Without testing, nobody knows the optimal fee for the ETH/BNT pool.
The fee has been at 0.1% for a long period of time and data on performance already exists for that time period.
This proposal suggests a fee increase to 0.15% in order to evaluate any changes in volume, fees generated, vBNT burned and fees/TVL.
Current Fee
The 0.1% fee associated with the ETH/BNT pool has been static for a long period of time. Certainly long enough for a wealth of data on:
Volume
Fees
vBNT
Fees/TVL
Now that we have a baseline measurement, this proposal suggests that we increase the fee by a very modest 0.05% in order to monitor the consequences on the measures above.
If this performance of the protocol improves, a future proposal could suggest a further increase. I performance sufferers, a future proposal could suggest a reduction in the fee.
A Philosophical Argument
The Bancor Protocol has many parameters that can be adjusted in order to fine-tune its economics. Another such example would be the Vortex Fee.
This proposal argues that in the rapidly changing world of DeFi and AMMs, we should actively be experimenting with and testing these parameters in order to consistently improve the performance of the protocol over time. This approach is similar to Toyota’s Kaizen philosophy.
To quote the great Mr Dylan, “he who is not busy being born is busy dying”.
A Rational Argument
The chart below is a chart from Bancor’s public Dune Dashboard and I suggest we use it as a base line.
If fees and APYs increase, then it is a strong signal that the change had a positive effect.
*Note: this will actually be evaluated statistically by controlling for other factors such as TVL change, total Ethereum DEX volume and more factors.
I would certainly like to see what the market effect for adjusting fees is but I think this is one of our largest pools so we should be cautious to test things out on it without clear metrics that would prove its benefit.
I think if a test like this was going to happen, the ETH pool makes the most sense as it’s likely to be the most resilient.
Applying Kaizen thought process to every pool, why is it that pool swap fees are different from each other? Wouldn’t we want pool swap fees to be as high as possible without hurting volume in that pool? How often (if ever) do pool fees change?
I certainly don’t know the history here. Before brainstorming solutions, can @mbr or someone else from the protocol tell us why the fees are structured like this? Or is it just that the pool creator set the fee and we’d prefer not to change them.
We should be really careful here since this pool is typically the first hop for many of the trades that occur on our DEX (an assumption on my end given that we operate on the Ethereum network and most people use ETH). E.g. people typically go from ETH → BNT → TKN and at most you are doing two hops. We have to take into account the overall fee because higher fees in the first hop means that the overall transaction will also go up. If I have to take a guess, I think that this trading pair has a lower fee for that specific reason since you do end up paying .3% in the overall transaction (the majority of our secondary pools have a .2% fee) and that puts us in the same ball park as other DEXes.
With that said, not to discourage the OP but typically for these types of proposals we should be holding ourselves to a higher standard (See Bancor Improvement Proposals in A Guide to BancorDAO Due Process) and the onus should be on the proposer to gather supporting documentation, analysis, metrics, expected outcomes, etc… for these types of proposals. We can’t expect the community to make an informed decision without any supporting data.
Yes, most transactions use ETH and they will increase in cost by .1%.
At the same time, this should increase APYs and the rate of vBNT burning.
As for your comment on the higher standard, i agree. When putting this together I wanted to add as much data as possible, but in reality, we really only have baseline measure.
I didn’t want to add a dozen charts that would make the proposal look good unless they added value for the reader.
I like this proposal. We should experiment, observe what happens, and maybe we’ll eventually come up with a set of rules for setting fees, and then this will lead to us automating the setting of fee levels.
@mbr - Can you please have a look at the proposal above? If you think it is ready, please elevate it to Level 2 with the following header:
This proposal is expected to appear on Snapshot for voting on Monday 17th May 2021 at 12:00 pm UTC. Make sure to stake your vBNT for voting before this date and time to participate in the DAO decision.
I find the proposal lacks the standard that it should need for such an important change so just from that, it shouldn’t pass.
