Thank you. What a great explanation!!! We gotta keep burning that vBNT especially with the higher protocol fees being discussed near 100% its the perfect weapon in our arsenal to rapidly close the deficits and make this a rapid recovery! I’m getting pretty pumped!
3 Likes
Some thoughts from my end with either solution
- On ETH/DAI as distribution:
Both of these models required us to source ETH or DAI (DAI to pay LPs and ETH/DAI for the insurance bucket) using outside sources which means that the Bancor DAO will be paying fees to another entity. I would much rather that we do not do that and we keep as many of the fees internally to us. For this reason, I like the “DAI” solution that collects a portion of fees in the native TKN and keeps it off the curve to cover deficits for that pool. I also like the “protection” solution for the specific reason that LPs will not be getting DAI but rather fees in their own native token. If we want LPs to have access to fees being collected in their native TKN (for quick cash), we don’t have to add it to the pool but rather make it available to be collected separately (they can deposit back into Bancor if they want and we can give them the option to do so).
A lot of LPs in Bancor are LPing with these tokens and HODL them because they want to accumulate more of the same token that they are staking. If you are not bullish on the tokens you are holding, then why even HODL or LP with them? I think this is an important value proposition that should not be ignored. Yes, we can always offer LPs the option to convert their DAI distribution to another TKN when they collect their fees but then again, you have already lost value by converting these TKNs to DAI in the first place (payed fees to someone else).
- On the topic of LPs:
On the DAI model, a compelling case can be made for TKN LPs that if we have pools in surplus and there is a bunch of TKN outside of the curve, a portion of this can be distributed to everyone based on some cadence. We obviously want to keep some surplus or reserves “off the curve” for any insurance payments in the future and we can figure out what that percentage should be. This might be a benefit to attract LPs into Bancor. I am not sure if BNT LPs should be entitled to a portion of these tokens since there are other benefits of being a BNT LP.
- Protocol Treasuries
We should take into consideration treasury deposits from other protocols that are TKN LPs. Some preliminary comments received are that selling TKN for DAI is not a net positive. Essentially, it can be seen as putting selling pressure on their TKN and they see benefits from collecting fees in their native token instead (not selling their native TKN for DAI or ETH).
- Withdrawal Fee
There is no longer a cooldown anymore and I would also consider removing the .25% withdrawal fee to make it attractive for TKN LPs. Consider collecting a security fee on the fees generated (e.g. when distributing DAI or TKN to LPs). This essentially means that we are not collecting fees on the principal but rather on the fees that are earned.
3 Likes
I like how both of these methods are alleviating the ILP burden on BNT. A possible solution could be a mix of both:
BNT DAI model (modified to describe my observations)
100% network fee {90% network fee - adjustable}
Distribution (TKN1->TKN2 trade):
TKN1->BNT → 100% vortex {90% vortex; 10% to LPs}
BNT->TKN2 → 100% {90%} of fees in TKN2 moved outside the curve in that pool (ILP reserve)
{10% to LPs}
- Notice that only the exiting side of the trade (TKN2) accrues off curve TKN reserves for ILP
- This method builds up ILP reserve on a per pool basis
- IL/ILP differs for each pool
[ If the ILP reserve on a particular pool is far greater then the IL on that pool, the surplus should be transferred to LPs, or the fee distribution changed; the network fee doesn’t accommodate for this (% of ILP reserve redistributed between pool LPs by governance vote?] - Requires micromanagement of the pools
- IL/ILP differs for each pool
Treasury as a ILP reserve
TKN1->BNT → same as above
BNT->TKN2 → 100% {90%} converted into DAI and moved to protection bucket/treasury {10% to LPs}. x% of the treasury could be staked in curveFi (dependent on total IL and the reserve amount)
- If it’s triggered every trade it could be costly (I’d suggest an off the curve TKN2 reserve threshold)
- The ILP coverage is in a stable asset and it’s earning fees, not an idle reserve in the pool
- DAO & the public can clearly see how much funds the protocol has for ILP
- There is no divergence of ILP reserves per pool
- More simplistic and straight forward
- The treasury could grow with other sources of revenues
I’d refrain from DAI distribution as rewards:
- There was/is always a problem of claiming rewards for small stakers on Ethereum (see the popularity of YFI, migration of SNX etc.)
