Proposal: Stablecoin pools fee experiment and analysis

DAO Archive (SUBMITTED)
1

This proposal is expected to appear on Snapshot for voting on 2022-01-02T00:00:00Z. Make sure to stake your vBNT for voting before this date and time to participate in the DAO decision.

TLDR

  • This proposal seeks to set the USDT/BNT, USDC/BNT and DAI/BNT pool fees to 1%, 0.5%, and 0.2%, respectively, from 0.2%.
  • There isn’t enough evidence to conclude that there is a linear relationship between pool fee and volume, given that many other factors influence volume in the pools.
  • The three stablecoins serve as an interesting experience, as we can set their pool fees to different values and evaluate the impact of the change.
  • The goal is to ultimately increase APYs for these pools.

Abstract

Weekly volume and fees from the USDC, DAI and USDT pools from the week of the 30th of August have been consistent (Figure 1 and 2).

1600×414 202 KB

Figure 1 - Weekly Fees for DAI, USDC and USDT pools.

1600×409 210 KB

Figure 2 - Weekly volume for DAI, USDC and USDT pools.

Total volume on Bancor for intervals ending in November 2021 can be found in Figure 3.

769×784 38 KB

Figure 3 - Total Volume on Bancor for intervals this year

And since there were fee changes for all the stablepools to 0.2% in February 2021, we can look at the Bancor Volume at the 7, 14, 21 day intervals either side of the change (Figure 4).

1280×673 65 KB

Figure 4 - Volume on Bancor either side of Fee Change.

Motivation

An analysis on the effect of the pool fees on pool volume conducted on the USDC/BNT, USDT/BNT, and DAI/BNT pools showed that there is not enough evidence to support a linear relationship between them (Figure 5, Figure 6, and Figure 7).

For the DAI pool (Figure 7) there is a minor (not statistical) decrease in market share upon changing the fee 0.1% → 0.2% as measured at the 7, 14, 21, and 30 day intervals. While the overall market share has significantly increased (300 days either side).


Figure 5 - USDC pool fee changes with market share before and after the Fee change. The 1000, 2000 and 3000 x axis values represent a 0.1%, 0.2% and 0.3% fee, respectively.


Figure 6 - USDT pool fee changes with market share before and after the Fee change. The 1000 and 2000 x axis values represent a 0.1% and 0.2% fee, respectively.


Figure 7 - DAI pool fee changes with market share before and after the Fee change. The 1000 and 2000 x axis values represent 0.1% and 0.2%, respectively.

Mark’s comment on the TRAC pool fee experiment sheds some light into the pool fee topic.

In conclusion, there isn’t enough data to conclude that changing the pool fees is affecting the volume. With this experiment, the USDC/BNT, USDT/BNT and DAI/BNT pool fees will be set to different values, and an analysis and comparison between them, similar to the analysis on the TRAC pool, will be performed.

In the end, this trial alongside the TRAC pool one, will hopefully provide more data to assist in choosing appropriate fees to optimize APYs for the LPs. Further experiments may be presented to the DAO if the results from this one are inconclusive.

For

  1. Change the pool fees for the USDT/BNT, USDC/BNT and DAI/BNT to 1%, 0.5%, and 0.2%.

Against

  1. Keep the pool fees for the USDC/BNT, USDT/BNT and DAI/BNT at 0.2%.
3 Likes
2

Same as the AMP pool, we are currently subsidizing these pools with $BNT rewards and any changes in volume that could affect the fees collected should be offset by LM rewards. It is good for us to experiment with the stable pools to figure out what the optimal fee should be. Are we planning on raising them all at the same time so that all three stable pools will always be at the same fee level?

With that said, I think we should also monitor the volume on the wBTC and ETH pools as I suspect that with higher fees on stable pools that people coming in from stable coins might actually be swapped to ETH/wBTC before routing through Bancor. I am still under the impression that most of our traffic is ARB-based with some from aggregators. I expect that for either of these scenarios that we might see the volume on the other two pools go up once these stable pools hit a high fee percentage.

Will the schedule to follow for changing fees (1%, 0.5%, .2%) be on a weekly basis?

2 Likes
3

THANK YOU!! I have been wanting to do this for a very long time.

I think I actually wrote a similar proposal a few months back.

I see no reason not to raise the rates at least for a test. I fully expect the 1% pool to be the more performant. But who knows, I’m a degan, not a psychic.

2 Likes
4

This is a really well thought out proposal, looking forward to maximizing our yields on stables.

1 Like
5

I need to be more involved in the forums…

I like this proposal and think it’s good to try such experiments and see what happens. As @glenn says, it would be good to see if people / bots are getting around it. In addition to the ETH and wBTC pairs we could see trade routed by curve or uni 0.05% stable coin pools.

I suspect it will take a while to see such routing in the wild, but it will happen. (obviously, in the long term I would expect all stable coin pools to be within 0.05% of each other)

I would have liked to see a sunset (4 weeks?) where the change would be reverted to 0.2% unless there was a proposal to fix it.

Also in the spirit of governance minimisation, a planned experiment e.g. fee for each pool changing in 2 week blocks over 6 weeks. Might be more useful in generating data.

1 Like
6

This is a good point and I had set a reminder to look into setting the stablepool fees back to the original values or others, should the change be justified, in 2 weeks.

In favour of this as a follow-up proposal if we don’t collect enough information to help us make a decision on what the stablepool fees should be!