FIP-6 PCV Initial Stable Coin Investment

Thanks for pointing that out! I feel there is not enough emphasis on the fact FEI is trying to build a decentralized stablecoin. If the strategy is based on centralized stablecoins we can throw our vision in the same bin :confused:

1 Like

A couple of more things.

Regarding preserving decentralised ethos discussed by @joey and @DioDionysos. I think we should separate short-term and mid/long-term goals. In the short term the major objective is to survive possibly coming bear market/eth crash. For this purpose USDC is perfect. In the mid/long run more intricate PCV portfolio is needed. Then not only the decentralisation ethos is important, but if FEI is just a proxy for a basket of other stablecoins, then what’s its value proposition?

@Bruno Thank you for being responsive regarding USDT being unfit as a PCV investment target. And as a follow-up:

Let’s say USDT is swapped out of your proposal. Hence you’re proposing to buy USDC & DAI in equal amounts: $150M each. How do you come up with these figures? At the very least trading volumes (4.5B vs 1.2B) and market caps (16B vs 4.5B) are far from equal for USDC vs DAI, be it for de-risking or market making purposes, or even from the transactional POV.

2 Likes

PCV Diversification TRIBE Update Week 2

Just a quick refresh of the activities of this Tribe.

May 4 - Kickoff: we proposed the working group and started to work informally

May 12 - Week 1: we posted the proposal of an initial stable coin investment and continue to work on the Investment Policy Statement (IPS), focusing on bringing references and drafting the first version of it.

May 21 - Week 2: we are making some simulations based on the Modern Portfolio Theory applied to crypto to bring a benchmark to PCV management, we advanced in some specific parts of the IPS and discussed some ideas to deploy stable coins (we still need more time to bring a robust analysis to community discussion).

Regarding the IPS, we are considering the following structure: (i) Introduction, (ii) Objectives, (iii) Benchmark, (iv) PCV Cash Requirements, (v) Investment Period, (vi) Risk Tolerance, (vii) Asset Allocation, (viii) Implementation of an Investment Strategy, (ix) Risk Management, (x) Performance Report, (xi) References .

The heart of the IPS is the objectives for PCV management. While we are advancing with other parts of the document, we would like to hear community feedback about these suggested objectives:

Objectives for PCV Management

The three main objectives of PCV Management are (i) FEI exchange rate stability against the US dollar (“peg maintenance”), (ii) providing liquidity for FEI markets and (iii) having a sustainable and growing FEI DAO.

This informs the approach to reserve management, governed by values of, in order of priority, safety, liquidity and return. The portfolio needs to achieve real growth, after inflation, with a level of risk that is appropriate for the PCV’s return and generate a significant amount of portfolio income.

Additionally, by holding tokens, FEI DAO can take advantage of meta-governance and influence other protocols’ decisions. As a stablecoin, Fei has a central role in DeFi, and meta-governance can be valuable to foster a thriving environment with other DeFi projects.

Another front we are working on is the investment in stable coins. After receiving the feedback from the community, we are analyzing two new alternatives:

1) ~25% of PCV on USDC&DAI is an option optimizing for diversification and increasing the liquidity of FEI with major stable coins

2) ~12,5% of PCV on DAI is an option optimizing for decentralization

We need to highlight that the vision for FEI is to be a decentralized stablecoin. Holding a centralized stable coins exposure is necessary to provide liquidity for FEI, incentivizing its use. For the future, it could be good to provide liquidity against a basket of other decentralized stable coins.

We would like to gather community sentiment on these two options.

Considering the liquidity of these two stable coins, both would support the estimated investment amount. We are suggesting having an equal amount of USDC & DAI to diversify idiosyncratic risks. Although USDC has more liquidity, it also brings more regulatory risk.

Considering the current status of PCV, 25% of it would represent ~ $200M (changing very quickly due to volatility).

1 Like

Still don’t get it: how do you measure idiosyncratic risks, how do you come up with specific figures? Why not 60/40 or 30/70? Could you be more specific?

And again USDC is roughly 5 times bigger than DAI both in 24h trading volumes (7B vs 1.5B) and market caps (19B vs 4.2B). BTW, the gap is widening – at the time of your initial proposal it was 4x. USDC is gaining market share at the expense of DAI.

Also

Why these specific figures?

1 Like

Here we converge.

Maybe, but it should be approached very cautiously at the very least, to the point of nonaction: it could backfire from the reputational PoV as FEI with its resources may well become perceived as a megawhale hijacking smaller protocols within the DeFi community – we all know what reputation whales have regarding their role in deDecentralisation of crypto.

1 Like

Hey Pavel,

Your question is good and we already discussed a little bit about it on discord. I would like to add more info about it.

The analysis below is inspired on the risk methodology of Aave, one of the most relevant lending protocols in DeFi.