- I presume that it would make the protocol more costly to use, which the v3 was aiming to lower
3 Likes
Why not change the burn rate to the deficit % ?
I think right now we are ~$34m in the red. So IF we have to make a change to the Burner, why not a dynamic % of BNT buybacks.
34% bnt burn/66% vbnt burn.
AS the deficit decreases the ratio switches back to being in favour of buying vbnt over bnt.
(I am against buying/burning bnt personally)
2 Likes
So If I LP to a TKN pool that’s in deficit, what is my kickback ( other than DAI div’s)?
Is it more risky to deposit then have to wait for the pool to close the gap? Would I not expose myself to the same issues v3 is facing with potential further deficits?
Could we also further incentive LP’s that are adding to riskier/ deeper shortfall pools?
For instance Link being 45% short. Could adding more DAI rewards to LP’s help keep the pool TVL’s high/ keep new LP’s happy? ( this would fluctuate as the gap tightens)
2 Likes
do you understand that when you sell BNT for vBNT and then burn the vBNT you bought that the BNT you sold is trapped unless ppl then dump vBNT?
2 Likes
This is how high and what the trend for the vBNT rate is with vortex to date
This is how much BNT sits under the vBNT liquidity on B3 alone, therefore unable to be bought (same as burning) until/unless vBNT is dumped.
If vBNT/BNT ratio is trending up then more BNT is being trapped in the trading liquidity for vBNT over time.
If vBNT starts to dump relative to BNT then this will release some of this 1.2 million BNT back into circulation.
Look at the charts and think for yourself about where the vBNT/BNT ratio might go under different tokenomic models.
vBNT pumping also pumps BNT, the reverse can be true if it moves BNT from the vBNT trading liquidity to TKN liquidity.
2 Likes
The only people that can sell vbnt are:
-
BNT LPs that received vbnt and decided to abandon their BNT position. This is the most idiotic scenario since you would be getting back a fraction of the BNT that you would if you just withdraw your BNT at a greater than 1:1 ratio with your vbnt in B3, due to fees.
-
BNT LPs try to temporarily short vbnt thinking that they’ll be able to buy it back cheaper at a later date. @thedavidmeister 's graph above shows that with B3 that is a bad bet since vbnt is only getting more expensive relative to BNT over time. Assuming BNT’s price also stabilizes with no more ILP or LMR, then it becomes a really bad bet. The vortex should be more than happy to take that vbnt and burn it, resulting in that trader losing access to their BNT forever.
-
People with surplus vbnt selling. This is the only scenario that can actually create BNT sell pressure again since they will likely take TKN from the trade. BUT, in order for this group of people to have bought extra vbnt, they also bought BNT in the process - TKN → BNT → vbnt. So this group of people provided buy pressure on BNT previously, the same way that the vortex continuously does it. The liquidity of the vbnt pools is also ridiculously small, so they can’t apply much pressure at any single point in time. And once again, while vbnt is under fair market value, the vortex should be more than happy to take that vbnt back and burn it at a cheaper price.
So unless there’s another source for people to retrieve vbnt and sell it for TKN, the BNT that is going into the vbnt pool is being locked forever, the same result as burning BNT. You’re just buying additional and future BNT by burning vbnt directly.
4 Likes
Beautiful I love it. Love the idea of the treasury reserve bucket earning yield of curve or a yearn vault. Just use nexus mutual cover for protection. I also agree that the DAI model should be modified so that we don’t burn 100% of the BNT fee collected because BNT Lp’s would like to earn some fees. However I would be ok with it being at 100% for a few weeks to accelerate the recovery of the deficit of ILP and then later adjusted so that Lp’s earn more fees.
