Perpetual Protocol Private Markets - Part I [RFC]

Hey everyone, Jonathan from Delphi Labs here!

I’m very excited to present Part I of our Private Markets proposal to the Perpetual community and look forward to your feedback regarding this proposal.

If you prefer to watch a video walkthrough of the proposal you can do so on our portal.

Let’s dive in!

Introduction

The objective of this engagement was to design a framework for Private Markets (PrM) on Perpetual Protocol. For the context of this piece, we refer to Private Markets as perpetuals markets that can be deployed on the Perpetual Protocol in a permissionless way. These markets would coexist and complement the current markets deployed on Perpetual Protocol (which we’ll refer to as Public Markets).

The process of designing the Private Markets framework, however, required us to first take a step back and review the high level architecture of the whole system. Thus, this post will be divided into two parts:

  1. In the first one, we’ll go through our proposal for the high level architecture of the system and how the Public Markets would work under this new architecture. We see this first part as a prerequisite for understanding the rationale behind the Private Markets framework.
  2. In the second part, which will be published at a later date, we’ll explore the Private Markets architecture.

High Level Architecture

The Public Markets architecture would be comprised of 4 main elements:

  • PERP DAO: The main DAO of the protocol, responsible for overall protocol governance and Public Markets creation. Users who stake into the DAO are entitled to protocol fees and governance participation and, in exchange, they act as the second tranche in case of protocol deficits (we will explore this dynamic further in the next sections). DAO participation is tokenized in the form of sPERP.
  • Treasury: A DAO-managed fund dedicated to the long-term sustainability of the project.
  • Global Insurance Fund: The first backstop of all the Public Markets. Upon any shortfall event, the funds in the Insurance Fund are the first to be used to make users whole.
  • Public Markets: Markets approved and launched by the PERP DAO that are backed by the whole Public Markets architecture.

The following figure summarizes the above architecture.

Figure 1. PERP High Level Architecture.

Governance

PERP holders who stake into the PERP DAO receive sPERP back, which represents their share of the DAO and entitles them to governance rights and fees generated by the protocol, while at the same time serving as second tranche in case of a protocol deficit. Being the second tranche in case of a deficit means that sPERP holders will only be slashed whenever a deficit wipes out the Global Insurance Fund. In that case, we propose that sPERP holders be slashed up to a maximum of 30% of their holdings. If there’s a deficit event that wipes out the IF and 30% of the assets in the PERP DAO, the DAO can decide either to mint PERP, issue IOUs to be paid in the future or use Treasury funds to cover the deficit.

As part of the new architecture, governance will need to decide on a number of important parameters for the system to work as intended. The following is the list of parameters and some of our recommendations (see Figure 2):

  • Cooldown Period: There should be a cooldown period that stakers in the DAO need to wait before being able to withdraw funds from the DAO. This cooldown period is necessary to avoid the case where stakers withdraw their funds from the DAO immediately after a deficit takes place, potentially leaving the system at risk. For the system to be robust, funds staked into the DAO need to be a credible source of insurance in case of deficit events. We propose a cooldown period of 1 week.
  • Insurance Fund Target: Ideally, the IF should be able to cover all deficit events by itself, without the system needing to use DAO funds. For this to be the case, the IF needs to have an appropriate size based on the risk the protocol is taking. As such, we propose that the majority of the fees flow to the IF until this level is met. From that point onwards, fees would flow to DAO stakers. We’ll explore this in more detail below.
  • Treasury Target: For the protocol to be self-sustainable in the long run, the DAO should control a Treasury to be able to allocate resources to operational expenses, a Grants Program, etc. In a similar way to the IF Target, we believe that a portion of the fees generated by the protocol should be allocated to the Treasury until it gets to a reasonable size and, from that point onwards, fees would start flowing to DAO stakers.
  • Fees Distribution: The fee distribution between the IF, the Treasury and the DAO. As was explained before, this distribution would depend on the state of the IF and the Treasury. Below we propose a fee distribution model in more detail.
  • Treasury and IF Asset Composition: What assets should the Treasury and IF hold.
  • Treasury and IF Asset Strategy: How those assets should be used.

Insurance Fund Target

The IF Target is one of the most important parameters of the system, as it determines to a certain extent how secure and robust the platform is to unexpected shocks. We propose setting the IF Target dynamically based on the risk the protocol is taking. In the case of a perpetuals platform, that risk is determined by the open interest in the perpetuals offered by the protocol. The higher the open interest, the higher the risk the protocol is undertaking and thus, the larger the IF should be. As such, we propose the following methodology for determining an appropriate IF size, given the protocol risk:

  1. First, we define the Insurance Ratio, which is the ratio between the IF and the open interest. In a sense, this ratio determines the current riskiness of the protocol. The lower the ratio, the higher the risk for protocol users, all else being equal.

  1. Setting an appropriate Target Insurance Ratio for the protocol. This should be done by the PERP DAO.
  2. Adjusting the flow of fees (to the IF or the DAO) based on where the Insurance Ratio is. If the Insurance Ratio is below the Target, fees should flow to the IF. Otherwise, fees should flow to stakers as that would indicate that the IF is large enough given the current open interest.

The following figure summarizes this methodology and offers some additional color on it:

Figure 2. Target Insurance Ratio Methodology.

Having this methodology in place, we offer an initial consideration for setting the Target Insurance Ratio based on a benchmarking exercise that includes some of the leading derivatives exchanges (specifically focusing on their BTC and ETH markets). What we observe from this exercise is that insurance fund ratios vary considerably between the analysed exchanges, from ~13% to ~37% (see Figure 3). It’s worth noting that these exchanges have been operating for longer than Perpetual and have been able to amass significant insurance funds over time.

We expect this exercise serves as a starting point for the community to set the Target Ratio. Given that Perpetual v2 will be built on a completely new architecture, we suggest establishing a conservative Target Ratio initially. As the model matures and enough empirical evidence is gathered, the Target Ratio could be adjusted downwards when needed.

Figure 3. Insurance Ratio Benchmark. Data as of August 23, 2021.

Fees Flow

The last piece we’ll discuss about the Public Markets functionality is the fees flow. Given that on Perpetual v2 liquidity providers (LPs) will do the heavy lifting within each market, we suggest the majority of the fees flow to them. Specifically, we suggest that ⅚ of fees generated flow to LPs and the remaining ⅙ flow to the protocol.

With regards to protocol fees, they should be divided between the IF, the Treasury and PERP DAO stakers. As was described previously, whenever the Insurance Fund is below the target, the majority of the fees should flow to the IF. In a similar sense, whenever the Treasury is below target a portion of the fees should flow there. On the other hand, whenever the IF or the Treasury hit their target, fees should be redirected to the DAO. We propose the distribution of protocol fees to be as follows:

  • 80% to the IF until it hits its target, deposited in USDC. After it hits its target, fees are redirected to PERP DAO stakers.
  • 10% to the Treasury until it hits its target, deposited in USDC. After it hits its target, fees are redirected to PERP DAO stakers. Given that the Treasury is already well funded for the foreseeable future, no fees will be initially flowing to the Treasury.
  • At least 10% to the PERP DAO, deposited in PERP. Ultimately, the share of fees that effectively flow directly to PERP DAO stakers will depend on the state of the IF and the Treasury. Initially, 20% of fees will be flowing to PERP DAO stakers, as the share of fees directed to the Treasury will be set at 0%.

Notice that the above distribution will be adjustable by governance at any point if considered necessary.

The next figure summarizes the mechanism presented above.

Figure 4. Public Markets Fees Flow.

Conclusion

Throughout this post we outlined what we believe to be a robust architecture that will allow Perpetual Protocol to safely grow going forward. Additionally, this architecture will serve as the foundation for the Private Markets framework that will be explored in the upcoming Part II of this post.

We look forward to your feedback regarding this post.

2 Likes

What will the trading fees expected to be? 0.3% (0.25%, 0.05% split)?

Would be interesting for this to be adjustable by governance. E.g. to:

  • Undercut fees against the competition.
  • To favor LPs even more until volume picks up.
  • Make fees more competitive if there’s sufficient volume.
  • Start off at a higher rate to encourage liquidity.

Overall, I like this proposal. Prioritizing treasury DAO at the beginning makes sense.

3 Likes

Is there only one DAO called PERP DAO, and specifically what mechanism controls the Treasury funds? It says “DAO-managed fund”, is that a Treasury DAO or PERP DAO?

Is the 1/6 + 5/6 fee flow suggestion based on technical risk or some qualitative assessment of the business case for LPs relative to PERP DAO? How do you arrive at this ratio?

Why are PERP DAO dividends paid in PERP when the rest in USDC? Where is that conversion taking place? As in, where are the excess PERP located, the Treasury?

4 Likes

Yours is a great point.

For example, market makers on stock exchanges receive a maker fee, however futures exchanges offer volume discounts because there’s sufficient liquidity. At some ratio of OI/liquidity, PERP DAO is overpaying LPs when there’s more liquidity than taker demand.

Ideally, the LP/DAO splits would be dynamic based on OI/volumes/taker demand so that the system is optimally incentivising makers relative to its DAO, and not overpaying for liquidity.

Direct video link to Jonathan’s tl;dr - Perpetual Protocol Private Markets Part I from EA on Vimeo

Thanks for the questions @mykha2 & @almccann !

What will the trading fees expected to be? 0.3% (0.25%, 0.05% split)?

@mykha2 I would expect fees to remain at 0.1% max, since this is already on the high end for perpetuals. Governance will be able to propose and vote on fee adjustments, for sure.

Is there only one DAO called PERP DAO, and specifically what mechanism controls the Treasury funds? It says “DAO-managed fund”, is that a Treasury DAO or PERP DAO?

@almccann afaik there is only the Perpetual DAO, controlled by PERP stakers. The DAO has funds in various forms - this makes up the treasury.

Ideally, the LP/DAO splits would be dynamic based on OI/volumes/taker demand so that the system is optimally incentivising makers relative to its DAO, and not overpaying for liquidity.

Since makers (aka LPs) receive fees based on their proportion of the pool they’re in, I don’t see how liquidity could be overcompensated (ie. LP APR dynamics should behave like any other AMM liquidity pool). In fact compared to spot, they are underpaid, which in theory is made up for by higher volumes in derivatives markets.

Fee flow and PERP dividends will have to be answered by someone else :sweat_smile:

1 Like

wouldn’t having a higher % of fee flow to the protocol incentivise more people staking, thus improving the protocol in the long run?

higher fee’s going into protocol → more people staking perp + increased insurance funds → improved reliability + confidence in protocol

maybe change it to 2/3 to LP and 1/3 to protocol?

The thinking at the moment is the most critical asset an exchange has is liquidity, so this is the group that needs to be rewarded the most. If you have liquidity, traders will come, and fees will flow, ensuring the other parts of the system (stakers, insurance fund) are sufficiently funded.

maybe change it to 2/3 to LP and 1/3 to protocol?

1 Like

So just adding to Lee’s point there’s some fundamental changes in V1 → V2.

In V1 because of the long-short skew issue the insurance fund had to actually play a part in paying or receiving funding. That meant that the risk to stakers is significantly higher, especially if there would be periods where the draw down of the insurance fund is prolonged and then depleted the insurance fund.

However in V2 each trade has a counterparty and as such funding simply flows between both parties. Given this, the actual risk for stakers actually has significantly reduced, hence the flow of fees. We are also investigating potential solutions to further mitigate risk of insurance fund draw downs through optimisation of the liquidation mechanism

1 Like

This is a very good proposal, and I like all the parameters that can be decided on by the DAO.

Most fees go to LP’s in the beginning which makes sense. But has it been considered to have a liquidity mining program in the beginning with PERP tokens? dYdX will be aggressive on this front. Even 100% of trading fees to LP’s might not be enough in the first months until volume picks up.

With a liquidity mining program to LP’s we could send 100% of fees to the insurance fund. A well capitalized insurance fund will make traders trust the platform.

1 Like

Hi guys! Vishal Kankani from Multicoin Capital.

Thanks @JonathanErlich for crafting the proposal. I like it directionally. A few questions -

  1. How do you think about Target insurance fund level when accounting for non BTC & ETH assets?
  2. Should there a minimum targeted dollar amount for IFas this changes take place?
  3. Why is it suggested that IF and Treasury funds be all in USDC? Why not a different mix?
  4. What is the thought process behind choosing the number 30% (% of stakers’ position as the second back-stop after IF runs out?).
  5. For point 4, when IF runs out, its highly likely that PERP token price nosedives. That leads to stakers’ purchasing power crashing. i.e. the effectiveness of the backstop is weaker – any ideas on how to address this?

