Lido for Aave - Proposal by Delphi Digital

[lukedelphi]

Lido has established itself as the leading solution for ETH staking, and is quickly expanding into other PoS chains (e.g. Terra and Solana) and leading dApps. As investors and believers in Lido, Delphi Labs is excited to propose a liquid staking solution for AAVE.

We want Lido to be the default choice for anyone considering staking to any of the main PoS chains and dApps. This can only be achieved through wide coverage and brand trust, earned through Lido integrations consistently providing significant advantages over direct staking. Aave is the biggest DeFi app by TVL with currently $21B locked, so to achieve this vision a Lido liquid staking integration for Aave is important.

This development proposal will be suggested to the Lido DAO. We will give a few days for feedback on this forum thread and then initiate the on-chain vote where LDO holders will be able to voice their support.

Drawbacks of current solutions on AAVE

There are currently two main ways for users to stake their AAVE tokens: 1) through the staking contract provided by Aave, and 2) through 3rd party staking services, such as xToken’s xAAVE contracts.

When staking with AAVE’s contract, users receive 1 stkAAVE token for every 1 AAVE token staked, with the staking rewards accounted for separately. Users who wish to compound their gains need to actively claim the rewards and restake which can incur very high gas costs.

When staking with xToken, users get back a certain amount of xAAVE tokens for every AAVE token deposited. As staking rewards are accumulated, the exchange rate between AAVE and xAAVE changes; specifically, each xAAVE becomes worth more AAVE. This design may be less intuitive for users new to DeFi.

Moreover, xToken’s contract does not allow xAAVE holders to vote in Aave governance. Instead, the xToken team votes on behalf of users. For this purpose, two variants of xAAVE tokens are created: xAAVEa and xAAVEb, which votes on the basis of left- and right-wing political ideologies respectively. We consider this a less than ideal solution as not only are the users stripped of the opportunity to voice their own opinions, but also someone has to interpret what ideology any given vote corresponds to, with that governance power currently being concentrated into the hands of a few xToken team members. This is an especially serious drawback in a money market protocol, in which governance decisions determine the balance between capital efficiency and protocol solvency. Stakers underwrite protocol risk and are on the hook for up to 30% of their principal in the case of Shortfall Events, giving them skin in the game incentives to participate in governance.

Another adverse effect of splitting xAAVE into two variants is the fragmentation of liquidity. As there is a 10-day cooldown period for unstaking, xToken created 50%/25%/25% xAAVE/AAVE/WETH pools on Balancer, so that users can off-ramp using the exchange without waiting. However, since liquidity is split into two pools containing each of the two xAAVE variants, each pool is shallower and large trades may incur slippage.

Our solution

Upon examining the drawbacks mentioned above, we propose the development of a liquid staking solution for AAVE (stAAVE) with the following characteristics:

• The staking rewards are automatically compounded;
• The balance of stAAVE token in each user’s wallet automatically updates every block, reflecting the staking rewards accrued (similar to stETH);
• Holders of stAAVE can vote in Aave governance, the same way as stkAAVE holders can;
• Low-slippage trading between AAVE and stAAVE utilizing a Curve pool.

Note: stAAVE will have to be approved by Aave governance as a voting token through a poll before holders will be able to vote directly. Before this is accomplished, the Lido DAO has the power to decide how to vote in each Aave governance poll. Lido DAO can relinquish this power at any time.

In addition, as the Aave Safety Module naturally evolves over time to become a sophisticated insurance product (as per our latest proposal), stakers will have additional options to underwrite specific markets and take on differing levels of risk. We envisage our stAave product streamlining this process for users.

Timeline and Future Work

• Completion of smart contract development [May 2021]
• Completion of two audits and further testing [June - July 2021]
• Mainnet release [July 2021]
• Lido governance proposal to incentivise liquidity for stAave pool [Q3 2021]
• Aave governance poll to approve stAAVE as a voting token [Q4 2021]

Suggested Fee and Incentive Structure

We split the fee and incentives into three parts to best align us with future success of Lido, the success of the stAave integration and help us with incurred costs.

