Proposal: TRAC-BNT fee change schedule and performance analysis

[tfns]

Proposal: TRAC-BNT fee change schedule and performance analysis

For this proposal to pass, it requires 35% quorum and 66.7% supermajority.

This proposal is expected to appear on Snapshot for voting on 2021-12-19T17:00:00Z. Make sure to stake your vBNT for voting before this date and time to participate in the DAO decision.

TLDR

• Schedule to follow 1.0%-2.0%-4.0%-2.0%, with weekly changes.
• We will analyse TRAC Daily Volume, Daily fees, Total Volume and Total Fees vs fee change data on Bancor as well as TRAC volume and fee share across DEXes on Ethereum with every fee change.
• In the end, this trial will provide more data to assist in choosing the fee that seems to maximise fees/APR for the LPs.

Motivation

Here are some examples of graphs that will be provided as an analysis once the experiment is over.

Figure 1 - Average volume per day in the TRAC pool, in USD, for two different fees.

Figure 2 - Average fees per day in the TRAC pool, in USD, for two different fees.

Figure 3 - Total volume per day in the TRAC pool, in USD, for two different fees.

Figure 4 - Total fees per day in the TRAC pool, in USD, for two different fees.

Figure 5 - Bancor Daily Volume in the TRAC pool, per day.

FOR
Change the fee weekly according to the schedule: 1.0%-2.0%-4.0%-2.0%.

AGAINST
Keep the fee at 0.5%.

[mbr]

It’s not a bad idea.

Analysis of fee changes the DAO has made up to the present suggest we are still in the noise - in general, fee changes make zero difference to trade volumes, and understandably so. To a wide variety of users, the difference between a 0.2% and a 0.5% swap fee is essentially zero. More importantly, lower fees to drive higher volume is not commensurate with increased returns; Uniswap v3 has attained some truly remarkable volumes, but is currently running at a $20M loss. This is of course an exceptional case - their concentrated liquidity model makes it much easier to lose money, but also attracts more volume, and therefore still serves the point of the discussion.

It is not the case that we ought to be pursuing volume for volume’s sake. More often than not, the fee wars that are driving the fight for increased trading volumes might be missing the point. The customer (trader) is only an asset if their activity benefits liquidity providers in the long run. Looking around the ecosystem and observing how much damage low fees are causing to liquidity providers in rival protocols, I am very interested in an investigation into higher fee models as a means to fortify APYs in the network. To some pool operators, the obviousness of this suggestion is comical.

Since its inception, the wNXM pool has maintained a relatively high swap fee, and as a result has consistently performed well - especially for its relative depth (which is considerable). The ARCONA pool has also performed exceptionally well - and I still consider it the gold standard in Bancor whitelisting practices - thanks largely to its sensible swap fees. It is easy to get distracted by the flashy KPIs and impressive volumes that are thrown around by commentators, but as a community we should be paying attention to our bottom line.

In addition to the analysis proposed by the author, I think we can afford a slightly more rigorous handling of the data to evaluate the financial consequences of the decision.

I am looking forward to supporting this.

As an aside - the Automaton will continue to observe its own policies, and will vote AGAINST this proposal. However, I think the results of this experiment (and the report being compiled from the DAO’s previous trials) will be sufficient to change the Automaton voting behavior with confidence. Anyone who has delegated to the Automaton, and is in favor of trailing these fee experiments, is reminded that they can simply vote manually to override the Automaton’s decision.

[glenn]

Thanks for putting this together. I support the experiment and will be voting “for” this proposal. I think that the analysis will provide us with some good insights as to what the correct fee should be for pools with the highes depth on Bancor. My thesis for tokens that we have the largest moat for has always been that we can charge a higher fee premium (up to ~1%). I think that this intuitive since if you have no competition then there isn’t an alternative for buyers. The opposite is also true, you can’t charge a higher fee premium if you have competitors with equal or higher liquidity for certain tradings pairs.

Typically, I have been careful on going above 1% since this is the highest fee that one can set on uniswap v3 for volatile pairs. Granted, that the tokens that we are examining might not be volatile at all and the experiment will still be valuable since we will be able to gleam some useful data from our end for potential fee improvements and how high we can go. With that said, what ever we accomplish here should probably be reviewed at some interval every year in case another pool that’s more competitive and with equal depth comes online that starts undercutting our volume for a certain trading pair.