As the overall risk of the two options are B+ (for different reasons as you can see by the figure), the idea was to buy the same amount of DAI and USDC.

However, as we are a stable coin, the subjective analysis of counterparty risk need to be more cautious. The regulatory risk could be even greater in our case. Additionally, as some community members pointed, the vision of being a decentralized stable coin is very important to FEI.

That’s why we suggested two options for discussion with community:

1 Like

OK, you clearly have a discrepancy with Aave’s analysis of February '21.

USDC: Counterparty – B (Aave) vs B- (you);
DAI: Counterparty – C+(Aave) vs B+(you);
DAI: Liquidity – B-(Aave) vs B1/2+(you);
DAI: Market risk A-(Aave) vs A+(you).

Could you explain this discrepancy? What changed and/or what Aave gets wrong?

I updated to the date of 05/24 the numbers and separate the category of liquidity risk from the market risk that is more related to volatility.

Mkt cap at that time of DAI was $2.3 bln and now is $4 bln.

Volatility was the same, A+ for DAI. However, as they considered together with mtk cap and volume the total market risk was different.

Here is a more subjective take. It starts with the specific number of holders number but then make some additional analysis. As a decentralized stablecoin, we should value more decentralization and be even more cautious with regulatory risk.

DAI starts with B+ (analyzing holders number) and stay with the B+ due to the decentralization aspect, even holding USDC as part of their collateral. USDC starts with A-, but fall to to B- due to centralization and regulatory risks within US jurisdiction.

There are two main decisions to be made: the % of the stable coin investment in relation to PCV and the composition of this first investment (the two options we have been discussing here are DAI and USDC).

To help in the % decision, we provide below some numbers about the current PCV situation regarding volatility and Value at Risk (VaR) and simulate some scenarios considering an investment range from 10% to 25% of PCV.

Besides analyzing the impact on those risk metrics, we also show how much of User Circulating FEI and Non-TimeLocked FEI would be backed by stable coins after this purchase.

image

We would like to hear from the community about this investment size.

About composition, we provided more analysis here: FIP-6 PCV Initial Stable Coin Investment - #22 by Bruno

P.S.: Just a technical note, one way to see VaR is that 18.25 days in the year the portfolio will lose a minimum of -10.59% in value and in 346.75 days in the year the portfolio will lose no more than -10.59% in value. VaR is one of the most widely known measurements for risk assessment and risk management. We recognize that the VaR method used, assumes normally distributed returns which in Crypto is not necessarily the case.

Hi @Bruno,

It seems that the investment size will depend on risk appetite of the WG and the wider community.
I have some questions:

  1. The PCV diversification is being taken from Fei. As you’ve highlighted ‘circulating FEI’ and ‘locked FEI’, is there any difference on which type of FEI would be chosen? Which pool would you choose?
  2. I am curious about simulated models. Is it possible to simulate impact to PCV using the various investment scenarios with different percentages using the market rout in the past week or so and understand how much PCV would be impacted?

As in regards to the 2 questions you have posed: while DAI fits the ethos of decentralization, it also does mean that we are concentrating diversification on another singular asset. As a PCV I would be very risk-averse and would prefer to diversify on a basket of stables. With USDC I understand the centralization aspect but it depends on whether the we intend to hold USDC for a long time. I am neutral to holding this as long as this is wind down to other types of stables in the future.

If decentralization is a fundamental red line we need to hold, then I suggest (picking random numbers) we diversify say 20% in DAI and 5% in a ‘basket’ of other decentralized stables (eg liquity, sUSD, others)

I think we should look more to non-timelocked FEI, as this is the real circulating FEI in each moment.

This model was updated on 05/24, so it is already considering all the last week volatility. That’s why the daily volatility calculated considering the last 1month is so high comparing to last 3months and 6months.

About the composition, I am just more cautious about the basket of stables in the first moment. I think we need a very deep analysis of each asset we are investing following a rigorous risk mgmt strategy. Many decentralized stable coins are new and not battle-tested yet. I think they would not perform well in our framework for risk analysis. However, some day in the future I think this could be a good option.

Why did you choose VaR as a risk measure?

VaR is one of the most widely known measurements for risk assessment and risk management.

Actually it’s outdated and being phased out due to its inability to capture tail risk.

See, that’s a fine line: if you were even slightly less superpositive about DAI and less supernegative about USDC even in your own assessment USDC would look more preferable.

OK, let’s read Aave then:

DAI Counterparty Risk: C+

DAI is the stablecoin of Maker DAO which has been functioning efficiently since inception. Anyone can mint DAI by opening a CPD. The Maker team also has some control over the minting. Following the Black Thursday crash of March 2020, there was some deficit in the backing of DAI which required the minting of additional MKR. Additionally, there are some concerns related to the centralisation of the price feed, with oracles failing to keep the prices up to date during the said Black Thursday.

