Continue the discussion on Treasury Diversification

Hi all, this @moss and Ainsley, posting from Avantgarde Treasury (the Treasury Management arm of Avantgarde Finance, core devs of Enzyme).

We have been reading the interesting discussion on Treasury Diversification, initiated by @MC_MARK and we are eager to actively participate in the conversation and potentially assist in achieving tangible DAO objectives.

Our takeaways from the previous thread

Some of the suggested diversification options from the DAO members include:

  • Diversifying stablecoins: consideration is given to diversifying between stablecoins such as USDT, USDC, and USDP. This approach aims to reduce reliance on a single stablecoin and spread the risk across multiple options.

  • US Treasury bills: while they are considered low-risk options for gaining yield on USD holdings, there are concerns about locking up liquidity for an extended period, which may not align with the DAO’s objectives and could generate some unwanted regulatory risks.

  • Exposure to Bitcoin and Ethereum: while these assets carry higher risk compared to other instruments, they are viewed as having long-term growth potential and being better aligned with the DAO’s ethos and objectives.

Overall, the thread shows broad agreement on stablecoin diversification and accumulation of bluechip assets, but the specific approach and composition are still under discussion within the Merit Circle DAO community.

We think our expertise + infrastructure could be helpful. :point_down:

A useful summary of our portfolio construction framework applied to stablecoins

This section is curated by our Head of Asset Management Ainsley.

We believe that for any universe of assets, the same three components are needed as inputs into an informed portfolio choice: Expected Returns, Risk, and Diversification.

To give a flavour of how we think about each of these key inputs, in this section, we will focus on USDC, USDT, USDP, DAI, and LUSD as our stablecoin universe for illustrative purposes.

Expected returns of different assets are often the only thing that less informed traders focus on. However, they are also the most difficult component to pin down - in equilibrium, the future movement of asset prices is necessarily uncertain since any certain profits available in the market are rapidly arbitraged away. With this caveat in mind, we can reference the current levels of yield available through supplying stablecoins in a blue chip lending protocol such as Aave as a starting point. The variable rates go through periods of volatility but the current range is between ~1.5% to 3%.

Yield volatility aside, there are potential downsides to holding the stablecoins over a period of time in order to accrue the income, which brings us on to risk, the second key input in our portfolio construction process. There are many risks to this asset universe that need to be considered. For the purposes of this summary we won’t be covering all of them and instead focus specifically on price risk - unlike government bonds, these stablecoins don’t return your capital at maturity, thus there is the risk that at the end of your investment horizon you sell them at a lower price than you bought them for. Despite having the word “stable” embedded in their name, the stablecoins had an annualised volatility ranging between 5% and 10% on daily data over the last 18 months (USDT being the lowest and LUSD being the highest) - for context, this is orders of magnitude higher than traditional cash-like investments such as short term Treasury bills or money market funds, and more in line with risky asset classes such as credit or corporate bonds.

Looking at more granular data illustrates that stablecoin prices can move significantly intraday. The chart above shows the hourly price moves over the weekend that Silicon Valley Bank failed during March of this year. Price moves do not always correspond to changes in underlying fundamentals but in this particular case the causation was more clear as the reserves backing some of these stablecoins were deposited by issuers in some of the affected banks.

This highlights another important aspect of our risk inputs within our portfolio construction framework, that not all risks can be captured by price volatility. A more holistic approach to assessing risk, including fundamental inputs such as the types of collateral backing each stablecoin in this case. The collateral breakdown from on-chain data and the latest attestations of the fiat-backed stables provides some clues to the drawdowns during the US banking scare in March - USDC and USDP have the exposure to bank deposits in their reserves in bank deposits. Despite not being a fiat-backed stablecoin in the traditional sense, DAI saw the second worst intraday drawdown during the episode, with a large exposure to USDC through its Peg Stability Module. Worthy of note is LUSD, which has no fiat collateral whatsoever, and despite having the highest price volatility of the stables in this analysis, was relatively resilient during the USDC drawdown (it actually moved to trade at a premium in the following days). This brings us onto the third component of our portfolio construction process, diversification.

As the saying goes, diversification is the only free lunch in investing - though opinions on perceived risks and expected returns will differ across market participants, the benefit of true diversification is a mathematical identity. The challenge is in measuring diversification effectively, since a naive equally weighted approach (whilst not without its merits) can miss nuances such as overlapping risks - don’t put all your eggs in one basket, but also don’t put them in separate baskets that happen to be tied together.

The traditional way to measure the diversification benefits of mixing different assets is through correlation of their returns. The correlation matrix above shows varying degrees of diversification benefit you can get across this group of stables: USDC, USDT, and DAI have shown more co-movement, USDP’s correlation to those three has been moderate, whilst LUSD is the least correlated to the others. This statistical look at diversification supports the fundamental evidence from the collateral breakdown earlier in this section that LUSD is diversifying to the other assets, a benefit that needs to be considered against its potentially less favourable characteristics on a standalone basis, such as its shorter track record, higher volatility, and more volatile collateral.

Hopefully this simplified illustration gives a flavour for our approach to portfolio construction. If there is interest in understanding our framework in more detail, including how we apply it to other assets such as BTC and ETH, we’d be happy to continue the conversation.

