Change ETH pool fee to 0.15%

Community Chatroom (DISCUSSION)
1

Proposal Draft: Change the ETH pool fee to 0.15%

TL;DR

  • The ETH pool is the second deepest on Bancor v2.1, with $291,217,616 in liquidity.
  • ETH is involved in most trades on Bancor v2.1 - increasing the pool fee would impact the ETH pool and the trades going through it with ETH as a target token.
  • This proposal drafts seeks to spark debate about increasing the ETH pool fee from 0.1% to 0.15%.

Discussion

  • The DAO has been experimenting with pool fees to understand how to maximize swap fees in the protocol, and thus increase APRs.
  • ETH is the token target token in 15.7% of trades and the pool accrued $4.3m (8.3%) of fees in 2021.
  • In the last month, Bancor v2.1 did $80k of ETH fees (5.2% of the protocol). An increase from 0.1% to 0.15% would be equivalent to $120k, assuming no volume change.
  • With this 50% increase we could take a 30% hit on volume and still positive on fees.
  • All multi-leg (>=3) txns only account for 1.3% of fees ($680k) since 2021 so ETH fees here are not considerable.
  • Weekly fees and volume is shown in Figures 1 and 2.
  • If the APR in the ETH pool fee lowers and the experiment has failed, a proposal to change the fee back to 0.1% will be posted.

image
image2090Ă—614 98.6 KB

Figure 1 - ETH pool Weekly Fees since beginning of July 2021.

image
image2092Ă—614 104 KB

Figure 2 - ETH pool Weekly Volume since beginning of July 2021.

  • Change the ETH pool fee to 0.15%.
  • Keep the ETH pool fee at 0.1%.

0 voters

2

OK, I’m struggling here.

When you say

Would a trade ETH → BNT → LINK be considered “multi Leg” (or is that a 2 leg)?

Where I struggle is in the second order effects when we change the fee on one of the “high volume / key tokens” (ETH, wBTC, DAI etc):

As an example, consider ETH ↔ BNT ↔ MKR trades. And assume that 50% of the ~$50,000 daily trade for BNT ↔ MKR pool is 2 hop via ETH (e.g. Arb trade to Uni pools).

MKR ETH Total
Fee 1.00% 0.10% 1.10%
Volume $25,000 $25,000 $25,000
Fee $250 $25 $275

Then BNT holders (Stakers or protocol) are earning $137.5 per day from the MKR ↔ ETH trade.

If we increase the ETH pool fee from 0.1% to 0.15% we expect to get some reduction in the pool trade. If we also assume (!) that the trade drops by 5%, then I calculate:

MKR ETH Total
Fee 1.00% 0.15% 1.15%
Volume $23,750 $23,750 $23,750
Fee $238 $36 $273

So,

  • the ETH pool is now +ve fees after the increase.
  • the MKR pool is -ve.
  • overall, the BNT holders are down to $136.5 (~ 0.7% reduction).

I suppose one way to express this is:

“If a fee increase is neutral for that pools total income, then the overall effect will be -ve for overall protocol income. This is due to the reduction in 2 hop volume with no corresponding fee increase on the 2nd hop.”

Note, Thinking about is this effect will be larger when looking at a low fee 2nd hop (0.1% + 0.1% → 0.1 + 0.15). So a 30% hit on volume would be ~ neutral for the ETH pair, but -30% on the other hop.

I suppose the good news of the 2nd hop effect is that it turns a 50% increase in fee on the single hop into a 25% to 5% increase on the combined swap… So we would expect a reduced hit on overall volume via the pool…

I’m sorry if this is repeating old discussions, but I’m still getting my head around fee changes. Particularly when there is a significant quantity of 2 hop trades (which is my assumption for Bancor v2.1).

Say we are looking at MKR with a 1% fee. and we are interested in ETH → BNT → MKR (and reverse trades) Then, these are paying 1.1% fee on those trades, if we increase the fee by 0.05% for the ETH leg, the overall fee will be 1.15%.

Assuming 50% of the MKR volume is via the ETH pool (e.g. arb to Uniswap), and that todays $53 K volume is representative. Then we have $265 income in the MKR pool

1 Like
3

While its great that we are looking to optimize fees, I think that for this specific primordial pool we should wait until after B3 is out to see what the effects of single hop swaps have on the fee revenue generated by the protocol. Right now trades from TKN to TKN need to pay:

  1. Ethereum fees 1st swap
  2. TKN to BNT Fees
  3. Ethereum fees 2nd swap
  4. BNT to TKN Fees

In B3:

  1. Ethereum fees for swap
  2. TKN to BNT Fees
  3. BNT to TKN Fees

only a single fee is payed for swapping which should almost cut in half if not more (due to optimizations) the apparent cost of swapping on Bancor.

4

This, would also impact fees when ETH is the source token no? The fees would just accrue on the BNT side but a higher fee tier for the BNT-ETH pool would also affect ETH-BNT swaps.

5

Yeah you’re right, changing the fee would impact ETH swaps whether it was the target token or not.

This discussion is driven by a debate with the Economics team and some analysis performed by them - I didn’t collect or analyse the data, just compiled the info for the proposal draft. Happy to discuss and understand if this is a change we want to explore.

Agree, changing the ETH pool fee has an indirect impact on all of the swaps that go through ETH. Furthermore, as @glenn mentioned, Bancor 3 swaps will have significantly lower gas costs, which means that volume from smaller swaps should route more through Bancor 3 than Bancor v2.1 as there will be a significantly lower relative fee on these (Ethereum + pool fees).

Couple comments:

  • Changing the fee now before Bancor 3 might be a useless exercise - BIP17 sets the ETH pool fee on Bancor 3 to 0.20%.
  • Bancor 3 swaps will have significantly lower gas costs than Bancor v2.1.
  • Swaps with >=3 legs (or >=2 hops) brought only 1.3% of fees since 2021 on Bancor v2.1. However, these swaps tend to be smaller-sized and increasing the ETH pool fee might hurt the volume on these swaps on Bancor v2.1
2 Likes
6

I must admit that I find that number surprisingly low.

I suspect I’m not seeing the entire picture as from my etherscan observations (looking for other purposes…). However, Most transactions I look at that involve Bancor are Arbs and generally look like:
1 ETH → BNT
2 BNT → TKN
3 TKN → ETH
(or the reverse)

{The 1.3% isn’t only capturing 3 Leg trades generated by the Bancor UI (and so missing 1inch, Matcha and arb bots) ?}

1 Like