Temperature Check: Aera for Compound

Ideas
1

The Gauntlet team recently launched Aera, an autonomous treasury management protocol. This post is a temperature check to assess the community’s appetite to trial Aera for a portion of Compound v2’s reserve funds.

Introducing Aera

Aera is a solution for optimizing DAO funds autonomously and on-chain. For most DAOs, insurance funds (e.g., reserves, treasuries, safety modules, backstops) are not actively managed or adjusted based on market conditions. For DAOs, this can lead to an inability to maintain runway, cover liabilities, and benefit from growth in the market. Traditional institutions can allocate funds to more nimble managers who make day-to-day decisions, but DAOs face numerous challenges with this model including governance and creating strong incentive alignment with external managers.

Aera provides DAOs with a one-stop solution for managing insurance funds efficiently and transparently. The Aera protocol consists of vaults, which are constructed on a per-protocol basis and can hold a combination of stablecoins, native tokens, and other cryptocurrencies. The objective function of the vault is determined by each DAO and is highly customizable ranging from simply keeping fund proportions in line with borrows, to complex hedging strategies using on-chain options. Vaults are automatically rebalanced by multiple actors (Guardians and Arbitrageurs) who compete on-chain to propose the best combination of assets in the portfolio. This ensures that the vault objective is met across a wide range of market scenarios and time horizons.

For more detail, here is a quick video that walks through how things work.

Product___Aera_Finance
Product___Aera_Finance3088×1246 148 KB

  • Guardians are experienced risk analysts and can be institutions or individuals. They compete by periodically submitting asset weights to rebalance the vault, which evolve over time to keep up with the objective and market variables. Their submissions are aggregated amongst other Guardians and weighted based on historical performance. To participate, Guardians must stake their own assets and reimburse the vault if their decisions underperform.
  • Arbitrageurs in the open market execute transactions to rebalance to the Guardians’ preferred allocation to earn profit

Key Benefits

The Compound v2 market on Ethereum has accumulated over $40M of reserves, which are now primarily in the form of USD stablecoins ($22M DAI, $14M USDC, $3M USDT), WBTC ($3M), ETH ($1M), and BAT ($1M). These reserves are static and don’t track risk exposure. In previous discussions on reserve management, the community has shown interest in reallocating reserves to a more optimal asset mix. With the launch of Aera, there is now a purpose-built solution that simplifies and automates much of this optimization.

A few key benefits for Compound:

  • Aera helps to minimize bureaucracy. The DAO doesn’t need to plan strategies, but rather just pick assets.
  • The Aera protocol continuously rebalances Compound’s fund portfolio based on actual liabilities and market conditions
  • Aera allows for coordination of a decentralized network of actors working together transparently on-chain to optimize the vault

How It Works

  1. Compound governance deposits a portion of reserves into an Aera Vault
  2. Compound governance works with a Guardian to set the objective and selects a set of assets for the vault. Gauntlet will serve as the initial Guardian with all fees set to zero. Later in 2023, a new version will launch with the ability to assign new Guardians and enable vault fees to promote Guardian specialization and competition.
  3. Guardians and Arbitrageurs continuously rebalance the vault based on the objective and market conditions
  4. Compound can view vault performance at any time through the public dashboard and have instant access to funds due to the self-custodial design of Aera

FAQs

  • What does this cost?
    • Free during trial phase. Fees enabled in late 2023 during scaled rollout phase.

Next Steps

It would be great to hear the community’s feedback on trialing Aera and happy to answer any questions – please comment below. If the community is supportive of a trial, we will follow up with a detailed proposal to vote on.

8 Likes
2

We support the idea of diversifying and optimizing the treasury, as this is crucial for the sustainability of the protocol in the long run. From our perspective, it would be great to first launch it with one specific treasury asset (like the one llama proposed for $CRV on AACE).

We would be happy to vote for it if the mechanism has been tested by the market while generating consistent yields.

3 Likes
3

This is a very innovative product with the potential to automate treasury risk management and other novel use cases. I would support the idea of Compound exploring it initially on a limited basis. However, as Compound gets its feet wet, I would also like to see additional details as follows:

  • Frequency of weight submissions by the Guardians has an impact on transaction costs incurred by the DAO. I think it would help to know the comparative cost estimates, say for daily submission vs. weekly submissions.

