Auction to Buy FEI - Smart Use for ETH Reserves

Based on option 4) “Using the reserves to stabilize the price more directly by selling ETH for FEI with some spread below the peg price. Requires more substantial contract changes but can likely be written, reviewed and voted on within 1-2 weeks. This creates a price floor on FEI and could be low like .95.” Joey shared on the Post Launch Peg Discussion, I think an auction to buy FEI could work.

People can say how much discount on $ 1 they accept, if any. And you pay first the ones with the high need to sell FEI right now. Actually this is how central banks work in exchange markets.

Auction to buy FEI

  1. We define how much ETH we would use for buying
  2. People send the accepted discounts
  3. FEI fills the orders of high discount first, until we reach the amount defined for this auction.

I invite some feedbacks about this and discussion about how this could be technically implemented.


I think this is a very good idea and it prevents the disappointment retail investors would have when attempting to sell on uniswap for ETH or for TRIBE while not knowing the technical implications of their sale;

That would give predictability & confidence that exits are possible at a small cost using the reserve; they would also be less than the burn rate suffered on the established incentivized market so that makes sense to use too

I think that for starters folks could (perhaps privately?) send quotes about the amounts they would want to exit & sign with their quote with their wallet holdings through the UI so that the team has a grasp of the exit size & multiple choice vote then on an the amount to test it with for starters? good ol’ order book :orange_book: either locked or nah idk the ramifications tho :slight_smile:


This is interesting discussion on which is the better way to execute this. It could be done on the First, it could be force auction. But later we could think about begin automated as reweight.

Actually governance token is meet to do that …
Write down your proposal, have discussions around it, make a poll then vote on-chain


but I am not certain the better way to implement it. If it is via website or a Balancer Smart Pool. Just talking to some people to se what can be the better way. And in any case, I think FEI team needs to see it first, because they know the protocol much more.;jsessionid=408C44AFA21E07C57B816940930B93B0?sequence=1

this is a study related to how the European Central Bank (ECB) used different mechanisms with banks to keep the interest rate pegged steady. They use a posted price mechanism with full allotment (PP-FA). Our scenario is the price here yet i think the way to do it is pretty similar. Would love any input :smiley:

“In the posted price mechanism, the ECB sets the policy target rate p0 at which the banks may purchase as much liquidity as they wish. In this mechanism, the price does not reveal the signals of the other banks. After receiving signals, banks condition their reserve requirements only on their own signal and demand”

So, if I understand it correctly, perhaps the expected rate p0 could be our discount factor, it would be the minimum & folks could agree to sell at that price for funds from the reserve while not knowing the price of other participants?

There could be a time period of several hours in which FEI holders can make their quotes & sign & agree on individual rates lower than current burnt rates. I think this would allow retail to sell their FEI without going through higher burn rates & still accommodates the FEI/PCV ratio while addressing the unknown buy pressures & prevents smaller folks from having failed transactions.

To add give some extra functionality, the discount factor would be lowered proportional to the amount of $tribe that participants hold up till some p0-0.7*p0 or smth? this is direct incentives for holding $tribe;

So back to my previous reply yeah I think that amounts quote for sale should be hidden, signed with balances of individuals & discounted proportional to position discount factor for $tribe holders & $tribe-fei LP’ers.

Introduction of paper

"We have presented a model of strategic bidding in multi-unit auctions that incorporates the main feature of the ECB liquidity auctions: The presence of secondary market. The novelty of our model is that the secondary market equilibrium generates the marginal valuation functions upon which bidders base their equilibrium bidding strategies in these multi-unit auctions. We have also used the auction allocations to model the outcomes in the interbank market and the need for the banks to turn to standing facilities of the ECB.

Conclusion of paper

We have compared four different mechanism to sell liquidity: The discriminatory price, the uniform price and the Vickrey auction, and the posted price mechanism with full allotment. All except the Vickrey auction have been used by the ECB in practice at one point or the other. Our main objective was to compare which of these mechanisms is the best at achieving the stated goal of the ECB: The implementation of the target interest rate (reservation price) to the interbank market. We find that the current mechanism of posted price with full allotment is by far the most superior mechanism in this respect. Moreover, mechanism selection involves only limited trade-offs, since the posted price with full allotment is more effcient than even the Vickrey auction in our model. The only trade-off that emerges from our simulations is that the discriminatory price auction is optimal and thus generates more revenue than the posted price mechanism.