• Lido Token Incentives: Using vested tokens distributed according to milestones reached.
• Revenue Share: An ongoing revenue share between Delphi Labs and the Lido DAO.
• Delivery fee: To help us cover costs incurred with development and bringing the integration to market.

Each of these are outlined below.

Lido Token Incentives

To align us long-term with the success of the Lido DAO we propose the following terms:

200,000 LDO tokens issued with a 1 year cliff and 1 year vesting when Lido for AAVE manages to capture 2.5% of the staked Aave supply.

Revenue Share

An ongoing revenue share aligns incentives between Delphi Labs and the Lido DAO whereby generated fees are split with 20% of fees generated paid to Delphi Labs and 80% to the Lido DAO treasury.

We believe that this simple structure is best suited to incentivize independent investment and expansion of integrations on our end by tying our revenues to those generated for the DAO at large.

Delivery Fee

While the above incentives align us with the success of Lido DAO and our Aave integration, an amount paid on launch of the implementation would be helpful to cover costs associated with development and bringing the integration to market.

For this we propose $60,000 to be paid in ETH from the DAO treasury once smart contracts are audited, deployed to mainnet and the staking widget is integrated into Lido’s UI.

Model

To help understand the fees at certain percentages of stAave market share we have produced the following model.

At current prices, given a 1% performance fee and 20% market share for our stAave product, the model shows we will generate $122k for Lido per year, repaying Lido’s investment in approximately 4 years. In a worst case scenario of 2.5% market share (the minimum threshold at which we receive our LDO tokens), Lido would make back its investment in ~15 years. We also believe the 1% performance fee can be increased over time as stAave becomes adopted.

Next Steps

We have already been in contact with various stakeholders from the Lido and AAVE ecosystems and started the development of our proposed Lido liquid staking solution for AAVE.

Our next step is to gather more feedback from LDO holders and other ecosystem members on this proposal and the proposed spec, which we are encouraging with this forum post.

On June 7th, we will issue a Snapshot vote to determine whether the Lido DAO is in favor of supporting our proposal. If the proposal passes, we will speed up development of the Lido for AAVE program in accordance with the spec and iterate on the spec with the goal to release a useful version to integrate into AAVE.

Why Delphi Labs

Delphi Digital has built a strong reputation as the leading independent research boutique covering the digital asset market, with our client base including some of the industry’s most prominent funds, financial institutions and investors. Through our consulting services (now Delphi Labs), we’ve worked closely with top projects in the space such as Aave, Synthetix, Axie Infinity and others to model, optimize and redesign their economic models.

More specifically, our team deeply understands AAVE’s safety module, having spent significant time vetting and contributing to AAVE’s insurance model and token economics to refine the viability of the design (see our proposals here and here). Having an intimate understanding of how the AAVE network is designed and functions will aid us in our development of stAAVE.

Was are also investors in LIDO at Delphi Ventures, having contributed 303 ETH to the Lido DAO in the recent on-chain purchase alongside Paradigm, Coinbase Ventures, and others. We are big believers in both Lido DAO and its founders, who we consider some of the most impressive founders we’ve met in the space. We intend for stAave to be the first of many products we build for Lido in the coming years.

[JK_stakefish]

Interesting proposal! I think it’s great to see Lido expanding to making all tokens liquid.

• Has this been proposed directly to the Aave team? The stAAVE product could be rendered obsolete if Aave changed the way their stkAAVE token model worked. I’m curious if this is a possibility.

• How does the re-balancing of stAAVE token work? Can this be easily fetched from the Aave’s staking contract?

• Would you be open to applying this same model to tokens that function similarly (like SUSHI)?

[monet-supply]

@JK_stakefish

Would you be open to applying this same model to tokens that function similarly (like SUSHI)?

I don’t this this would be suitable for xSUSHI because there’s no lockup and the rewards are auto compounding. stkAAVE doesn’t have these features so might benefit more from a wrapper.

[lukedelphi]

@JK_stakefish thanks! Yep we’d like for people to always go to Lido when they want to stake as they assume there’ll be an integration and that it will have advantages to them. To answer your questions:

Has this been proposed directly to the Aave team? The stAAVE product could be rendered obsolete if Aave changed the way their stkAAVE token model worked. I’m curious if this is a possibility.

Yes we have presented this idea to the Aave team previously and they are aware we’re doing it. We are also actively involved in discussing Safety Module economic designs with the Aave team and community, and have made several proposals on this front (see here and here). We believe changes to the Safety Module are more likely to increase the utility of a solution like stAave by introducing additional options for stakers, each with their own risk/return. At this point, similar to other Lido staking tokens, stAave will have to provide a baseline staking strategy for users (or a few, but this comes at the cost of liquidity fragmentation)

How does the rebalancing of stAAVE token work?

It works similarly to stETH - when a user deposits AAVE to the contract, he is assigned a certain amount of “shares”, which represent his share of the contract’s NAV. The NAV is calculated as the amount of staked AAVE + staking rewards to be claimed - performance fee to be deducted. As rewards are accumulated, NAV grows, while shares remain constant. When a user queries balanceOf, the contract calculates NAV per share, and simply multiply that with the user’s shares.

Can this be easily fetched from Aave’s staking contract?

Yes, we only use on-chain data for calculating the NAV, which can be queried from Aave’s contract and is updated automatically every block. No oracle is needed, unlike stETH.

Would you be open to applying this same model to tokens that function similarly (like SUSHI)

Yes, we are already considering other projects to build similar products for. We have our eye on SNX, although we’d like to wait until their staking architecture is finalised. We’re also considering creating staking tokens for Anchor and Mirror on Terra, both of which don’t issue tokens to stakers despite having over $2b in combined tVL.

While an stSushi would be interesting, as @monet-supply says xSushi is already auto-compounding since fees are used to buy sushi and added to the xSushi staking pool. It’s also widely accepted as collateral, including on Aave itself.

[calchulus]

This is very interesting to ensure maximal capital efficiency

[eliasimos]

thank you @lukedelphi —this is very interesting as to @JK_stakefish ‘s point it could open up to more of this work accross defi protocols.

my main question is around the payback period you project. what param changes to the model would reduce payback by as much as 75% (ie 1 year)?

initial reaction to 4-15 years payback is that…it’s long! repeated across multiple protocols it just pushes a lot of realized value for Lido deep into the future—which high level could mean a lot of OCC incurred.

[lukedelphi]

Hey @eliasimos. That’s fair and we had the same concerns initially. We kept the model quite conservative but there’s a fews ways the integration could pay back sooner than modelled:

1. There’s the potential to raise the performance fee from 1%. Given xToken’s integration charges a 1% performance fee + a 0.1% mint/burn fee, we believe pricing higher than that this would be a mistake initially as the goal is to grow market share. However as TVL in the pool, liquidity, integrations and brand awareness of Lido grow it would be fair to propose that the fee is raised as people will be prepared to pay a premium for their trust in Lido (see recent xToken hack for example). You might point out that the difference between compounded and non-compounded returns does not allow room to raise fees, but for many people there would be tax benefits to this approach which still makes it worthwhile.

2. We expect that the Aave insurance module will become more complex in the future, with stakers having additional options to underwrite specific markets and take on differing levels of risk. Our stAave token would make those decisions on behalf of stakers through this integration and for this a higher fee could be taken since it’s a higher value service.

3. We expect Aave to grow significantly and with it the value that each percent of market share represents.

These things are all pretty speculative and difficult to model however so we kept it conservative to show that even in a worst case scenario the ROI is positive.