DAI isn’t that decentralised. Specifically, Aave gives it C- for trust, which you don’t even mention.

USDC Counterparty Risk: B

As it’s backed by real US dollars, USDC is centralised. The technology to mint new USDC and hold the backed USD is based on a legal framework. It is currently maintained by the CENTRE consortium which is a trusted entity in the ecosystem. Furthermore, USDC is the first regulated cryptocurrency bringing a lot of legitimacy to the space. Still, the infrastructure is based on the Ethereum blockchain where regulators have little power.

See, Aave already takes your concerns into account, balances them with other things you don’t mention and the resulting picture is different.

It is still widely used by many banks and funds that I know. The model has its drawbacks as you pointed and should not be analyzed alone. Volatility and VaR are the two more common metrics used for risk management. Do you want to suggest an additional metric that we could analyze?

I see, this part of their analysis is very subjective. The regulatory risk for a decentralized dollar stablecoin is greater than for a lending protocol.

This analysis is not updated, they do not consider the current number of holders for the centralization score. Also, DAI did well during the crash last week, it is more resilient than last year. Anyway, I agree that we need to keep monitoring DAI risks.

Basel Committee on Banking Supervision ('13): “a number of weaknesses have been identified with using VaR for determining regulatory capital requirements, including its inability to capture tail risk”. Basel III has phased out VaR.

Expected Shortfall

I see, this part of their analysis is very subjective. The regulatory risk for a decentralized dollar stablecoin is greater than for a lending protocol.

Well, your analysis is no less subjective in this regard. Care to elaborate?

Care to give the breakdown of your scores for centralisation/trust for DAI/USDC and rationale?

Tks for pointing that, Pavel! Good to see the recommendation Basel Committee. I think that calculating the Expected Shortfall makes sense. I updated the models here only to change from VaR to Expected Shortfall (considering a 97.5% confidence level following Basel Committee guidelines:

image

Just to clarify to others, Expected Shortfall (also called Conditional Value at Risk) is the average of the “extreme” losses in the tail of the distribution of possible returns. It is how much you could expect to lose in the nine worst days of the year.

For sure, it is difficult to have a more objective analysis about trust and we will depend on how community see this. I think there are two main aspects regarding USDC, decentralization and regulatory risk. FEI is a decentralized stable coin and investing in a centralized stable coin would go against this importante value. Also, as a stablecoin, we can expect more strict regulation in the near future, so regulatory risk from government decisions will be even higher. USDC is under US jurisdiction, that in the last years was not very favorable to crypto. Many projects had difficulties in accessing US investors. I think US govt will embrace existing stablecoins in the future, but it could come with a heavy regulatory burden.

Considered them together. First calculated the score based in the number of holders then adjusted by the subjective analysis of trust.

I fully agree that the ordered priorities are stability, liquidity, and growth. A big part of the FEI vision is to be long ETH and other decentralized assets and to become a compelling alternative as stable decentralized collateral.

FEI’s value proposition is being able to deploy PCV to the benefit of the FEI ecosystem. An allocation to stablecoins can help further our goals on multiple fronts by being deployed with FEI as liquidity. In addition to reducing the volatility of the PCV, it can create efficient arbitrage opportunities and earn trading fees. We should aim not to become overly reliant on other stablecoins. It is difficult to unwind a precedent once set, so sticking to decentralized only collateral types is important until there is a strong reason not to.

With regard to how many stablecoins to buy we can err more on the lower side for now at around 10% of PCV and scale it up if the need and use cases arise. We should prioritize a way to deploy this liquidity to UniswapV3 or BalancerV2 as the goal for this allocation in a future proposal. BalancerV2 has an asset manager feature where we can take the majority of the assets out and redeploy them somewhere like Aave while still providing liquidity. This would be an ideal way to attain capital efficiency and get the yield as a bonus. A UniswapV3 0.5% slippage pool would be an excellent candidate as well. Neither pool would support reweights out of the box, but we could start with just the LP side.

The code @eswak wrote here can be used for more than just swapping into stablecoins, but also into any asset with a pool on UniswapV2 or Sushiswap. This can be used for future diversification or rebalancing.

I support a DAI-only allocation of stablecoins for this proposal with 30k ETH allocated. This keeps in line with the decentralized-only PCV vision. Additionally, future low-slippage stablecoin to stablecoin liquidity will be a benefit for the FEI ecosystem and a way to earn fees for the protocol

6 Likes

OK, I’ll try to calculate it then. Aave:

The counter-party risk is measured from the level of centralisation corresponding to the number of parties that control the protocol as well as the number of holders and the trust in the entity, project or processes.

Counter-party risk = centralisation (# of controlling parties + # of holders) + trust.

