GMX: Transition from “Buyback ETH and Distribute ETH” to “Buyback GMX and Distribute GMX”

[gmsolq]

Background

In revisiting my “Buyback and Distribute” (BB&D) proposal from last year, Buyback GMX: A Proposal for Reward Distribution Optimization - Buyback GMX: A Proposal for Reward Distribution Optimization - March 2023, and considering the recent introduction and adoption of BB&D by Gains Network, it has become clear that it is necessary to reassess and reintroduce this proposal. Additionally, feedback from the GMX community, especially the insights and discussion with Woju , and a detailed analysis of GMX’s current situation support the need for this change.

Overview

This proposal suggests changing GMX’s current revenue distribution model from “Buyback ETH and Distribute ETH” to “Buyback GMX and Distribute GMX”. This change aims to leverage the advantages of “buyback”, enhance GMX’s long-term value retention ability, and optimize user experience and protocol strategic flexibility while minimizing model changes and fully preserving the “real yield” advantages.

Details

1. Buyback GMX: (30%+70%) of fees from GMX V1 and 27% of fees from GMX V2 will be used to swapped to $GMX , instead of the current $WETH and $WAVAX. This will create continuous buying pressure for $GMX, promoting token price stability and growth.
2. Default Path to Distribute GMX: $GMX is obtained through swaps from GM Balanced GMX Pool, not minted.
3. Optional Path to Distribute ETH and AVAX: When users Claim or Compound, “Convert GMX to ETH” and “Convert GMX to AVAX” will be provided to ensure the actual user experience remains aligned with before.

Implementation Comparison (using GMX Arbitrum as an example)

Compound Rewards (Before):

Claim GMX Rewards
Stake GMX Rewards
Claim esGMX Rewards
Stake esGMX Rewards
Claim WETH Rewards
Convert WETH to ETH

Compound Rewards (After):

Claim GMX Rewards
Stake GMX Rewards
Claim esGMX Rewards
Stake esGMX Rewards
□ Convert GMX to ETH

Claim Rewards (Before):

Claim GMX Rewards
Claim esGMX Rewards
Claim WETH Rewards
Convert WETH to ETH

Claim Rewards (After):

Claim GMX Rewards
Claim esGMX Rewards
□ Convert GMX to ETH

Data Insights

Average $GMX price in the last 30 days: $27.19

Total fees generated by GMX V1 (annualized/30d): $17,902,288
Estimated 100% of fees from GMX V1 for buyback $GMX (annualized): $17,902,288
Total fees generated by GMX V2 (annualized/30d): $69,370,908
Estimated 27% of fees from GMX V2 for buyback $GMX (annualized): $18,730,145
Estimated total fees for buyback $GMX (annualized): $36,632,433
Average daily purchase amount: $100,362

CEX, DEX $GMX Balance (current): 1,798,405 $GMX
Equivalent $GMX for buyback (annualized): 1,347,215 $GMX
Estimated average daily buying pressure: 3,691 $GMX

Model Impact

1. Maintains the ETH and AVAX narrative while preserving the “real yield” advantages and leveraging the advantages of “buyback”.
2. Lays a solid foundation for $GMX to further become a general-purpose liquidity token and for potential large-scale deployment of GMX Single-sided GMX Pools.
3. Provides better liquidity for $GMX itself, effectively attracting more long-term and large-scale capital.
4. Using $GMX as the base for fee buyback instead of $ETH and $AVAX helps unify GMX’s solution for further deployment on more chains, compared to $ETH, $AVAX, $SOL or $BTC, etc.
5. Solves the issue of zombie addresses once and for all, no longer allocating $ETH and $AVAX to long-term inactive addresses, but using these fees to buyback $GMX.
6. Better value retention. Users are allowed to choose “Convert GMX to ETH”, but as long as the funds not choosing “Convert GMX to ETH” are greater than 0, it’s a strictly positive gain for GMX and $GMX.

Summary

‘Buyback GMX and Distribute GMX (or ETH)’ essentially strikes an exquisite balance between ‘Buyback GMX and Burn’ and ‘Buyback ETH and Distribute ETH’, positioning itself as a middle ground that harmonizes the advantages of both approaches.