From research to action: an automated way to implement your desired framework

One approach to implement the desired portfolio for the Merit Circle DAO treasury could involve leveraging an automated agent controlled by the DAO in a non-custodial manner. This agent would serve as an automated tool to manage the treasury assets according to predefined rules and parameters set and voted by the DAO community. By utilising smart contracts permissions and decentralised vaults of Enzyme, paired with an automation like Gelato Network, the agent could autonomously construct and rebalance the portfolio based on the desired allocation percentages of stablecoins and bluechip assets.

The automated agent could continuously monitor market conditions thanks to the built-in robust Chainlink price feeds and adjust the portfolio composition accordingly. For instance, if the allocation of stablecoins deviates from the desired proportions, the agent would automatically execute trades to rebalance the portfolio. Similarly, when the agent detects favourable market conditions for acquiring bluechip assets within the specified risk parameters, it could initiate transactions to gradually accumulate BTC and ETH (or Liquid Staking Tokens).

It’s important to stress that this non-custodial approach would ensure that the DAO retains full control and ownership of its assets at all times. The automated agent would operate within the predefined rules and guidelines set by the DAO, minimizing the need for direct intervention by human actors. By relying on decentralized infrastructure, the implementation could provide transparency, security, and efficiency, aligning with the ethos of the DAO. Also, thanks to the integration of several sub-graphs from The Graph, Enzyme’s UI would give a 24/7 live NAV & transparent historical activity log of all transactions executed.

Also, it’s vital to note that implementing such an automated agent does not exclude that a human operator can be added in parallel and able to take over in case of “emergency” (however that notion is defined). Enzyme’s smart contract roles & permissions would allow the trustless delegation to human operators too that can act quickly and efficiently in case of need. Additionally, community consensus and governance processes would be crucial in defining the rules and parameters for the agent’s operation or the occasional human intervention.

We Avantgarde Finance, as the core developer of Enzyme, are uniquely positioned to deliver an auditable automation solution that is customised and fully aligned with the parameters defined by Merit Circle. If desirable, we could also go one step further and develop a more complex solution that reads the outcome of a DAO governance vote (e.g. a new portfolio construction) and executes orders on chain in an automated manner.

More broadly, leveraging an automated agent in a non-custodial fashion could provide Merit Circle DAO with a powerful tool to implement and manage the desired portfolio for its treasury, while maintaining control, transparency, decentralisation and efficiency in the decision-making process.

Closing thoughts

We’re happy to provide help and would love to hear your feedback. We are aware of the formal proposal process and the proposal template that you’ve outlined and we would happily follow the formal governance process if we saw positive feedback on these initial ideas.

Thanks for proposal @Moss! As you pinged me about feedback I will put it here.

I see that you propose to automate the portfolio management with some tools. My opinion is:

  1. To consider the automation DAO first needs to agree on general approach and rules to manage portfolio (which kinds of assets do we want, ratio, triggers for making decisions). It didn’t happen yet, so we have nothing yet to automate.
  2. Myself I would hesitate to use any 3rd party tools - it adds additional risks. Not just the risks of additional smart contracts but also the risks of human mistakes in entering the automation rules. Such mistakes might be rare but may cost too much. Also I expect that MC treasury management should be conservative - with simple rules and rare actions taking. So the automation should not bring too many benefits.
  3. I also didn’t like the part of message dedicated to stablecoins correlation. We diversify stablecoins mostly because of different kinds of risks associated with their issuers (Circle, Tether, MakerDAO etc.) and approach (centralized or decentralized/algorithmic). Price correlation analysis here makes no sense.

Hi @timour, thanks for taking the time to post your response!
We love to see some constructive feedback. Here’s how we are thinking about this:

Absolutely! Our goal with this post is to assist Merit in developing a robust framework and provide a transparent, accountable and trustless approach to implementing any framework you create. If needed, we are happy to support you in shaping your vision and ensuring transparency and auditability throughout the process.

Yes, we understand your concerns regarding smart contracts risks and we take this matter very seriously. We’ve been on mainnet for over 4 years without ever experiencing hacks or vulnerabilities. Additionally, we undergo thorough audits by top firms like Chain Security and we have implemented a generous bug bounty program in collaboration with Immunefi.

We acknowledge that automation, while efficient, may still have its limitations. That’s why we provide the option to have a human operator who can supervise the actions taken by the automated agent. This ensures an additional layer of oversight and minimizes the risk of unforeseen failures.

On the other hand, when it comes to human action, is it really less risky? We think it’s important to have at least transparent reporting and trustless smart contract rules in place to assign/revoke specific permissions for a human operator, as we recognize that - best case - mistakes can occur even without malicious intent, such as operational security accidents. And regarding the costs of human vs automation, human mistakes may be rare but may cost too much, potentially costing higher than a supervised automation framework. While automation is not the end-all be-all, we see it as a net positive addition to the human-only approach.

We fully acknowledge the limitations of correlation, volatility, and other price related measures in fully capturing the many risk vectors associated with stablecoins - price is only one of the tools in the analytical toolbox. For the purposes of the illustration outlined in the post, we aimed to give a flavor for how we think about the main inputs into a traditional portfolio optimization problem: expected returns, risk, and correlation. The main utility of the correlation statistics in this instance was to highlight that on the face of it, asset price movements seem to support the preceding fundamental analysis of the different underlying risk exposures.

We hope this clarification is useful to spur the internal debate and we remain open to receiving more constructive feedback like yours! :eyes: :ear: :slight_smile: