Fei V2 Design Discussion

FIP: XX
Stage: Draft

This post kicks off the discussions around Fei V2 mechanisms! We think it is powerful to involve the Tribe as early as possible in the development. Of course, for the V2 mechanisms to be deployed, they must be approved through governance.

The core principles of V2 are:

  • Enhancing PCV Management Capabilities
  • Tightening the relationship between FEI and TRIBE
  • Backward compatibility with V1
  • Simple risk-minimized design

V2 will be rolled out in multiple phases to minimize risk of introducing too many new mechanisms at once and accommodate development and audit timelines. We see Fei as a continual evolution rather than a sequence of discrete versions. For example, the reserve stabilizer was added and direct incentives were removed already by TRIBE holders. That being said, the V2 branding helps solidify and clarify this suite of particular changes.

Phase 1 - Q4 2021: Tribal Council (Profit Sharing and Backstop)

In Fei V1, TRIBE is implicitly the backstop of the protocol, and not directly tied to the success of FEI or PCV management. V2 aims to directly link TRIBE to PCV performance through TRIBE buybacks and inflation.

V2 would add a mechanism where TRIBE gets minted and sold for FEI as soon as the collateralization ratio dips below 100% or some configurable threshold by governance. The mechanism would be similar to normal reserve stabilizers that sell PCV for FEI when the FEI price dips below a certain threshold, but with an additional check on the collateralization ratio.

TRIBE would earn a percentage of the PCV equity (PCV value - user circulating FEI) each year via buybacks, as compensation for bearing the risk of inflation in the event of undercollateralization. FEI would be minted to buy TRIBE off of the most liquid AMM (currently the FEI-TRIBE Uniswap V2 LP pool) and distribute back via staking rewards. A portion of this TRIBE would go back to the DAO treasury, which could be used to pay for development, liquidity incentives, or as a buffer against inflation.

The rationale behind using PCV equity is that TRIBE should only receive rewards for being overcollateralized, thus in turn incentivizing risk management and a longer time horizon. This feature would require a manipulation resistant collateralization ratio oracle on-chain which @eswak has been working on.

Likewise only distributing a percentage as opposed to 100% of the equity will maintain a healthy PCV buffer to absorb volatility without inflating TRIBE.

For example, if the PCV equity is 500m and the distribution rate is 5%, then 25m FEI would be minted over the course of the year to buy TRIBE and distribute back to stakers.

Open design questions: governance changes with any new staking derivative, potentially boosting for lockups (veTRIBE?)

Phase 2 - Q1 2022: Algorithmic PCV management

One of the most important aspects of Fei Protocol is the management of PCV. This PCV is the reserve used to both back the value of FEI and also provide utility for it through liquidity provision and other integrations.

V2 will continue to benefit from a variety of PCV deployments throughout DeFi. A prominent addition to PCV will be BalancerV2 investment pools. These will be able to:

  • provide liquidity simultaneously to multiple tokens
  • Automatically rebalance to preferred asset allocation (and earn fees in the process)
  • gain capital efficiency by rehypothecating tokens through the Asset Manager
  • potentially earn BAL rewards

Likewise there will be some BalancerV2 Stable pools optimized for FEI stablecoin liquidity, which benefit from the gas optimized Balancer V2 architecture and composability with the investment pool.

The Fei DAO would be completely in charge of the tokens and asset allocations between them, leaving the pools unfinalized. This means the parameters such as trading fee and weights can be changed by the DAO.

By introducing the explicit relationship of TRIBE absorbing PCV volatility, TRIBE becomes effectively levered on PCV depending on how much FEI there is in circulation. To take some extreme examples, if there are no circulating FEI then TRIBE would simply earn rewards from the entire PCV in perpetuity. If there was exactly as much user FEI in circulation as PCV, then TRIBE has no rewards and small movements in PCV have a large impact on PCV equity.

We will add in automatic adjustments to PCV composition based on the effective leverage factor. For example if leverage is high PCV can switch more into stable assets by increasing the weights of stable assets in the Balancer pools.

This treats the PCV like one giant MakerDAO vault, where TRIBE holders control the algorithmic knobs that rebalance PCV based on market conditions.

12 Likes

I really like everything mentioned above but would also like to open the floor and further highlight boosting rewards by locking TRIBE and having a veTRIBE.

5 Likes

I support allowing TRIBE to share profit and function as a backstop. It makes sense to implement these together. I will post a separate comment on the implementation of buybacks.

I do not think having users lock up TRIBE is a good idea. The point of having a secondary market for securities is to make the securities liquid for the investors while the issuer (the protocol) has guaranteed access to capital. We already raised capital (which is the PCV equity) and also have a liquid secondary market for TRIBE - I see no reason to take away liquidity from investors. Even if they are given the option to lock up TRIBE for higher rewards, this would crowd out investors who value liquidity.

I also support exploring Balancer pools. They are quite capital efficient, especially if combined with an asset manager. I would just point out that when we deposit XYZ in a Balancer pool with high fees, we would be trading off liquidity provision for XYZ, presumably against FEI.

Additionally, I suggest that we disallow Guardian from pausing reweights or 0.99 redemptions. This is to prevent TRIBE governance from doing a governance rug pull - pausing reweights and redemptions and then going through a DAO vote to divide the PCV pro rata among TRIBE holders without redeeming FEI at $1.

2 Likes

Good to see the first step of a v2! I liked the ideas! :slightly_smiling_face:

I share here the first thoughts on it.

I would like to see an increase in the capital efficiency of Fei Protocol. I missed some integration with Uniswap V3. This could release a lot of capital to earn more revenues on deployments. Uniswap is the biggest AMM and I think we could have something planned for it. Any idea on how to be more efficient on FEI-ETH LP? Or the best way to implement this (eg.: sth as Gelato or other project could help)?

This is a good idea, I support it. Instead of earning buybacks each year, I would suggest doing it quarterly, after the analyzing the quarter results.

I agree with @countvidal, this is a core issue in my opinion. One problem of DAO governance is that tokenholders tend to overvalue short-term price action instead of long-term strategy. This alignment with a more strategic view will help the DAO to make the best decisions. Lockup TRIBE would have more voting power.

I was wondering how do you see the role of new bonding curves on the v2 design?

I like the idea of using Balancer to simplify and improve PCV management. Also, using Asset Manager would be great to increase capital efficiency.

In terms of v2 marketing, I feel that is strange to have 2 phases of v2. Maybe it`s better to focus and call v2 what we can do in the end of this year and then let the other features to a v3.

1 Like

Hi all,

Excellent points all around. Just a few comments:

I would like to see this mechanism working both ways. For example, we set a target collateralization ratio of say 200% and a re-centering/damper speed of say 5% (this means that sudden price movements of ETH/other assets will not shock the mechanism from mass buying/selling so that it is dampened). If the collateralization ratio goes above this datum then the PCV will start minting FEI in batches to to both get the CR down and also buy up TRIBE.

I agree with @Bruno to use a quarterly buyback based on financial results.

Governance should be able to set the target CR and also perhaps a dedicated OA council to determine where staking rewards should be distributed via TribalChief.

I would still like to see some human review of the algorithms driving this on a quarterly basis.

I had a thought about this and I have to say that veCRV works because CRV is not explicitly used as a ‘control token’ to regulate liquidity in how Fei V2 is being proposed. Having said that most of the TRIBE at this point is still in the Treasury and through FEI-TRIBE so individuals should be able to still provide lockups (and maybe boosts as incentives) and enough liquidity provided by the protocol to transact. I believe that lockups will have a boost to TRIBE prices in the short term but the tokenomics will need to be planned to be robust to survive in a longer term horizon.

I support both of the above points.

I think this is a big step in the right direction for FEI/TRIBE. Glad we’re thinking manipulation resistant CR oracle onchain. I prefer quarterly buyback epochs

Super excited for the V2 :star_struck:

I’d support a veTRIBE model, but if we want to be softer, we could implement something like AAVE’s safety module. TRIBE could be staked in a contract, with a small lockdown period, like 10 days, and earn the TRIBE rewards from protocol buybacks as long as it stays there. But the TRIBE could be slashed by the protocol to be sold first in case CR drops below threshold. In this design, all TRIBE holders benefit from the buybacks (price appreciation), but only people with a skin in the game (risk of priority slashing) get the extra TRIBE airdrops. We could even increase the TRIBE rewards as CR drops closer to 100%, to compensate for the increased risk.

If we go for a veTRIBE model, I’d like to see a way to get out of veTRIBE back to TRIBE, perhaps with a slashing of 10% that would go back to the DAO treasury, or something similar. Curve’s 4-year lock is brutal :sweat_smile: And this ‘get out’ mechanism would be unavailable for 1 month after on-chain voting, 3 month after creating a proposal, stuff like that.

Do Balancer pool have a way to mitigate an asset depeg/flash crash ? At first glance, it seems like managing all PCV assets in separate pools would be safer than pooling everything in the same LP token.

I think if we have an on-chain collateralization oracle, this can even be done more often, in a streamed fashion. For instance, weekly, a keeper calls a function that mints X FEI on a PCVSwapper, corresponding to the share of extra PCV equity times the weekly TRIBE “management fee”, and during the week the PCVSwapper can swap FEI for TRIBE every few hours.

2 Likes

I would like to bring up a migration of FEI-ETH to Uni V3 as part of the FEI 2.0 designs; which will entail for all of the reweight mechanics be migrated to the uni v3 platform. As is uni v2 pool is an incredibly poor utilization of the capital.

3 Likes

For example, if the PCV equity is 500m and the distribution rate is 5%, then 25m FEI would be minted over the course of the year to buy TRIBE and distribute back to stakers.

Tribe still has no profit, because you just made a circle, and tribe still has nothing to do with pcv. I need to change it here. Either use $2500W to destroy the tribe, or actually recycle it, and don’t put it in the lp pool again.
If tribe wants to be associated with pvc, it has to get pvc tangible subsidies or returns, instead of making a cycle and changing the concept. This doesn’t do much for tribe at all.
I feel that the v2 solution is not really helpful to tribe, unless some details are modified. For example, when the mortgage rate is greater than 100%, the part of casting fei is part of pcv, part of tirbe pledge reward and part of tribe are jointly cast. In this way tribe can really benefit.
Otherwise, while pvc is getting stronger and stronger, the mortgage rate will only get higher and higher. Tribe does not get substantial benefits either.

Excellent path forward. I recognise many of the ideas we floated in the discussions on discord, glad they’re finding their place in Fei’s future.

For me the most important issue is this:

Clearly protocol design & decision support tools will emerge as gradually most important. How do we evaluate market conditions? How do we predict market conditions? How should the Tribe council translate these assessments into decisions about the protocol’s overall design and fine details?Exciting times are coming.

Also support this:

Finally, @cozeno is right that Fei should strive to embrace Uni3 .

2 Likes

Proposal: TRIBE Buyback Pilot Program

Authors: @Bruno, @cryptozen, @fei.saver

Summary
Introduce a TRIBE buyback pilot program by establishing a bonding curve that buys TRIBE with FEI.

Abstract
This proposal would be a pilot for the FEI V2 design and introduces a TRIBE buyback program. A percentage of PCV equity (PCV - user circulating FEI) is used to purchase TRIBE from the market. This is a profitable arbitrage for the protocol, aids price discovery of TRIBE, and makes a direct connection between FEI and TRIBE.

Motivation
A buyback signals the community’s confidence in FEI protocol, and a positive price action could raise FEI awareness in the DeFi community and bring in new investors. In the end, the best marketing is price action. More awareness about TRIBE could increase the use of FEI.

A buyback program brings value to TRIBE holders who believed and invested in the protocol. They take the risk to backstop FEI, if needed. It contributes to tightening the relationship between FEI and TRIBE, which is essential for the long-term sustainability of the project.

TRIBE has been grossly undervalued relative to the value of PCV. PCV is worth around $850M, while user circulating FEI is around 280M. Even if we were to wait until all 700M TRIBE is issued, redeem 280M FEI at $1 each, and then divide the PCV pro-rata among TRIBE holders, each TRIBE would be worth more than 80 cents. This is of course only a lower bound on valuation, since it does not consider the fact that 1 TRIBE today is worth more than 1 TRIBE tomorrow due to staking rewards. Nor does it take into account the growth potential of FEI protocol.

On the other hand, TRIBE has consistently traded around 65 cents in the recent bull market.

A buyback program would also contribute to price discovery and help the price of TRIBE to be closer to its intrinsic value.

As this is a pilot program, and considering the collateralization ratio of the protocol, we suggest a buyback of 2% of PCV equity calculated from Q2 (April, May and June).

Specification

Set up a bonding curve with the following parameters.

Asset sold to the bonding curve: TRIBE
Asset issued by the bonding curve: FEI
Scale: 10M
Cap: 10M
Discount: 50bp

References

Joel Monegro, Stop Burning Tokens - Buyback and Make Instead

Yearn forum post discussing the rationale behind their buyback program

I am against a TRIBE buyback program. I believe FEI being a stablecoin and with coming stablecoin regulation down the road, needs to be as diversified as possible. This means that we want to have TRIBE in the hands of as many people as possible. A Buyback program would cause the protocol to control more TRIBE. I would rather propose a revenue share to TRIBE holders. Additionally for better price discovery we can create locking mechanisms that boost revenue share and voting power. This will cause TRIBE circulating supply to go down while maintaining decentralization and increasing price.

This is something we are exploring that has some nuance to the implementation. First it would be a long term commitment as the vote locking can’t easily be undone. This means the initial parameters are extremely important. Second, all of the timelocked TRIBE would not be able to vote if we did a full changeover, so we’d need to establish some kind of joint governance between TRIBE and veTRIBE which could be complex. We’re actively looking into it and it may deserve its own thread for discussion.