Happy to brainstorm and contribute in designing the proposal further. Twitter DMs are open.

Disclosure: Multicoin Capital owns PERP.

2 Likes

Why is it suggested that IF and Treasury funds be all in USDC? Why not a different mix?

USDC is the only collateral type for all markets, so it makes sense that IF only contains USDC to cover potential loss. Treasury could be in other tokens (e.g. ETH or WBTC), but then the community has to make decisions regarding treasury management.

I have been a longtime proponent of making Perpetual Protocol truly multi-collateral, where we have markets priced in ETH or DAI as well as USDC.

I’m looking forward to Part 2 that describes private markets and permissionless pair listing. I hope permissionless collateral listing is part of the proposal, but I don’t expect it for V2. Maybe V2.1.

2 Likes

Hi everyone,

it´s been a while but I like how the project is moving forward and will follow it more closely again.

I find the proposal well laid out and the questions being brought up are spot on. Let me throw in my 2 cents.

Coming from traditional finance, risk and volatility is often synonymous. I suggest the target IF ratio should be related to the underlying market volatility. Let´s say BTC/USDC is the standard market with a target ratio of 40%. Lets assume SNX/USDC has 1.5 times the volatility of BTC, the ratio should be 1.5*0.4 = 60%
This ratio could go up to 100% - or a lower hard cap determined by governance.

With all due respect to Delphi Labs, I think these ratios are often just copied from other projects where it works out. the 1/6 + 5/6 fee distribution, to my knowledge, was first introduced by SUSHI and is seemingly successfull, while the 30% slashing cap reminds me as strikingly similar to AAVE´s staking model. It kind of works but I have to note, we cannot know the optimal numbers as long as incentives outweigh tweaking effects by orders of magnitude. For example, Sushiswap still rewards LPs generously with liquidity mining so they don´t need to be concerned with earning 0.25% instead of 0.3% on volume.

In my opinion the fee split LP/protocol should not be hardcoded but must be controlled by governance to be future proof. Or find an algorithmic way to balance them and reduce governance (preferred but hard).

I suggest we borrow from AAVE again with a few adjustments and incentivize liquidity for PERP as well so that a backdrop event doesn´t completely smash the price.

My suggestion is a 2 tier model:

  1. staking PERP → gives you full governance rights and serves as slashing backstop → receives 1/3 of protocol rewards
  2. staking 80%PERP/20%ETH Balancer LP Tokens → no governance rights, → receives 2/3 of protocol rewards

ETH as counterasset in the pool seems reasonable because Ethereum is the base risk for any project built on top of it. Building on Ethereum means, naturally you have to be bullish on the future of ETH. I would not like to see Perpetual becoming even more dependent on the well-being of one particular stable coin. In fact, we should seek to minimize that risk - but different story.

The 1/3 + 2/3 split is based on the assumption that reality falls within the realm of the following extremes:

  • The market values governance rights at 0$ → There will be double as much liquidity as staked PERP. (Yield equilibrium) This makes the price of PERP very resistant to shocks.
  • The market values governance rights as high as the return of the governance dividend → There will be as much liquidity as staked PERP. This still provides very decent liquidity, given that Perpetual flourishes.

Another thought regarding the IF funding:
Why don´t we kickstart it with a portion of the Ecosystem Rewards fund? As @tongnk wrote in the Volume Program thread, there are 21M unlocked tokens which equals to over 400m USD at current prices. Taking 50m USD or so to kickstart the IF could go a long way and leave enough room for other operations.

2 Likes

Hey guys, thanks for all the feedback!

Below some of our comments:

We see the Treasury as a fund that will be controlled by the PERP DAO. Initially I can see the Treasury being managed through a regular governance process where each spending proposal is submitted to a vote, but eventually I think the best approach would be to designate a Treasury Committee with certain pre-determined attributions (such as a budget, spending use cases, etc) that can make decisions in a more efficient way.

We see sPERP working similarly to xSUSHI, where it continuously accrues value as fees are deposited into the PERP DAO; as such, over time each sPERP should be worth more PERP as fees flow to the DAO.

The rationale for this was: 1) To put continuous buy pressure on PERP (as that PERP would be bought off the market) and 2) We assumed sPERP holders are bullish PERP and would prefer to receive those fees in PERP rather than USDC or other asset.

Yeah, this is a good point and I think the team is already looking into it for v2.

