This proposal is expected to appear on Snapshot for voting on 2022-07-31T15:00:00Z. Make sure to stake your vBNT for voting before this date and time to participate in the DAO decision.
- The LPL-BNT pool is currently the deepest pool, with it beating the UNI based pool by almost x10. This basically makes us the prime provider for LPL
- Current swap fee is on UNI is also 1%, while having more than 10x less TVL than Bancor atm. Thus it should definitely not be strange to increase the current swap fee
- An increased swap fee will help us siphon more profits from arbitrage trades and us being the prime provider, LPL trades will always flow through Bancor
Figure 1 - Current information for the LPL pool on B3:
There are currently two pools besides Bancor on Uni V3 and V2, with 16.71k & 9.7k liquidity, respectively.
As seen in Figure 4 a swap of 5 ETH to LPL currently has a 34.35% impact on UNI V3 and a 4.85% impact on Bancor.
This confirms again that Bancor can provide the best price for LPL traders and thus set an increased fee on our pools without the risk of being undercut by the competition.
I propose the following schedule for the gradual increase in the trading fee:
- 1 to 3%
- 3 to 5%
- 5 to 3%
- 3 to 1%
- Evaluation of all %, and final decision on what to do.
All steps will be performed weekly with the last week gathering all the data and analyzing the results.
Vote FOR to have the Trading Fee be changed according to the proposed schedule
Figure 2 – Weekly volume for the BNT-LPL pool (B3 only)
Figure 3 – Weekly fees for the BNT-LPL pool (B3 only)
Figure 4 – Comparison between 5 ETH swap on UNI V3 (Left) and Bancor (Right)
Vote AGAINST to keep the Trading Fee at 1%
I think we should do this in a step-wise manner since we did a similar experiment with $TRAC (raised them all the way to 4%). We found out that 2% was the optimal fee to get the most profit for LPs.
I suggest something similar:
go to 2% then 4% then 5% then back down to 4% then 2% to see what fee data looks like.
I would say in this case we’re not even competing with rival pools.
The liquidity difference is just so big that even with a 5x
increase in trading fees, we’d still have better deals for Traders.
But I do like your suggestion.
What if we did the following:
1 to 3%
3 to 5%
5 to 3%
3 to 1%
Evaluation of all %, and final decision on what to do.
Is that all right? @glenn
@glenn @AnimaDunk what if we split the liquidity in half and run one pool on 2% and one on 5%?
we can take data from both pools in parallel
Hey, it was the same with the TRAC pool as there was little to no liquidity anywhere else . I do like your suggestions though and I think it is a good start for a fee experiment. We can always revisit 2% if 5% or 3% don’t yield the estimated results.
Not sure about the feasibility of this on our end (having multiple pools at different fee levels) but also splitting the liquidity might make the pool less efficient.
Not sure about the feasibility of this on our end (having multiple pools at different fee levels)
Please ask about it, maybe @yudi knows
splitting the liquidity might make the pool less efficient.
“might” => yes, this is an experiment, it may also create B32B3 arb opportunities that don’t exist currently because there’s no liquidity elsewhere
v3 doesn’t support multiple pools per token.
everything is of course doable but that would mean a pretty large redesign and will also make the (smart contract) interface not as clean.
and it will also most likely reduce the efficiency of the protocol, so I’m generally opposed.
@yudi “less efficient” does that mean “more fees per TKN overall”?
because apparently we’ve been seeing a lot of our recent volume coming from v2.1 to v3 swaps
i’d be willing to chase more data on that
The higher the trading fee is, the less frequent trades bancor will receive.
This is also not a guarantee to collect more fees as volume has the potential to decrease.
Lastly, we have an ongoing conversation on where to “keep” these fees. should they be saved on-curve or off-curve.
- on-curve - fees are also subject to pool risks and are utilized for trading liquidity.
- off-curve - fees are kept as super-fluid liquidity and not utilized for trading.
bottom line, what i am trying to say is that increasing fees might not generate the result you are after (which seems to be more fees that can help close the deficit and benefit the LPs)
optimizing fees and volume is the bread and butter of the DAO whether or not there is a deficit
the default state of an AMM is to bleed TKN over time, so fees simply help capital to stop evaporating
as B3 uses shared pools, bancor does not have the luxury of foisting the evaporation back onto individual LP positions like other AMMs do, the overall system has to be net positive in fees otherwise there’s few reasons to deposit
i’d be interested in whether we get more trades and fees overall with 2x small pools with a low fee than 1x larger pool with the same fee, or even 1x large pool with more fees
I mean that if the liquidity is distributed between different pools, the trade size will be limited.
but tbh, single token per pool is one of the fundamentals of v3, it will require a huge benefit to justify such a change which at this point is hard for me to say the tech implications of, other than the fact that it’s most likely a huge change (token address today == pool address).
Seems like this can be accomplished with dynamic fees though, that would be the (technically) preferable solution.
@yudi right now we have:
- 2 million link generating 550 link per week
- 7000 ETH generating 4 ETH per week
- many project pools where the only meaningful onchain liquidity is bancor itself, with nowhere to do price discovery against, other than near zero fees then a sudden “jump” that brings IL
the main “users” of bancor are aggregators and arbitrage bots, not retail end users, the data i saw on that was that this is by an order of magnitude
aggregators and arbs like liquidity that they can trade between, not one giant pool to buy a token and hodl like a user would
anyway, just food for thought at this stage…
we can and are going to plot all the liquidity sizes for every token against the historical fees it brought in
with that data in hand i will try to see if there may be benefits to a multi pool setup for a single token
@yudi the bnTKN could still represent a single stake but there could be multiple liquidity curves
why is it that we can say that “B3 will bring new revenue streams like native staking and lending” but we can’t say “the liquidity we have for TKN is assigned to two trading curves”?
if we discover that having 2x ETH curves with 3000 ETH in them brings in more than 1x ETH curve with 6000 ETH in it, why not just treat that as two “revenue streams”?
I think we should stick to the custom of posting to Snapshot on Sunday’s.
If we post around the week, we are making things more difficult for voters and proposal might be less likely to hit quorum.
Having said that, it’s just a suggestion.
That exact thought came to mind, because I was thinking it would be two different pools, but because everything is virtual anyways in the vault, there can be multiple logical curves in a pool, perhaps aggregators would have to code for it thus to know impact, or somehow Bancor contract choose between each.
feels to me that this conversation is going all over the place.
increase fees → yes. i believe we should find the right fee point to maximize the revenues from the pool. this type of testing was done on the stable pools in v2.1 and identify that 1% (i believe) is the right place for the fee to be.
where to hold fees → on-curve vs off-curve. this is something to be discussed.
multiple pools per token (multiple curve per token) → my vote for this is no. reason is that we are fighting over price and slippage. shallow pools create a situation where gas costs are a huge share of the cost and prevent frequent arbs to perform trades. less trades seems to generate less fees (especially in times with less volatility).
The discussion here is good but we should stick to the proposal at hand. I think having the fee experiment on $LPL (and other tokens in the future to figure out optimal fees) is worth doing. As to whether Bancor should support multiple pools with different fee tiers (a change in architecture) then we should probably start a new discussion around that (as it has had some input already here).
FYI, Uniswap already has different fee tiers for the same types of pools and I am guessing there should be some literature out there on whether this is actually beneficial for LPs
The reason we can increase fees on this pool is because there basically no where else to trade LPL.
I support raising the fee on this pool.
I agree that doing in in steps makes sense, but I support it either way.