3 Likes
what is the point of the DAI model? Why not just leave 100% of the TKN side split fees to build the pool reserves? i could care less about getting 10% worth of TKN fees in DAI. There should be no arguments other than to clear the protocol deficits. 10% in TKN fees is barely anything. We should focus on the goal of making LPs whole.
4 Likes
It’s easier to bring back TKN through rebalancing than to make up the deficit in fees, aka, make Bnt go up vs TKN. That’s why the 90% share is focusing on buying back and burning Bnt/vBNT.
4 Likes
Kudos to the OP and DAO feedback - all we need to know Bancor is truly ALIVE with plenty of energy to research and safely implement principled short & long term solutions.
My thoughts so far, after reading 91 posts - rather than planning the quickest “fix” (which maybe burning vBNT as fast as possible but lack the required sustainable long term LP profit solutions to keep LPs, and just lead to the LP exit cliffs, as suggested) the DAO and team could seek an earlier open B3 plan for the graduated & sustainable market based solutions to make the protocol more profitable, attractive for LPs & for Arbs who will pay the fees.
If B3 was planning more protocol utility for “off curve” capital, then let’s remain focused on getting those right and bringing them to market. → Token holders, Stake and earn up to 3 ways via staking once in the Bancor liquidity protocol.
As for the other “rabbit holes”:
Dynamic fees (demand surge fee, dynamic along a demand curve relative to an average over / epoch?) is an improvement for LP Bancor’s profit sharing scheme, shared with the ARBs - this is our core business.
Protocol level arbitrage would compete with ARBs? Would early demand surge be partially value extracted by the protocol and keep the surge suppressed, simply leaving less opportunity for the ARBs to capture? How to share opportunity?
The combination of “fair - market will absorb” Dynamic fees and a “fair” protocol ARB and well researched parameters (probably responsive to demand) could seek to keep ARBs sufficiently happy, while the protocol takes it’s fair ARB & fee share - that’s a long term protocol design market solution. A continuation of research done in Bancor v2.1 & B3 in 3 phases.
4 Likes
Thank you for the in depth post @glenn
Regarding protocol treasuries - I actually think DAI is a great fit. Imagine you are the TKN DAO. I assume you have plenty of TKN in your reserves but spending it is essentially putting downward pressure. However, staking it and earning DAI provides liquidity (single sided) for TKN and allows the projects treasury to build up DAI that can be spend without downward pressure on TKN.
Regarding the withdrawal fee - In both options above, I suspect we will find product/market fit with the long tail tokens moreso then the ETH/WBTC. However, in order to get volume from trades, we will need deep liquidity in the ETH pool. The idea above is to use that withdrawal fee to build up protocol owned ETH (that won’t withdraw) to provide that liquidity. This seems much more sustainable than an incentive (LM) for LPs in the ETH pool, but I am certainly open to suggestions that allow protocol owned ETH on curve without spending BNT.
1 Like
Keep in mind the 10% is just a parameter that can be changed by the DAO at any time.
When the number is low, it is intended to reduce the deficit, when it is high, it to benefit LPs.
due to the implementation unknowns, my vote is to go with option 3.
hybrid-ish solution where we can plug and play in a matter of days (instead of months) and free the DAO to focus on the next fee generating solution.
my suggestion is to go with option 3 called: Yield Throttling
the only difference in this model vs the DAI distribution is that all TKN fees that were planned to be swapped for DAI, are kept in TKN.
this allows TKN fees to be socialized and help close the deficit in the pool.
very much in the lines of, one for all and all for one (per pool).