  • The AERA concept is fairly straight-forward under normal market conditions. However, things can get complicated under volatile markets. It would be helpful to see examples addressing scenarios involving multiple rounds of mid-epoch trading by Arbitrageurs, price ”whipsaw”, reversion to the mean etc.

Additionally, one general observation regarding the use of options in a framework such as AERA’s is the following. (I understand that Gauntlet is not proposing to use options in the initial trial for Compound, so this discussion may not be immediately relevant)

My concern is that having options as part of the asset mix during volatile markets can lead to adverse results (in terms of portfolio value), given that Arbitrageurs can trade any number of times as prices whipsaw, with their primary concern being correct asset allocation and arbitrage profits, and not overall portfolio value. To prevent loss in portfolio value, Guardians would be required to predict not only the underlying asset prices correctly (where they would end-up by the end of the epoch), they would also need to predict the trajectory of prices through the epoch, so that they can estimate implied volatility and option prices. When underlying prices revert to the mean, or if volatility suddenly subsides, option prices can collapse, locking in losses. This means that Guardians may be correct in predicting underlying prices, but can still incur slashing. In theory, they are supposed to be experts and be able to predict even options implied vol, but it’s not clear how this will work in practice if the options markets are illiquid and bid-ask spreads are too wide. However, on the other hand, without using such hedging instruments, protecting against crashes may be difficult.

Options are generally cheap just before big price moves, and not optimal to deploy once volatility has already taken afoot. One way to deal with this situation is to run the epoch based on some chart events on a reference asset such as ETH (say, a moving average crossover, which is blunt with false positives, but more sophisticated ones can be constructed) rather than on a periodic basis such as weekly epochs. This is just a thought.

3 Likes
4

Thanks @dtalwar for the post!

We are in support of trialling Aera for Compound especially as multi-chain deployments increase and subsequently, so will reserves.

Assuming the temperature check passes, we look forward to a detailed proposal outlining the initial TVL, objective function, and asset selection!

2 Likes
5

Thanks for this initiative. I’m generally supportive of having DAO treasuries actively managed to allocate capital efficiently and help maintain runway. However, I’m not sure whether this specific implementation would be the best way to do it. DAO treasuries should work more like endowments: focusing on keeping the principal amount intact through close management of underlying risk factor exposure.

The concept of a “decentralized network of actors working together” seems appealing at first sight, but complexity opens the door to unexpected outcomes:

  • What happens in an extreme market event e.g. sudden price drop; depeg; protocol hack? What if none of the depositing DAO executors was aware on time to withdraw the funds?
  • What happens when a vault guardian amasses a significant amount of voting weight on the aggregation function to a point where it becomes economically viable to collude with arbitrageurs and drain the CFMM pool?

Has Aera been tested before? If so, could you please share any meaningful results?

1 Like
6

Thank you for all the thoughtful questions and comments!

The objective function (e.g. yield generation, maintain runway, etc) will ultimately be decided by the community for the pilot. We’ll share more details on this in an upcoming forum post. Aera has been audited by Spearbit and has been in testing for the past 6+ months on Polygon.

For context, there are two transaction costs - (1) gas fees, which are paid by Guardians and Arbitrageurs and (2) arbitrage loss, which is incurred by the DAO. It’s difficult to provide cost estimates for daily vs weekly submissions given the amount of variables involved (e.g., the asset, along with its historical volatility and on-chain liquidity, etc). Once the community selects the assets for the pilot (more details on assets and the selection process will be included in the upcoming forum post), we will determine the optimal epoch length and can share this data with the community as well.

You are correct that things get complicated under volatile conditions. It’s worth noting a few things about the incentives and roles of Guardians and Arbitrageurs during these conditions. Guardians are incentivized to predict future volatility and incorporate this into their submissions to avoid incurring losses. Arbitrageurs are constantly rebalancing the vault to achieve the optimal weights - this is similar to a Balancer pool. Arbitrage loss would be equivalent to a Balancer pool and Arbitrageurs pay fees to the DAO.

As you mentioned, options will not be part of the pilot. Worth noting that options will be over collateralized, which reduces the likelihood of adverse events resulting in volatility. There would also be constraints on the percent of the portfolio allocated to options - this would also depend on the DAO’s objective function.