Metrics such as volume on the pool and % of ETH TKN swaps on Bancor weren’t analysed. A global picture of the fees that ETH TKN swaps have on competitors vs Bancor wasn’t done. An analysis of the impact of reducing the fee previously wasn’t performed as well. And finally, it would be relevant to know how many ETH TKN swaps come from aggregators.
My apologies for not participating in this thread before. Anyways, this was just a “we can tweak it, so let’s try it, can always go back later” approach and while I don’t condone it, I’d prefer that our community was more diligent when it comes to proposals.
If I’m not mistaken, e.g., Sushiswap and Uniswap swap fees hover around 0.30%. @glenn has already explained this well enough, but I’m afraid we might lose aggregator volume if we increase the fee on the pool. Given how relevant ETH is, this is arguably the pool that should have the lowest fees on Bancor, such that ETH TKN swap fees also hover around 0.30% and we remain competitive.
My opinion could change, but not without a thorough analysis on the metrics stated above, to properly gauge the impact of changing the fee. One of the proposal’s tldr points is “The fee has been at 0.1% for a long period of time and data on performance already exists for that time period.” so we should start by using that data, debate on the forum for longer, and once a sensible and well informed fee proposal can be drafted, then it should be voted on.
If not clear enough by now, this is a hard AGAINST from me.
As the author of the proposal - I fully understand your concerns here. However, let me promise this was not rushed or half thought through.
Rather, I went through lots of charts on the public Dune Dashboard and I’m very familiar with how Uniswap and Sushiswap do this.
My approach, and maybe I’m wrong, is that what we have is data since 2017 showing a baseline measure of the fees on the ETH/BNT pool. The only thing we really know is how it performs right now. If we make this small adjustment, then is the point where we can dig deep and compare results.
But like in the scientific method, this is just a hypothesis. I propose we make this change, collect data and analyze the results.
I stumbled on this thread (looking for something else actually) but it seems really useful for this discussion.
In particular, this post where a large pool provider lays out the rationale behind how pool trading fees are chosen.
It’s a long discussion that defies summarizing. Separate to that, I think the ultimate breakdown is this: higher ETH pool fees would make LPing in that pool more desirable, except that until LM rewards end our data is too dirty to rely on. A small (relative to LM) APY change from fees won’t tell us much about LP habits. Until LM stops we may not be able to know what’s correct in terms of drawing LPs.
Decreasing fees is hard because APYs are already bad, it may disrupt that pool’s economic ecosystem creating a downward spiral, and it will add to the “Bancor is completely propped up by LM” narrative. Increasing fees is hard since we already struggle to get traders and every little advantage helps. (Will traders even notice? Who can say?)
We can only learn what optimal fees are through testing, so it’s hard for me to argue against the proposed change because it’s a reasonable and small amount, and we have to start somewhere or else just admit we can’t try anything.
Generally when companies raise prices, especially a small amount, the effect tends to be net positive for revenue, even if it means a small loss in user base.
That being said, imo, we should be focusing on user/volume growth and not revenue/profits right now. Take a page out of the tech handbook and run at a breakeven to loss level right now to grow market share. Then you can raise prices later.
FYI, this proposal passed and the new fee is active on the ETH-BNT pool. The thread that @bias referenced is a good one in addition to the repeal thread. I guess time will tell what happens and hopefully we will get some useful data in the next 30 days.
I personally (because I have a vested interest) will continue making my swaps through Bancor even if cost me an extra .05% in fees when going from ETH to TKN. I can’t say the same for others who would go to our competitors for pools that have similar liquidity and therefore their orders might not incur much of a price impact. At that point, all they really care about is the exchange that has the lowest fees and to get the best outcome for their trades (I would as well).
Below are some pools which I think we should look at closely in the next 30 days. We have the deepest liquidity and have essentially cornered the market for these tokens on the Ethereum network. :