But aside from those points, we believe there’s value in Lido establishing itself as the go-to staking venue for all leading protocols and Aave is undoubtedly one of those. Over and above the ROI, we believe there is still significant brand value is covering all these protocols which can pay dividends in the future.

Edit: xToken’s mint/burn fee is often overlooked, and can add to user’s cost.

Say someone deposits 100 AAVE. He pays 0.1 AAVE mint fee upfront. With the current 6% staking reward rate, he earns 6 AAVE rewards in a year, for which he pays 0.06 AAVE performance fee. The total cost for the user is 0.16 AAVE, which is effectively 2.7% on the 6 AAVE income.

This strengthens the argument for raising the stAave pool’s fee in the future.

[vsh]

Main points in favor of this to me are:

• aave staking, IMO, will be getting more complex and involved with time, which warrants a niche for liquid staking here
• I am pretty sure we’ll see more growth in Aave TVL/fees over time

This is a venture to me, with significant risk, but one that makes sense.

[FelixLts]

Agree here, AAVE will likely grow and stay one of the core DeFi protocols. To have a team like Delphi that is as familiar and will stay in touch with their tokenomics work on a liquid staking solution for Lido feels like it will likely be long-term beneficial for LDO holders.

Thanks for putting this proposal forward!

[Drag0nflai]

I agree with @eliasimos that a 4 year payback is way too long. The fee model presented shows a 29~30% fee will be charged by Lido where 80% goes to Lido and 20% to Delphi as ongoing revenue share. Is this correct?

The question is what value-add there is and what to anticipate in user adoption over the long term. @JK_stakefish asked some very good questions. Based on my understanding, as an AAVE holder, staking directly on AAVE

Pro: more flexibility of underwriting specific markets based on my risk tolerance

Con: gas fees to compound interest

Staking via Lido

Pro: auto compound feature without incurring gas fees (but with a Lido staking fee of 30%)

Con: loss of flexibility in choosing a staking strategy that is the most suitable for the user

Have I missed anything @lukedelphi?

Currently the gas fees have come down significantly and if gas fees continue to trend down lower, I wonder how this solution would gain much traction as I expect a lot of AAVE stakers would prefer to stake directly via AAVE and periodically pay gas fees to compound reward. This is the first proposal I have read that is about a dApps staking solution. In my humble opinion, PoS staking often comes with many more limitations for ordinary users that Lido is in a great position to address (e.g. the 32 eth min). But in this instance, this seems to be a high cost project (one-off delivery fee; performance fee; plus ongoing revenue share) for solving a short term problem of high gas fees.

[frontalpha]

I am questioning whether this proposal expands Lido’s mission beyond what was intended.

While Aave’s staking module has the word “staking” in it this product seems more suitable as a yearn strategy, or something similar. At a glance, this proposal might suggest Lido take an active approach as Aave creates more sophisticated staking options in the future. So when you say, “We envisage our stAave product streamlining this process for users” what does Delphi Digital think that will look like? Would Lido incur overhead managing Aave insurance products for users?

I also looked up what happened with the snapshot vote that was supposed to happen June 7th, but couldn’t find it. Am I looking in the wrong place? Snapshot

Also, are there any other communication channels between Delphi Digital/Aave/Lido regarding Aave staking that the public is welcome to join?

[lukedelphi]

Hey @frontalpha I’ll leave to others to comment on your main point. The snapshot wasn’t launched on June 7th as we wanted to give more time for responses given many people were in Miami. However, I’m planning to create the snapshot vote today and will update here when done.

[frontalpha]

@lukedelphi anything new with the Aave staking project since the vote?

[lukedelphi]

@frontalpha everything is built - just waiting on the all clear after fixing issues raised from the Quantstamp audit, then we can deploy to mainnet and move forward. There then needs to be another discussion around whether we want to use LDO to incentivise AAVE to move over, and to provide Curve liquidity for stAave, but we’ll cross these bridges when we come to them.