Our highest priority with Aera is building a trusted product with a strong track record given that we are working with treasuries. For this reason, we are launching small pilots (like this one we are proposing with Compound) to complement 12+ months of rigorous research and testing where we’ve simulation extreme market events and built an incentive mechanism that responds to volatility.

The design of the protocol is such that the staking and slashing costs of collusion are extremely high (the whitepaper unpacks the math behind when and how this would happen if you’re curious). Aera will start off with a high collateral ratio for the amount staked by Guardians and will reduce this ratio over time as the protocol matures.

2 Likes
7

@dtalwar Thanks for your reply. Could you please share the results of this testing period so far?

1 Like
8

Thanks @dtalwar! Can you elaborate your point 2 above? Is it just price slippage or there other reasons too? An example would help.

Whitepaper (p. 19) says the following: “The goal of the parameter submitters
is to enable the strategy utilized on-chain to be dynamic and competitive, reducing the
overall adverse selection and transaction cost for a DAO.
A unique challenge faced by vaults offering dynamic rebalance strategies is deciding on an
optimal rebalancing frequency. If a vault rebalances too frequently, then it will realize losses
from excessive transaction costs.”

The above seems to be referring to the impact of submission frequency on the transaction costs and related losses. Can you provide more details?

1 Like
9

Even without using options, volatility can wreak havoc. Last 24 hours is an object lesson. Due to Cirlce’s exposure to SVB, USDC ( and thereby DAI) have depegged from dollar. Now, FDIC is known to act with extreme efficiency, and the most likely outcome on Monday or Tuesday morning will be an announcement that SVB depositors will be able to recover most, if not all, of their deposits. This will probably cause the peg to be restored at a very rapid clip. Under AERA’s regime, if Arbitrageurs had rebalanced due to depeg, they may not have enough time to rebalance to adjust to the repeg when the announcement hits, and can cause losses to the portfolio. This again means that even if the Guardians are correct by the end of the epoch in their price predictions, they may still get slashed because of loss in portfolio value.

This points to the core limitation of AERA model: the incentives of Guardians and Arbitrageurs are different / misaligned. Guardians want to protect the value of the portfolio over a longer timeframe (duration of the epoch), while Arbitrageurs only care about asset allocation weights and immediate profits.

Of course, losses to the portfolio are recovered by the DAO (not sure if this make-whole?) by slashing the Guardians. But then, this raises the question if Arbitrageurs are truly adding value or making things worse, and if it’s possible to create a model where only Guardians exist. In this model, if the Guardians are wrong, they get slashed, but Arbitrageurs’ actions won’t cause additional losses for them.

In my view two aspects of AERA are worth examining further:

  1. If it’s possible to create a system where only Guardians exist, or with two classes of them: Submitters vs. Executors, and both do staking.
  2. If it’s possible to run the epochs not periodically, but based on some chart events that approximately portend a significant price move.
2 Likes
10

Hi, I’m Ignacio from Avantgarde Finance.

We’re excited about Aera. However, we’d also like to put forward an approach to managing these funds and I’d imagine others would like to put forward other approaches too. We’re wondering if an open application process would make more sense.

We’d recommend that an RFP goes out for a period 2-3 weeks giving people the time to propose alternative approaches. At that point, the DAO votes on the 2-3 approaches it likes best with a path to ramping up exposure over time once the approach is proven & tested.

To further support the idea of an open application process, we can draw inspiration from the successful approach taken by the Auditing Compound Process, which was won by OpenZeppelin.

11

Of course. We have been running a vault on Polygon (submitting weights daily) with $220k of our own funds. This vault models the loanbook of another client and was positioned as insolvency coverage. The portfolio has largely maintained a composition of 90% ETH and 10% USDC as the risk in the loanbook of the client was low and thankfully had no insolvencies or drawdowns that caused a liquidity crunch during this period. The vault is valued at around $250k today.

Yes - it is price slippage and loss versus rebalancing.

The point is for them to have opposite incentives. The Arbitrageurs have a low time preference and the Guardians have a high time preference. By being as close to zero sum as possible, you ensure that colluding cannot drain the vault. If they had the same incentives, that would effectively mean they could drain the vault.

We have deliberately chosen to bring Aera to market slowly through rigorous research and testing both internally and through very small (and free) external pilots like this one to inform development and get feedback from the community as we build. Given Aera’s stage (pre-launch, mechanism and fee model still in development), there isn’t a great way to compare it to other options.