Nonetheless, optimality is probably of second order considering the role of ECB and the role of liquidity auctions. However, if the central bank values the information about the market that the bids provide, they should adapt an auction mechanism over a posted price mechanism, because in auctions, the central bank learns the entire demand function, whereas in posted prices the bid is only a single point in the price-quantity plane. Despite these consideration, the auction
mechanisms seem inferior to the posted with full allotment in the relevant policy dimensions.
At the end of the day, we think that the mechanism design in the ECB liquidity auctions should be decided by the main purpose of the ECB: The implementation of monetary policy. Thus, we conclude that the ECB should continue using posted prices with full allotment, even after the current crisis. This conclusion is very intuitive. If the goal of the regulator is to regulate price, it is best achieved by regulating the price directly and letting the quantity adjust, instead of regulating the quantity and hoping that the price will adjust to the desired level."


Auction to buy FEI

  1. People send their quotes, they receive quoted discounts, they agree on their own quotes as a factor of the burn rate, inversely on holding size & tribe/lp holder discounts
  2. Team sees quantity & quote data & proposes to vote on ETH quantity to fulfill sell orders
  3. PCV fills the orders to quoted FEI amount at the rates proportional to their ETH & tribe/lp holder discounts & the ETH sent in a vault from which people can claim it;

Allowing them to also individually agree on a discount factor relative to the burn rate contributes to PCV/FEI ratio, provides confidence in the system’s capability of pegging, you can also soft/hard commit your FEI to a first sale with rates better than the ongoing steady high burn rates that are currently going on

1 Like

Interesting. Could you explain more how 1) works?

And why in 2) Team need to see quantity & quote?

Another option is to use the Balancer Smart Pool: Liquidity Bootstrapping FAQ - Balancer


People go on the site and commit their FEI for sale back to ETH. Their holdings are calculated and based off of that people would get a burn rate. If they own TRIBE/LP tokens, the quantity of burn they suffer is lower. I backtracked my thoughts on the holding size cause I get it would punish folks for just having more capital xD; This then allows some off the counter exit at a loss with money from the PCV

In 2 keeping the quantity hidden could prevent a crowding in effect of too many folks wanting to exit at once because other folks are exiting too, would show data on their propensity to stay which perhaps may be kept sensitive?

No clue on whether the idea of “private” quotes is worth looking into that much,

I guess there might be more downsides with the lack of transparency than benefits in preventing exiting hysteria but yeah i thought that might help keep things calm-ish

So yeee the simplest version of this is just a convert function which slightly punishes less than the current burnt suffered on uniswap or the interface. People seem okay with voting on a percentage they would take for off the shelf stable rate burning fee so thats why I tried to expand on that

I think the balancer smart pool seems pretty effective & I have seen it in action with radicle so far;
Would then the PCV also participate in the LBP to keep it pegged? so that folks can sell at FEI starting starting from peg and the PCV repurchases to keep it like between 0.95 and 1?

Cause if so it seems pretty easy to do :stuck_out_tongue_winking_eye:

1 Like

thats the idea. Selling ETH at Uniswap pool price plus a “penalty”, and then let the market decide the price, respecting a minimum selling price for FEI at peg price. Let say we start to sell ETH at 2,500 and if there is no such demand ends at the peg 2,114.

If you want to chat: Bruno Rodrigues #1710

Just announcing using the PCV to buy back FEI is going to help. It will change the entire narrative around this situation and put people at ease.

Then when the actual buying happens via a Dutch auction, i would not be surprised if the sell side pressure is a lot lighter than expected. Confidence in the peg being restored will be huge. It will entice speculators to buy FEI.

A series of tweets stating that a Dutch auction via a Balancer Smart Pool will really help.

Plus the Balancer Smart Pool option can be implemented immediately and requires no dev work in itself.

If slippage is a concern, the trading fee can be set low. Impermanent loss is less than the burn function built into Fei.

Liquidity can be removed at a later date, so the FEI buyback via PCV will actually increases the book value of its FEI that the treasury has. Add in BAL rewards, trading fee income and that the transaction just swaps ETH for FEI crating a higher FEI price - this might actually increase the protocols NAV.

Also a smart pool, the swap function can be turned on and off. Any issues, it just stopped, so the risk is minimised. The sentiment change with announcing this will be huge! It means speculators will want to buy FEI for a quick profit!


From discussions with @Matthew_Graham:

Create a Balancer Smart Pool with FEI and ETH. FEI starting price can be the Uniswap last price (let say the equivalent of $ 0.80).