The current TribalChief also doesn’t support veTRIBE and isn’t upgradeable, although we can investigate adding upgradeability before deploying.

Also it isn’t clear that incentivizing locked TRIBE is the way to go, as @fei.saver mentioned it would reduce the liquidity, so there are tradeoffs between liquidity (price stickiness) and lockups (incentivized supply reduction/long term alignment)

This is a very important idea long-term, but it isn’t simple to remove the possibility yet. I think further down the line we should look into “un-governance” types of decisions like this.

UniV3 is pretty straightforward for stablecoin liquidity, although I feel balancer would be higher ROI for us in general due to BAL rewards. For non-stable liquidity UniV3 has difficult issues with IL that often make LPing unprofitable for all but sophisticated market makers. We don’t necessarily want to be profit maxis but I think the capital efficiency isn’t as straighforward of a benefit as there are costs associated. We should look more deeply at Gelato for something like this.

Bonding curves and reserve stabilizer will probably remain the primary peg mechanisms, as the protocol would always aim to be fully collateralized and able to defend the peg directly.

BalancerV2 isn’t fully rolled out yet either :slight_smile: I think we should just market all the changes as V2 and release them as they are developed and audited. Its also safer to do fewer changes at once.

Fully agree, I’m keen to explore the design space but not in a rush as it is a big commitment.

I like this idea, although anything over 100% isn’t enforceable because new FEI purchases lower the CR . Its quite possible we could lower the target CR below 100% somewhat to reduce the probability of an inflation event.

These are awesome ideas but add meaningful complexity to the veToken model. Best case would be to use the battle tested implementation from a security perspective, but I think this could have tangible value add so I’m open to exploring it.

This was an earlier idea we had, I’ve leaned away from it because if we ever do hit near 100% CR then the slashing becomes quite likely and the reward profile for staying staked is unattractive. All TRIBE holders should be on the hook and all TRIBE holders should be incentivized to participate in governance and manage PCV.

Yes, their new investment pools will have circuit breakers.

I’m imagining more continuous buybacks as well using this model, we can optimize the frequency/cost of triggering

3 Likes

A bonding curve wouldn’t be the right vehicle for this, I also think it deserves its own thread

1 Like

I am also wondering if using newly minted FEI to buy back TRIBE is the right approach. Why not distribute the minted FEI directly to the TRIBE holders? It would be a bit simpler in that a buyback doesn’t need to be executed. This also rewards TRIBE holders for doing a good job and doesn’t concentrate the supply of TRIBE to @countvidal’s point. Some may choose to buy more TRIBE, but others may not want to.

Curve had the same debate on what to do with profit sharing. They originally wanted to do buybacks of CRV on the open market, but instead decided on stablecoin dividends. The main reasons were fears about being front-run during buybacks and the idea that CRV holders would value having stability in their dividend income w/ stablecoins.

I’m okay with either option, but think it’s worth making a conscious choice. I also favor quarterly over yearly.

Phase 2 looks great. Excited for the developments!

1 Like

That is actually quite a good idea and using PCV profits to mint FEI and pay back pro rata to TRIBE holders could work. Technically we could wrap TRIBE into an xTRIBE where it accrues profits in an agreed time interval. This would work similarly to xSUSHI.

I’d like to reiterate that a DAO should strive to nurture real, effective decentralisation manifested in DAO governance participation vs formal decentralisation manifested in the number of TRIBE-holding wallets. For example, at the moment there are 14,382 TRIBE holders. One could argue that this figure is a sign of relatively high decentralisation. Yet only 17 wallets voted in the last Snapshot. Effectively 99.999% of wallets are governance-wise dormant and hence effectively excluded from governance. Let’s not fool ourselves: if we stimulate further scattering of TRIBE voting power it will effectively spur centralisation in the hands of incumbent whales.

1 Like

@pavel,

That is an observation I had also pointed out. Would you have any ideas on how to ‘democratize’ voting? This is no different to traditional company shareholder rights where if you own a lot of shares, you get more voting power.

Unless you really make it as ‘one wallet one vote’ (not game-proof, but makes it harder for whales)…

Just off the top of my head:

  • reward participation in governance, like if we eventually converge on some form of revenue sharing, whitelist only those wallets who voted;
  • develop semidirect democracy governance tools, like optimistic approval, guardians, multisigs etc.;
  • stimulate delegation, maybe also with some rewards participation: if you don’t want to follow daily intricacies of the protocol’s governance, ok, but at least delegate your votes to somebody active, like a member of some future TRIBE council, holders of multisigs, Guradians etc.

In other words, if centralisation in the hands of the most active ones is inevitable, which it is, let it happen in the controlled and most decentralised fashion possible.

@arcology

3 Likes

:brain: that would be a really cool idea