DAI

In February Aave gave DAI C+ for counter-party risk. Back then it had 312K holders, so as per Aave methodology, it gets B+ for the # of holders. It got B- for centralisation, hence it got C for the # of controlling parties. Pretty obvious the # of controlling parties hasn’t changed any significantly since February.

You gave DAI B+ for Counter-party risk.

It gets B+ for 385K holders and C for the # of controlling parties. These are objective parameters. So you have B- for centralisation. Therefore, for trust you give A vs C- as per Aave. How did you come up with A vs C- for trust?

USDC

In February Aave gave USDC B for counter-party risk. Back then it had 685K holders, so as per Aave methodology, it gets A- for the # of holders. It got C for centralisation, hence it gets B/B- for the # of controlling parties. Pretty obvious the # of controlling parties hasn’t changed any significantly since February.

You gave USDC B- for Counter-party risk.

It gets A for 1M holders and B/B- for the # controlling parties. These are objective parameters. So you have A-/B+ for centralisation. Therefore, for trust you give C/C- vs A as per Aave. How did you come up with C/C- vs A for trust? Please, explain.

Tks for taking the time to make the calcs, Pavel.

Although the Aave is incredible good reference, we do not need to copy all of it, but adapt to our stable coin reality. Especially this analysis of counter party risk is highly subjective. Even considering the number of DAI and USDC holders, it does not seem to me as a good proxy for centralization as they are not governance tokens.

In the end, I think the value of this framework about counterparty risk is to raise questions to be discussed with community. It would be good to have more perspectives on DAI risk related to some control over minting and trust.

1 Like

Hi everyone,

The snapshot vote for FIP-6 is coming soon ! So I wanted to share more specific technical details about the proposal. I think we need to have a few last discussions & settle on some parameters before we vote. I will give you my opinion, that is the result of my analysis + discussions with other members of the community, and all feedback is welcome to fine-tune the proposal before it goes live :grinning:

The new PCVSwapperUniswap is configured to swap on a Uniswap V2 or a Sushiswap pair. Therefore, we have the choice of which DEX to use. The current liquidity of the DEX pairs are the following :

  • Sushiswap :
    • 107M DAI, 44k ETH
    • 200M USDC, 77k ETH
  • Uniswap v2 :
    • 42M DAI, 17k ETH
    • 137M USDC, 54k ETH

In both cases, for USDC or DAI, the Sushiswap pairs are more liquid. This is an important consideration, because we will be able to do bigger swaps (less gas fees), or swaps with less slippage (more stablecoins for the PCV!).

We could use both DEXes, but for simplicity, I think we should just focus on one (introduce less new contracts in the core protocol, and easier to monitor activity), and I think it should be Sushiswap, because it has the most liquid pairs.

The PCVSwapperUniswap swaps a fixed size of X ETH, and can be called every Y minutes.

  • The larger X, the more slippage, but the less time it takes to convert ETH to stablecoins.
  • The larger Y, the more time for arbitrageurs to do their work, and it lets the market recover from that large volume of ETH we are selling.

I think an interval between swaps Y = 10 minutes is reasonable, as arbitrage should happen pretty quickly.

For the swap size, there are 2 scenarios :

Scenario 1 : we want to swap as fast as possible
In that case, we should target one that incurs ~0.5% of slippage per swap, for instance 100 ETH for the ETH-DAI Sushi pool. To swap 40k ETH, it would take 400 swaps, and if one swap happens every 10 minutes, we’ll be done in 3 days. Total cost for the PCV would be around 477k$ (437K$ slippage + 40k$ FEI incentive).

Scenario 2 : we want to optimize cost for the PCV
In that case, we want to reduce the swap size to have less slippage, but not too much because each swap costs gas, and to cover gas fees of the caller, the swapper offers a 100 FEI incentive, so we don’t want to do too many swaps.

The sweet spot seems to be around 40 ETH per swap. It reduces the slippage to ~0.2%, and takes 1000 swaps / 7 days. Total cost for the PCV would be around 275k$ (175k$ slippage + 100k$ FEI incentive).

One last thing to consider: there is the possibility that the protocol will be targeted by sandwich attacks. That attack consists of manipulating the market to decrease the price of ETH before the swap, for instance with a flash loan, call the swap() function, and swap in the opposite direction afterwards. To protect from this, the swap() function will revert if the obtained amount of stablecoins is less than 99% what is predicted using the Chainlink oracle price of ETH / USD. It means that, if we choose scenario 2 to minimize the slippage to ~0.2%, there is still the risk that the protocol could be sandwiched and all swaps would happen at 1% slippage.

There is also a risk that if the Chainlink oracle is not aligned with spot prices on the Sushiswap pair, swaps could revert unexpectedly. In that case the protocol wouldn’t lose money, but we may have to make a second DAO vote to tolerate more slippage.

For these reasons, I think the scenario 1 is better, even if it seems to cost more to the PCV.

3 Likes