3 Likes
@TheOneJM @thedavidmeister
thank you for all the details and information.
this is truly a topic few people fully understand.
i would take a step back on this as @yudi suggested and focus on the core solution before trying to change/fix additional parts (especially given that even change can create a massive impact which sometimes can be overlooked).
i would still claim the following:
- vBNT burning is too complicated. people outside of you two (and maybe a few others) do not understand how it works, what it does and/or what is the benefits. for this, we need to improve either our communications or the solution to better fit our users.
- vBNT did not meet any real goal. it is hard to indicate if the burning cause the uplift in ratio or the LM ending. we might need to give it more time and evaluate in 3 months. however, my feedback here is that if we do not see vBNT<>BNT ratio moving to where we anticipate it to be, and if we do not see the impact on BNT, we should consider a change.
4 Likes
vBNT burning in v2.1 and v3 has different mechanics and different goals
if i was to summarise my understanding, i’d say:
v2.1 the goal was to mitigate the negative impacts of minting BNT and it was much more indirect, based on the idea that “someday” BNT stakers would buyback vBNT to access their stake
v3 the goal is to amplify fees that we know will never outpace IL without amplification, and is more direct as the protocol first sells TKN to buy BNT then moves the BNT to the vBNT owned liquidity pool, taking it off the market for every other TKN
i agree the v2.1 vBNT model was insufficient, even if it was effective it wasn’t effective enough in light of the minting that was done
v3 vBNT burning may also be insufficient, we don’t have much data on it as b3 hasn’t been live that long and the migration was suboptimal
i’m also not surprised at all that the DAO is trying to manage b3 as though it was v2.1 because many of the new mechanics are very new and very different, even if they superficially look like v2.1 mechanics
what i’d really love to see is a whitepaper on the impact of vBNT burning fee amplification, as a follow up to https://arxiv.org/pdf/2111.09192.pdf
does it actually work? how much is the amplification? is it enough? are there scenarios in which it breaks?
the fact is that my own understanding of the design and impact of vBNT burning in b3 is based only on reading BIP15 and what other people have said (a lot of which is contradictory) and doesn’t come with:
- formal analysis
- backtesting
- simulations
- comparisons with other models
imo somehow we have to take all the complicated variables and distill it down to a single number, like an “amplification factor” that can be shown on a dashboard, and we also need to know how much amplification is needed to offset IL expenses
if we don’t do that then how can regular people possibly be expected to understand and participate?
once we have that information then we can say things like, “oh, X% should go to vBNT burn and Y% can go to a compensation fund, Z% can go to a DAO treasury, etc.”, otherwise we’re literally just saying random numbers that sound nice, with no idea of what the protocol can safely afford (i.e. without being a ponzi)
7 Likes
i fully agree that burning vBNT hasn’t been shown in isolation to be sufficient, all my comments about vBNT burn are re: suggestions to disable/limit it based on not understanding it, or the hope that a short term but unsustainable BNT pump is achievable or useful via. fee based tokenomic changes
all of these rabbit holes can be pursued in parallel in an open and inclusive way by several independently funded teams
that would be AMAZING if done well
my biggest gripe about the b3 development so far is that it all sat behind NDAs and was dumped on a team that is perpetually too small and too overwhelmed, then when it finally becomes public the DAO is collectively scrambling to understand what even happened so we can take responsibility for managing it
4 Likes
As a link holder who already lost 5k Link (half of my life saving), there is something that worry me a lot, and this question does not seem to have been addressed yet.
We all know Eth merge and Link staking are coming in the next months. It will certainly have positive impact on price, increasing deficit in both pools.
Have you thought about that? What are your margins of maneuver?
We already lost 45% of our linkies since ILP has been disabled. We can’t accept to lose more if Link price starts to skyrocket. Those who are staying in the protocol and therefore support the protocol should not lose even more than those who already withdrew with haircut. It would be very unfair and it will hurt Bancor image even more, for sure.
3 Likes
which is why i suggest implementing small changes that would be quick to develop and release and evaluate their impact while designing the next feature in line.
out of all the proposed solutions, we should consider time to market as a key make or break in the decision making.
1 Like