Feedback Request: Potential Direction for Recovery

Which option do you prefer?
  • Dai Remittance
  • Protection Fund
  • Yield Throttling
  • Increase fees going to the Vortex & focus on yield generating features (revisit in 4 weeks)

0 voters

TL;DR

  • This proposal reflects the personal opinions of the author
  • This is not a fully baked plan, rather, this is a direction that I think is appealing and I’m asking for feedback, questions, further contributions, support, counter arguments and comments
  • The short term goal is to create the best possible scenario for all parties
  • The long term goal is to create a robust model that we as a community can further build upon
  • Both models outlined below stick largely with the Bancor V3 system as designed and therefore present no radical changes to the way the pools actually function
  • In both options BNT is not longer minted for ILP
  • In both options, about 95% of protocol fees are going to reduce the deficit (Vortex)
  • (NEW) I would also like feedback on two other questions.
  • (NEW) Both of these models are very similar and include increasing fees and using those fees either to fund the DAI distribution or the protection. Would a more effective approach for the DAO to first focus on those new features and collect the fees for now and later when we discuss which model we choose we already have the funds ready?
  • (NEW) Should we go ahead and increase the V3 vortex to 90% if we know we are headed in that direction anyway?
  • (NEW) Added Option 3 - Yield Throttling - that was suggested in the discussion below:

Summary

  • The BNT supply has grown and been distributed via LM during 2.1 and the events of June 2022
  • The TKN deficits have therefore grown
  • Below are two models (DAI Model & Protection Model) that could help reduce the deficit - both are very similar with a few differences
  • (NEW) I have now added a third option, Yield Throttling

DAI Model

  • Turn the network fee on V3 to 100% - This means all fees generated by trades are used by the protocol and not immediately passed to LPs.
  • All fees in BNT are sent to the vortex
  • All fees in TKN and treated as follows:
    • 10% of that fee is taken and converted to DAI and sent to the distribution bucket
    • 90% of that fee is left as TKN but moved to outside of the curve on that pool
  • DAI bucket gets distributed to LPs on a monthly basis (month is a parameter, can be changed to any period)

Example

  • Trader trades WBTC for ETH
  • WBTC gets converted to BNT - produces a fee in BNT that is 100% sent to vortex
  • BNT converted to ETH - 90% moved outside the ETH curve and 10% converted to DAI and sent to the DAI bucket.

Therefore

  • Assume the protocol generates $35,000 per day in fees (numbers used here are for ease of explanation - the math is simple enough anyone can test with their own assumptions - accurate 24 fees can be seen on the UI)
  • I have seen that on some dune pages the fees displayed ignore the vortex/network fee
  • Total fees would be about $1,000,000 for the month
  • About $500,000 BNT gets sent to vortex
  • About $450,000 of TKN gets moved outside the curve in the respective pool
  • About $50,000 DAI gets added to DAI bucket and distributed to LPs
  • It is very easy to allow LPs to claim any supported TKN which would also further increase trading fees

Key Parameter

  • The split between the DAI bucket and the TKN out of the curve
  • Note that as the deficit decreases, the 90/10 split can be easily changed by the DAO

Protection Model

  • Turn the network fee on V3 to 95% - This means 95% of all fees generated by trades are used by the protocol and not immediately passed to LPs.
  • All fees in BNT and treated as follows:
    *95% sent to vortex
  • 5% of that fee remains with the LP
  • All fees in TKN and treated as follows:
    • 95% of that fee is taken and sent to the protection bucket in the form of DAI and ETH
    • 5% of that fee remains with the LP
  • Protection bucket is available to all LPs seeking to exit. This is not first come first serve, but based on the LPs stake as a percentage of total staked in the protocol.

Example

  • Trader trades WBTC for ETH
  • WBTC gets converted to BNT - produces a fee in BNT 95% sent to vortex and 5% to LPs
  • BNT converted to ETH - 95% is sent to the protection bucket as DAI and ETH and 5% given to ETH LPs

Therefore

  • Assume the protocol generates $35,000 per day in fees (numbers used here are for ease of explanation - the math is simple enough anyone can test with their own assumptions - accurate 24 fees can be seen on the UI)
    • I have seen that on some dune pages the fees displayed ignore the vortex/network fee
  • Total fees would be about $1,000,000 for the month
  • About $475,000 BNT gets sent to vortex
  • About $475,000 of TKN gets moved to protection bucket
  • About $50,000 goes to LPs

Key Parameter

  • Network fee
  • Note that as the deficit decreases, this can be changed to shift balance between fees to lPs and protection bucket

Yield Throttling

This solution diverts the majority of fees to reduce the deficit, while simultaneously sending BNT to the vortex. Note that version 3 of the vortex is not yet live.

  • Turn the network fee on V3 to 95% - This means 95% of all fees generated by trades are used by the protocol and not immediately passed to LPs.