Arb bots would buy FEI cheap via uniswap pool and sell FEI for ETH in Balancer. Using Balancer, PCV will be buying FEI with ETH at a price below $1. Starting with the price of Uniswap let say 0.8. The price of FEI on the Balancer will be going up as time passes until it reaches 1.00. The dutch auction. The pool weights change, increasing the incentive to deposit FEI into the pool and purchase ETH.

This FEI would be purchased from the UniSwap. As the dutch auction continues, the net effect is to lift the ecosystem price of FEI towards the peg. FEI Seller could sell on Balancer or to arb bots on Uniswap.

The Smart Pool starts off mostly ETH with some FEI and finishes majority FEI. With the auction, PCV uses its ETH to buy FEI at a reasonable price (below $1) contributing for the increase of its reserves in the long term. In other words, the auction is essentially selling ETH and receiving the excess FEI.

At any point this does not go to plan, Fei Protocol can pause the swap functionality. Fei Protocol receives fees, BAL rewards and creates the much needed buy side pressure on UniSwap.

We can test first with a small amount. Actually, it would be good to do many auctions and see how the market agents play with it.


  1. Start with ETH Uniswap pool price
  2. Define a specific amount of ETH for the auction
  3. Minimum price for the auction is the Peg price $1

If this works we could let a permanent Dutch auction there starting at the current FEI mkt price and with the maximum price of $1.

Reference Material:


Using a Balancer smart pool sounds really cool. It’s like a continuous version of reweighting, where instead of going straight to $1, we have the price moving up gradually over time - a Dutch auction, as you said. If this works well, we could just move the whole PCV to Balancer. Then the main parameter would be how fast the price increases as a function of selling pressure.

There would still be arbitrage opportunities as the contract is “poked” to increase the price. But the gas auction could be less severe, since instead of competing to be the first one to poke at time t, you can just sell at time t-1 at a marginally lower price. So the gas auction is transformed into a price auction, which is much better for the PCV.

Perhaps there exists an ideal static auction that sells a certain amount of ETH and completely removes the gas auction. But the Balancer auction could come pretty close. We should definitely think more about this idea.


Tks for investing time on this idea, it was a little bit forgotten here! hahaha

I didnt understand what do you mean by gas auction. In my view, what is happening is that people will be selling FEI at a price that increase with time until reach $1. I am seeing this as a price auction.

Other way to look this is that PCV is selling ETH, starting at 3,074 FEI and going down until it reaches the peg 2,483 FEI. You can play with the simulation of this auction here:

Some questions:

  • How fast the price will increase?
  • What amount of PCV will be available in this auction?
  • How long will take each auction? The standard in balancer seems to be 3 days
  • When an auction ends the other can start at the same time? So we would always have an auction available? It seems to be an interesting option
  • How to guarantee we do not sell ETH under a minimum price?

Any more questions I am missing?

Reference: about the Liquidity Bootstrapping Pools that we would be using here → Building Liquidity into Token Distribution | by Mike McDonald | Balancer Protocol | Medium

I just meant that there could technically be a gas auction if multiple bidders want to sell at some price x. But rather than paying gas, one of the bidders could just sell at x-1, so it seems the gas cost could be small. Ideally, competition would mostly take place in the price dimension. This seems plausible as long as the price increments are small enough.
The simplest way to implement this might be to have the auction run all the time, either by moving the whole PCV to a Balancer pool, or staying in Uniswap but allowing reweights to happen in a continuous way.

1 Like

I see the benefit of the Balancer Pool in comparison with continuous reweights in the uniswap is that you give time to arbs and mkt to buy FEI and help the protocol to stay in the peg. It is more efficient in terms of PCV use than providing a continuous redemption option at $1.

The result of the simulation is that we start selling ETH reserve at the current market price (3,110 FEI) and step by step we sell it for a price near to the peg (2,435).

Source: FEI Dutch Auction Simulator.xlsx - Google Drive

It is just possible in Balancer, because on this platform the weights on the pool can change. In Uniswap it is fixed on 50/50, the quantity of tokens change, but the amount of value of each token in the pool stays the same.

The weights in this simulation would be the following:

Also, the smart pool in Balancer enables this price decrease step by step. We would be using this specific kind of pool called: Liquidity Bootstrapping Pool. Instead of using this for a Token launch or a token sale, we would be using it for ETH reserve selling.

When the price reach the peg for a specific amount of time we could pause the auction and let the market work. If the price is out of the peg (<$0.99) for 12 hours, it would automatically start the auction.



Good idea, I agree with you, the Founding team do as this ASAP

1 Like