Gauntlet Fee Update

Hey everyone,

We over at Gauntlet have been continuing our AMM governance research and wanted to give you an update on our approach to swap fees. You can read up on our current approach to AMM governance and some new results here:

We’d love to get your feedback on how to think about some of the tradeoffs between LP and Trader utilities and how they will fit into the communities vision going forward.

1 Like

tldr low fees good for traders (majority of whom are arb bots), high fees good for LP’s. Shouldn’t be a surprise, I’ve said it from the beginning.

Real issue is the belief that low fees will bring human traders, as we accumulate human traders we can begin raising fees. Hurt our LP’s in the short term to benefit them in the long run. As far as I’m aware, there is zero real world evidence to support this belief. Balancer goes on either way tho. I’ve learned to let this go and embrace low fee nirvana.

Comprehensive post with much content and data to unpack. I like the model of stakeholder utility functions. For me the takeaway of the post are two questions:

  • How can Balancer design the best combination of stakeholder utility functions?
  • On which stakeholder utility functions should Balancer place the highest weights?

A Brief summary of how I understood the article

The part on purely arb volume:

  • Your simulations show that with decreased fees, arb volume would increase (figure1)
  • LPs fare slightly better if swap fees are higher (figure2)

An LP, however, decides his capital allocation mostly on incentives. He is therefore most sensible on changes to incentives. He is therefor most sensible on changes to incentives. Based on this information one can reason the following:
Reducing swap fees could bring higher utility to a trader (arber) compared to what an LP would lose in utility due to lower return on fees generated based on his pool share.

Taking non-arb volume into account:
Hypothesis: LP revenue can increase if Balancer bumps fees up given enough non-arb volume. However, most of the volume as of now seems to be from arbs. Therefore this hypothesis needs to be treated with caution.

Comparing arb and non-arb volume:
as of right now the determining factor for an aggregator if a trade is routed through Balancer is pool liquidity which the simulations from gauntlet show.

Following up with my interpretation:

  1. To me, Balancer has increasing arb volume solved? - “Just lower fees”
  2. So as of right now, the more challenging part is getting non-arb volume.

It is hard to voice an opinion currently. Also:

As far as I’m aware, there is zero real world evidence to support this belief.

What would be good data points I could take look at to derive an opinion on this?

Looking forward to readings other people’s thoughts.

1 Like

Thanks for that summary @Mkflow27. A few responses:

Your simulations show that with decreased fees, arb volume would increase (figure1)
LPs fare slightly better if swap fees are higher (figure2)

Yes, arb volume should see a mechanistic increase with lower fees, simply do to the fact that more profitable opportunities will arise with lower fees. This again would help the Balancer spot-price more closely track external prices, so could potentially be a part of Balancer’s protocul level utility (e.g. Balancer is great because I can always expect the spot price to reflect the price off chain)

And yes, slightly higher fees will probably be slightly better for LPs, but LPs likely care much more about incentives, so the rewards program is likely the best lever for attracting capital.

Hypothesis: LP revenue can increase if Balancer bumps fees up given enough non-arb volume. However, most of the volume as of now seems to be from arbs. Therefore this hypothesis needs to be treated with caution.

Yes, we have estimates of roughly how much of this volume is from arbitrage, though, so we can take that into account when setting fees. The real question, though, is how to weight the tradeoff between raw volume and fees from that volume.


Compare the top two rows: In the first case, trader utilities are quite high and we see more retail volume; in the second case we expect lower trader utility and higher LP revenue. Should we maximize swap fee revenue (regardless of whether its from arbitrage) or should we instead try to favor retail volume at the expense of some extra fee revenue?

2 Likes