[BIP-893] Reconfigure the Protocol Swap Fee for the ReCLAMM Pool Type

Governance General Proposal
1

PR with Payload

Summary

This proposal aims to decrease the protocol swap fee for pools of the type “Readjusting Concentrated Liquidity AMM” (ReCLAMM).

Introduction

The ReCLAMM pool type was launched in July 2025. Several pools of this type have been deployed since launch. The dynamics of this pool type have been the subject of intense internal investigation, carried out both through simulation and analysis of real-world data. Comparing the evolution of deployed pools with simulations allowed us to validate the simulator as an accurate tool for interpreting live pool activity. The most salient observation at this early stage is that the protocol fee severely harms the LP positions.

Motivation

Currently, the protocol retains 50% of the swap fees. Compared to a scenario without protocol fees, our projections suggest that, in most cases, ReCLAMM LPs may be facing an annual cost of between 5% and 10%, with potentially higher costs for highly volatile pairs. For the sake of maintaining reasonable fees and ensuring the long-term sustainability of the ReCLAMM pool type, the Balancer DAO should reduce the protocol fee parameter for ReCLAMM pools from the current 50% to a significantly lower value. This modification will allow LPs to make their positions profitable, thus incentivizing liquidity provision and creating a positive feedback loop.

See this technical report for a quantitative analysis, and the following two examples.

Examples

  1. AAVE/ETH simulation.

One example from the technical report is a simulation for the pair AAVE/ETH during the period 2024-01-01 / 2025-10-01. We simulate four positions:

1. ReCLAMM, 10% p-fee (with parameters optimized for LPs performance)
2. ReCLAMM, 0% p-fee (using the same parameters as position 1).
3. Weighted, 50% p-fee.
4. Weighted, 0% p-fee.

The simulations perform arbitrage swaps only, according to the price history. We can visualize the results with the following chart.

imagen
imagen1059×656 68.6 KB

The difference between the positions 1 and 2 represents the impact of the 10% protocol fee on the ReCLAMM pool, while the difference between the positions 3 and 4 represents the impact of the 50% protocol fee on the weighted pool. By analyzing the data, we observe that the first difference corresponds to 1.2% annually, while the second amounts to 2.1% annually. For other pairs, such as ETH/USDC or GNO/ETH, these two values are closer to each other, and close to 2%.

  1. WXPL/USDT0 live pool alongside simulation.

The following chart represents the evolution of the value of an LP token from the WXPL/USDT0 ReCLAMM pool in Plasma chain (red line), alongside three simulated positions: two ReCLAMM pools with same configuration, with and without the protocol fee activated (orange and blue respectively), and a feeless weighted pool (green). The plot on the right shows the values relative to the blue line, for easier visual comparison.

imagen
imagen1296×662 75.7 KB

The proximity between the red and orange lines throughout the entire period indicates that the simulation fits the real-world evolution well. The final relative value of the red line is 0.938, meaning that the impact of the protocol fee on this pool has been approximately 6.2% over just 35 days, which is equivalent to almost 50% annual depreciation. This resulted from the pair’s extreme volatility during this period, combined with an inappropriate protocol swap fee value. The technical report presents the results of applying this method to other pools, with varying outcomes.

Proposal

Based on the quantitative technical report, we propose setting 25% as a tentative default protocol swap fee for every ReCLAMM pool with a price_ratio less than or equal to 6. This policy should apply to both existing and future pools. After implementation, we will continue monitoring the pools’ evolution to assess whether these values are satisfactory or require adjustment.

Technical Specification

Pool list: visit this notion page

The Balancer DAO Omni-sig at 0x9ff471F9f98F42E5151C7855fD1b5aa906b1AF7e will interact with the ProtocolFeeController at 0x212f884252792ebaaa811fb0678444b21c7c2879 and call setProtocolSwapFeePercentage with input values:
pool = each pool address,
newProtocolSwapFeePercentage = 25 * 10^16, corresponding to 25% protocol fees on swaps.

Edit:

Originally, the proposed protocol-fee parameter for ReCLAMM pools was 10%. After some discussions, the author considered 25% to be a more cautious value, mainly for two reasons:
-For some use cases, the trading volume is expected to be fundamentally limited by the market, which imposes a bound on protocol-fee collection.
-To avoid an overly drastic modification.

6 Likes
2

Discussions around protocol fee configurations are generally delicate as they would interfere with the core of Balancer’s business model. What is different here in my opinion is the fact that we designed the new fee model before reCLAMMs were introduced. Concentrated liquidity pools in general behave differently. Even more in the case of reCLAMMs where rebalancing results in “active realization of potential impermanent loss”. Therefore, the global swap fee of 50% could be too aggressive for LPs as outlined in this BIP. Furthermore, reCLAMMs would never qualify to be core pools, so this proposal would not interfere in any way with our current framework.
Given these facts, I would argue that it’s worth a shot to reduce the protocol swap fees to 10% for these pools. I would urge the community and BLabs though to reassess the numbers once the change is in effect to see what impact it had on pool performance.

2 Likes
3

Appreciate the research, but it appears that a drop from 50% to 10% is quite a bold move, especially now that we will face challenges scaling TVL/fees back up from the fallout of the v2 exploit.

Granted that it would be great to lower the fee and bootstrap adoption of this product—perhaps even making even more fees—I’m skeptical that this is the thing that will make it successful. ReCLAMM was poised to be a competitive product for v3 and one of the DAOs cash cows.

Is it possible that the suggested 10% fee distribution could be directed solely to the DAO Treasury and not shared with veBAL? This way the product bootstrapping would be supported by veBAL and not burnt runway. Without fees, I fear there will be low interest from growth/bd/mkt to go to market with this product and all the resources allocated this past year on this great AMM are going to waste.

1 Like
4

Unfortunately the data is showing that the protocol fee at 50% for reclamm is not sustainable. Annualized pool depreciation above 5%, sometimes closer to 10% or above due to the protocol fee for many live pools. This can be shown with the rigorous method illustrated at the BIP for the pair WXPL/USDT0, and contrasted with the simpler method of observing the daily collected fees and comparing that to the TVL.

This depreciation level between 5% and 10% by itself points to the proposed p-fee at 10%.

I believe we should allow more margin for LP positions to grow, so that we attract more liquidity and therefore more fees as a consequence. This policy by itself doesn’t guarantee success, but makes it much more likely.

5

AAVE/WETH has 12% BAL APR - doesn’t this more than cover any LP losses from a 50% protocol fee?

Generally the way I think about this is any BAL incentivized pool needs 50% (perhaps even 100%) protocol fee so Balancer gets as much value back for the BAL we’re spending.

For pools that do not get BAL a lower protocol fee like 10% is reasonable. Notably the plasma reclamm does not get BAL. Could try lowering that to 10%, though I don’t see any reason to believe this would cause TVL to increase.

1 Like
6

It should be noted that the BIP has been adjusted by @Sergio to propose now 25% protocol fees instead of the initial 10%. To me this is much more reasonable and is in line with the industry standard for protocol fee settings for concentrated liquidity products.

2 Likes
7

https://snapshot.box/#/s:balancer.eth/proposal/0x7514382977e4917928ae382f370ce0e589b027175649245d29cae9689145f840

1 Like