and then is factoring in that into his original estimate.

21

Yes,

All points are correct, it is right to say that the lender pays part of the interest to providers of the pool cover (10%) and another part (10%) to credit underwriter (Maven 11).
Indeed, the quote that you are referring to is accounting that the average interest rate will be 5% (Exclusive fees) so that the net yield in pure wETH will be 80% of that = 4%. MPL rewards are added on top. Originally, we planned the deposit couple months back so the rewards were higher than they are right now (plan is to phase them out with time so that the token is not a sell only asset but rather a way to participate in revenue sharing of the protocol). The rates on ETH are spiking tho, therefore we expect the weighted average to creep up to levels above 5% in the near future.

2 Likes
22

After the 3ac and contagion on lenders i’m less confident that maple yield is worth the risk. Would vote against and worth reconsidering imo, eth staking yield seems safer and positive expected value in comparison.