Summary
This post evaluates the potential implications of introducing the new fee structure suggested by @gmsolq for the GMX platform. We begin by clarifying our understanding of the proposed changes, followed by an analysis of potential risks and benefits. Additionally, we utilize historical data to assess how these changes might impact user experience and protocol revenue. We also discuss specific concerns regarding the shift to post-impact price fees.
Our findings indicate that while the lower fees could attract more users and potentially increase total revenue, the overall impact on users may be minimal. Given the modest benefits versus the potential risks, we recommend deferring the decision on this change to the community. The data provided aims to facilitate an informed discussion and decision-making process.
New Fee Mechanism Clarification
First, we would like to confirm our understanding of the new fee structure suggested in the discussion and help everyone else follow along:
We understand that the intention behind the new fee structure is to leverage price impact as a means to finance borrowing rates. In other words, the borrowing rate is utilized to mitigate the price impact fees.
To clarify, assuming that if fee > 0, the user receives money and if fee < 0 the user has to pay, we interpret the definition of post_position_price_impact as:
Lastly, we would like to address a point made by @gmsolq in a follow-up message:
âPlease distinguish between trading fee and order fee. The trading fee is the comprehensive trading cost, which includes the portion paid to LPs and the portion paid to holders of the opposite position. The focus of the current proposal is to maintain the portion of the trading fee paid to LPs while reducing the portion paid to holders of the opposite position, thus lowering the trading fee without changing the order fee, while still providing the same protection against attack risks.â
We want to clarify that both the borrowing fee and the price impact contribute, at least partially, to the returns of GMX holders. The only fee paid directly to other users is the funding fee, and altering it could compromise the OI balance. Thus, GM holders would be directly affected by this change. In the following sections, we will quantify this impact to provide a clearer understanding of the potential consequences for every participant.
Risks
While we acknowledge that the total fee will remain strictly higher than the previous impact fees, thus protecting the protocol from rapid price manipulations, we have identified other potential issues:
- Liquidity occupation: The proposed fee strategy will effectively decrease the cost of simultaneously opening long and short positions. We plan to elaborate more on this concern in future discussions.
- Max Price Impact fees parameters: Currently, there is a maximum price impact fee, and cap on max price impact taken into account in order to trigger a liquidation, currently set to 0. These parameters are important to ensure that the collateral is always sufficient to cover the price impact. Since higher impact fees are expected to be paid when closing a position if post-position price impact is implemented, these parameters may need to be redesigned. One possible solution could involve charging a pre-impact fee and offering a larger rebate at the end.
Benefits
We recognize that offering zero-slippage trades can be a powerful marketing strategy. Additionally, this new strategy primarily benefits long-term users rather than short-term traders, which helps to create a more stable OI balance and potentially stabilizes the APR of GM tokens.
Historical data
To evaluate the implications of the potential new fee design, we conducted an analysis of historical positions over the last month. This analysis helps us understand how the proposed changes could affect both user experience and the protocol itself.
In the attached image, we present histograms for each market showing the percentage reduction in price impact and the total fee for each position. Additionally, the title includes the following key metrics:
- Negative Impact Position: Total number of positions that end up paying impact fees, calculated by summing the impact fees paid at both opening and closing. These positions are the only affected by the new fee mechanism.
- Price Impact Decrease: Represents the total fee reduction as a proportion of the total price impact fees paid. This calculation does not consider positions that end up receiving impact fees.
- Total Fees Decrease: Quantifies the overall loss in fees the protocol would incur due to the proposed changes.
- Gini Index: Used here to measure the distribution of fee reductions across positions. A Gini Index close to 1 indicates that a few positions receive disproportionately large discounts, suggesting high market concentration. Conversely, a Gini Index near 0 implies that the discounts are distributed more evenly across all positions, making it more impactful for the community.
Arbitrum
Summary
| Total Fees |
3.8M |
| Total Impact Fees (over negative impact fees positions) |
1.4M |
| Total Hypothetical Fee Reduction |
0.1M |
| Zero Slippage trades |
8.5% |
Avalanche
Summary
| Total Fees |
242K |
| Total Impact Fees (over negative impact fees positions) |
130k |
| Total Hypothetical Fee Reduction |
8.8K |
| Zero Slippage trades |
17.6% |
Analysis
As we can see in the previous plots, the impact on Price impact decrease changes significantly from pool to pool. This could be due to higher borrowing rates or due to users holding positions for longer. However the decrease in the total fee is never significant, almost never passing 5%.
From the analysis of the previous plots, itâs evident that the impact on Price Impact Decrease varies significantly across different pools. This variation could be attributed to higher borrowing rates or the tendency of users to hold positions for extended periods. However, itâs important to note that the reduction in total fees due to these changes is relatively minor, with reductions almost never exceeding 5%.
Singular token pools exhibited the most significant effects under the proposed fee structure changes. However, itâs important to note that these pools do not incur price impact fees for now, which shifts the relevance of our findings. Additionally, in several smaller markets, particularly on the Avalanche network, nearly 50% of historical transactions could have achieved zero slippage. Despite this, the decrease in the total fees paid by users rarely exceeded 10%.
Risks of moving from pre-impact fees to only post-impact fees
While we recognize that pre-paying impact fees can be a barrier, for most of the positions there should be no expectation on receiving price impact fees when closing their position.
This rebate is only expected for a user who plans to significantly and temporarily unbalance the poolâtypically sophisticated traders who are likely already factoring potential rebates into their strategies.
Furthermore, the proposed change to charge fees only at the close of positions could complicate liquidations or the closing of positions at maximum leverage. GMX has established a maximum price impact fee for closing positions and consequently liquidations to mitigate bad debt risks. Implementing post-impact fees would likely necessitate the removal of these safeguards. Without them, positions could be forced to close due to high OI imbalances, potentially exacerbating these imbalances further.
Conclusion
We acknowledge that the new fee structure could alter user dynamics. In theory, lower fees might attract more users, potentially increasing the total revenues for the protocol. However, our analysis suggests that the change might not significantly impact users, raising concerns about whether the benefits justify potential risks and efforts of development.
While we believe it is possible to construct the new fee mechanism safely, the anticipated impact may not be substantial enough to warrant exposing GMX to new risks. Therefore, we propose leaving the final decision to the community. We hope that the data presented will assist in making an informed choice.
We are open to discussing new ideas and are prepared to conduct a more thorough analysis of the risks mentioned should the community find it worthwhile to implement the new fee structure. We look forward to engaging further and supporting the community in this decision-making process.