I think something like @DeFiChad’s proposed methodology makes sense here:

I think this will ultimately depend on how open interest evolves in v2. However, I’m assuming that the funds (or at least some of them) in the current IF will be used to initially bootstrap the IF for v2. CCing @tongnk here…

Regarding the IF: What @bardur suggested is spot on. The rationale here for having the IF in USDC is that the protocol’s potential liabilities in case of a deficit are in USDC as the markets are settled and margined in USDC. Having said that, I do think there’s potential for putting the IF assets to work on some low risk, yield generating strategies, and that should be decided by the DAO as part of its IF asset strategy.

Regarding the Treasury: Just to clarify, we’re not suggesting that the Treasury should hold only USDC, just that the fees generated by the protocol flow to the Treasury in USDC (at least initially). Having said that, we definitely think the Treasury should hold a mix of different assets (beyond PERP and USDC). Unfortunately defining a Treasury asset strategy was beyond the scope of this proposal but we do think there’s a lot of interesting possibilities to explore here. Eventually, whenever there’s a defined asset management strategy for the Treasury the asset (or assets) in which protocol fees should be deposited into the Treasury could be adjusted.

We wanted to strike a balance here between:

  1. The DAO serving as a significant backstop in case of a shortfall event.
  2. Incentivizing as many PERP holders to stake as we could. This would potentially make the community stronger as more PERP holders become actively involved in protocol decisions as they would have skin in the game.

In other words, we were looking to maximize the size of the DAO (adjusted for slashing) and the number of stakers. We also assumed that the higher the maximum slashing, the fewer stakers we would have, as the more risk averse PERP holders would be gradually left out.

Now, as @DeFiChad suggested, the inspiration for this system and the final figure (30%) were borrowed from Aave, where we think the model has worked well in achieving the above objectives.

However, we’re definitely open to exploring other suggestions here. Additionally, as with every other parameter in the system, this should be adjustable by governance going forward.

Indeed, this is an issue with all native token denominated insurance funds. Agree with @DeFiChad that implementing something like what Aave has done with the 80/20 AAVE/ETH pool could help mitigate this situation. Open to other suggestions from the community here as well.

I think there are two parts to this one:

  1. Setting the trading fee: While initially I think it makes sense to have a fixed fee for all markets (as suggested by @LeeKB) for simplicity and to aggregate liquidity at the initial stages of the v2 deployment, eventually I think the best approaches for defining the trading fee are: 1) A dynamic approach as suggested by @DeFiChad that takes into account the market’s state (volatility, liquidity, etc) and adjusts fees as needed or 2) A market-based approach where you allow LPs to provide liquidity at different fee levels (a la Uniswap v3).

  2. Setting the fee split between LPs and the protocol: This step depends on the first one directly. As in step 1, I can see different splits applying to different markets. For instance, less risky markets could have a higher protocol fee split versus riskier markets where LPs are taking a larger risk. Having said that, we suggested the 5/6 and 1/6 borrowing from Sushi, where it’s worked well (with the caveat of LM incentives - as @DeFiChad correctly points out). We think this split is a good starting point for v2 to start operating but definitely don’t see it as the final solution (something we should have been more clear about). Eventually, as the system matures and more empirical data can be gathered a more complex fee split methodology could be implemented.

Additionally, we definitely think these parameters shouldn’t be hard coded but rather adjustable by governance going forward.

1 Like

@JonathanErlich thanks so much for the thorough reply! Regarding the above, PERP stakers are bullish, but USDC staking rewards have been a big draw since the early days of the project and a lot of community members were disappointed when it became clear that v1 could not deliver on this promise. Additionally, the expectation was vested PERP rewards and liquid USDC rewards, giving stakers a liquid income. This is the sentiment I’ve gathered from the community over the past months.

I’m assuming that the funds (or at least some of them) in the current IF will be used to initially bootstrap the IF for v2.

Highly unlikely because v1 will continue running well after initial v2 launch. My assumption is the v2 insurance fund will be seeded at launch though, likely from the team treasury or Perpetual DAO, depending on amount of funds needed. To my knowledge this decision hasn’t been made yet but I will defer to Nick @tongnk on this!

2 Likes

Thanks for the color here! I think this is definitely workable and ultimately I think the decision of what assets PERP DAO stakers receive their share of the fees in should be made by the community.