All Fees in BNT and treated as follows:

  • 95% sent to vortex
  • 5% of fees remain with the LP

All Fees in TKN and treated as follows:

  • 95% of collected fees are collected in the token vault & used for building a surplus. These tokens are not made available for trading and are therefore not exposed to any deficits from AMM trading.
  • 5% of fees remain with the LP

Example:

  • Trader trades WBTC for ETH
  • WBTC gets converted to BNT - produces a fee in BNT 95% sent to vortex and 5% to LPs
  • BNT converted to ETH - 95% of fees are collected in TKN and sent to the vault to build a surplus in TKN and 5% is given to ETH LPs

Therefore

  • Assume the protocol generates $35,000 per day in fees (numbers used here are for ease of explanation - the math is simple enough anyone can test with their own assumptions - accurate 24 fees can be seen on the UI)
  • I have seen that on some dune pages the fees displayed ignore the vortex/network fee
  • Total fees would be about $1,000,000 for the month
  • About $475,000 BNT gets sent to vortex
  • About $475,000 of TKN gets moved to the vault to build up a surplus in TKN
  • About $50,000 goes to LPs

Key Parameter

  • On the BNT side, the split between the vortex and BNT LPs
  • On the TKN side, the split between TKN sent to vault to build up a surplus and TKN LPs
  • Note that as the deficit decreases, the 95/5 split can be easily changed by the DAO

In All Above Models

  • LPs can withdraw at any time
    • If the pool is in surplus, their full stake
    • If the pool is in deficit, based on the deficit
  • No more cooldown
  • Withdrawal fee stays at .25%
    • That fee is used to buy ETH with TKN that will become protocol owned ETH on curve
  • It might be time to enable BNT deposits as if people are interested in BNT, it is likely with a view to be an LP

Keep in mind

These models outlined above serve two main purposes.

  • Direct 95% of protocol’s revenue to be used in an effort to reduce the deficit in the near term
  • Establish a framework where the Bancor DAO, with only a few parameter changes, can redirect the protocol revenue back to LPs and away from deficit reduction when the deficit is closed.

This is just the starting point that outlines a framework that can be further built out by the community with additional fee generating features.

Such additional fee generating ideas that have been suggested by the community are:

  • Protocol level arbitrage
  • Use flash loans for JIT stakes on concentrated liquidity protocols
  • Integrations with off chain market makers to allow for support for additional tokens and slippage fee trades. Affiliate fees can be passed to the bucket.
  • Trading competitions
  • Stable swap. Use a dedicated bonding curve for stable to stable trades
  • Dynamic trading fees
  • Token launchpad
8 Likes

I’m a huge fan of the DAI model. I think that it actually serves us better on the protection front than the protection model itself - because it gives users an actual reason to be interested in providing liquidity on Bancor. The protection model simply provides a way for users to get to the exit.

The DAI model creates a compelling reason to come to & stay with Bancor.

I would love to see the math extrapolated on the model - IE how much a user could expect to earn in DAI based on providing $50k liquidity.

I think this clearly becomes a lot more compelling once we have worked our way past the deficit & can distribute more than 10% to LPs. Could we see how much that same $50k would be earning in a scenario where we are distributing 20%, 35%, 50% of fees? I think these numbers start to create something to look forward to.

I’m very excited about this proposal.

3 Likes

DAI Model → seems it is more in line of the following:

  • pros
    • deposit x tkn and always entitle to receive x tkn (there is no value appraciation)
  • cons
    • DAI distribution which will require claim (another transaction i assume)
    • more expensive trade (require a bit more work to get the DAI in every trade i assume)

Protection Model → seems it is more in line of the following:

  • pros
    • very similar to what we have today in terms of codebase
    • deposit x and be entitle to claim x+fees (5%). this indicate some APY and value appreciation.
    • i assume less complicated code which means keep transaction costs as they are
  • cons
    • initially, the accumulated value is small and still based on the pool deficit (meaning, if the pool is in deficit, i will not really have access to the extra value)

personally, i am in favor of the Protection Model because:

  1. it is easy to implement (fast delivery of code, can be implemented close to immediately)
  2. it is easy to understand
  3. it is socializing the tkn IL and can still have the potential to provide protection as intended
  4. gives LP an option to earn passively (no extra transactions needed)
6 Likes

DAI model is interesting because we allow users to claim DAI.
however,

  1. what if the token you deposit moons? would you not prefer to receive tkn instead of dai?
  2. what if this model force every trade to be more expensive? would it still be worth it?
  3. what if this model force LP to deposit their bnTokens into a new contract?
    • how would this affect the potential composability of the bnTokens?
    • would it be worth the extra transaction costs for the LP (stake, claim, unstake)?
  4. what if this model delay the roll-out by 3x time (vs Protection model)? (3x is a random number. it can be 2x or 10x, need to validate with devs)
3 Likes

Responding to your questions:

  1. what if the token you deposit moons? would you not prefer to receive tkn instead of dai?
    Of course, I’d always prefer whatever option results in me having more money - but I would be happy earning a consistent dividend over time.
  2. what if this model force every trade to be more expensive? would it still be worth it?
    That’s something we need to understand - if it forces every trade to cost more gas, I’m less inclined to pursue this model.
  3. what if this model force LP to deposit their bnTokens into a new contract?
    Would this be similar to staking in a standard rewards contract? If so, it’s definitely a much worse UX.
  • how would this affect the potential composability of the bnTokens? It would trash their composability, since they would not be yield-bearning tokens in the non-composable form.
  • would it be worth the extra transaction costs for the LP (stake, claim, unstake)? Possibly, but again much worse UX.
  1. what if this model delay the roll-out by 3x time (vs Protection model)? (3x is a random number. it can be 2x or 10x, need to validate with devs)
    I would be much more inclined to start with an immediate solution & pursue the more appealing option with room to breathe.

Based on this, I would be very interested in proceeding with something like:

  1. Implement Protection Model immediately - since this could be done very quickly & start digging us out

  2. Explore implementation of DAI model. Specifically I would like to know:

    • Is there a way to do this without increasing gas fees & hurting UX?
    • Could we implement this as a monthly airdrop, or take a snapshot of bnTokens that we use for a DAI distribution page?
2 Likes

Thanks for your feedback @lesigh

I also think the DAI model is likely more interesting for DAO treasuries.

If there is support for either of these models, then yes, happy to add some math or build an excel where people can input their assumptions and see the results.

2 Likes

the way i understand things based on the available information, is that there is not a clear number we can plug and play.
meaning, the indication is that 5% will go to LP (in the Protection Model), yet after 3 months of evaluation this can be changed to any number (assume 20%).

2 Likes

Just give me DAI

I can decide if I want to use it to offset IL or not.

2 Likes

If the protection model can be implemented immediately, I would love to start with that while we R&D a DAI model. It starts to fix the situation while giving us something to look forward to.

6 Likes

what i am referring to is iterative roll-out plan.
phase 1: Protection model. can be activated close to immediate from what i know (need dev feedback)
phase 2: improve the LP offering.

  • more fees
  • dai distribution
4 Likes

Instead of being paid in DAI, can the LP choose which TKN they want their distribution in?

Do LPs need to claim every month or will the distributions build for the LP until they claim?

Is there a concern about losing liquidity of TKN over time with the DAI model?

3 Likes

I see a couple mentions of the dune reports incorrectly reporting the impact of the vortex, but no mention of the vortex in these proposals. The vortex and burning vBNT has been extremely successful in locking up Bnt for the protocol. Does these proposals move away from that?

Why not move to a model where 50% of BNT is burned and 50% of vBNT is burned instead to keep some of the economical advantages that the vortex has provided the protocol.

It seems like making the most economical decisions for the protocol should be the focus right now. I’m not sure why moving away from one of the most successful economical designs in Bancor today would make sense at all given the current situation.

4 Likes

Yes, thats a very easy and value driving addition - adding it to the above @TheOneJM

Also, specified that the TKN sent to the protection bucket is now in the form of ETH and DAI @ashachaf

2 Likes

It is not easy on dune to see the total fees (vortex+LP).

I included this here to make sure the the reader doesn’t underestimate the fees by potentially looks at an incorrect or mis-labelled dune.

1 Like

What are the thoughts on burning BNT/vBNT at a 50/50 rate?

3 Likes

Honestly love all of these ideas. Directionally it addresses the primary structural concerns, and sets the scaffolding for new revenue to enter the system to continually renew the whole system. Also important to note it doesn’t require new deposits to “pay off” the existing LP’s, but rather they get to benefit from day 1. The optionality here is nice and given all the noise in most chatrooms etc., wanted to add some praise here.

3 Likes

I would personally prefer to try burning BNT and see how the market responds.

But you may notice I specified only V3, meaning 2.1 would still be burning vBNT

2 Likes

thank you and much appreciated @PaperStreetCapital

1 Like

V2.1 is irrelevant.

The vortex has bought back 3.5 Million BNT while only spending a fraction of that to do so. Considering that each vBNT represents more and more BNT over time, it is an absolute mistake to move away from this mechanism.

I understand people wanting to move to burning BNT directly.

So do both. Don’t move away from one of the most deflationary mechanisms of Bancor when you’re focusing on making BNT deflationary. It makes no sense.

3 Likes

As per my understanding, the protection model fees distribution is based on the total amount staked by LP of the total protocol stake and irrelevant of the fees earned of that specific LP in his/her specific pool (especially and LP might have a big stake in a pool but the fees are minor). The DAI model is pool specific. Hence i believe the protection model is more interesting to LPs

2 Likes