More importantly, this change will not affect user experience but can significantly enhance GMX’s value retention, provide strong buying pressure for $GMX, and fully leverage GMX’s advantage as one of the most important revenue sources in DeFi. This is likely to be fully reflected in GMX’s actual performance, breaking free from the long-term undervaluation situation.

Options:

1. Maintain Current Model: Continue with “Buyback ETH and Distribute ETH.”
2. Adopt New Model: Implement “Buyback GMX & Distribute GMX” (with the option to convert GMX to ETH).

Once again, thank you to Woju from the GMX community for their active feedback and discussion, to the GMX Core Contributors for their positive evaluation, and to Gains Network for their detailed explanation of BB&D and for being the first to introduce the BB&D experiment. As a GMX Delegate, it is my responsibility to bring forth numerous proposals and possibilities to GMX, which are then subject to democratic voting by the GMX DAO for decision-making. I believe GMX has always been blessed with a very responsible community, and we can work together to bring more innovative ideas to the development of the industry.

Q - Q's Delegate Profile

[gmsolq]

Reserved for potential future changes

[Seraph]

I am usually skeptical of buyback.
Supporting this due to two important differences

1. the gmx tokens is bought, not minted. Hence, no inflation issue.

2. an option is available to swap back to ETH or AVAX, hence those who prefer ETH or avax are not affected, can continue their preferred choices.

Support and hope a proposal would be out asap!

[0xjunwei]

I am keen on this idea, but i propose the other way around which might be easier for devs too, frontend during claim to have another checkbox ticked by default:

Convert ETH/AVAX to GMX

Link this up to a aggregator (e.g. 1inch) and swap the coin into gmx enabling seamless experience for the user.

This would allow users to choose between eth/avax/gmx, and not kind of forcing gmx onto stakers, still has the real yield narratives and holding eth/avax in the contract in theory would be less volatile than just pure gmx alone.

essentially does a twap buy in to gmx too, since everytime a user claims then their respective eth/avax is converted into gmx instead of every wednesday the contributors have to convert all the consolidated fees and purchase gmx from the market.

[gmsolq]

@0xjunwei

I am keen on this idea, but i propose the other way around which might be easier for devs too, frontend during claim to have another checkbox ticked by default:

Convert ETH/AVAX to GMX

Hi, junwei

I have indeed explored the method you proposed. It is very simple to implement, but I believe it may not fully achieve the effects of BB&D.

@0xjunwei

Link this up to a aggregator (e.g. 1inch) and swap the coin into gmx enabling seamless experience for the user.

There is no need to connect to any aggregator, as GMX can already complete all the swaps internally.

@0xjunwei

This would allow users to choose between eth/avax/gmx, and not kind of forcing gmx onto stakers, still has the real yield narratives and holding eth/avax in the contract in theory would be less volatile than just pure gmx alone.

In both approaches, users are allowed to choose between ETH/AVAX/GMX, and not forced to have GMX allocated to stakers, while still maintaining the real yield narrative. I am not sure this slight difference will not affect the choice of direction, as all are quite volatile assets.

@0xjunwei

essentially does a twap buy in to gmx too, since everytime a user claims then their respective eth/avax is converted into gmx instead of every wednesday the contributors have to convert all the consolidated fees and purchase gmx from the market.

“TWAP” must be implemented; it cannot be done on a weekly basis as this would lead to frontrunning issues. The best way is to set a relatively low threshold based on the aggregated swap fees, and execute the swap once the fees accumulate to this threshold.

[0xjunwei]

Hi Q,

As spoken on telegram:

thinking that if u do gmx to eth users be losing 0.6% to fees due to swaps from originally whatever the amount is to gmx using uniswap rate of 0.3%, not factoring gmx internal swaps due to rates being different. then if user chooses to swap to eth after that, it would again cause another 0.3% fee

while using the default eth now, if we do a eth > gmx if user choose to, then they are hit with 0.3% fee only, this still makes the process invisible but we reduce the number of contract interactions and technical feasibility…. twap purchases etc to buy the gmx using the fees accul would require rewrite of the contract, while technical implementation to do “eth > gmx” would just be frontend and hooking an aggregator/one contract instead of writing a new contract to handle swaps (similar to uniswap v4 hooks).

Also we must factor in pool imbalance within gmx, as handling large swap volumes might result in swap not being executed/infavourable fees.

impact on gmx pools are more than using external sources as shown in the image below:

by sending it to diff pools we allow lp to earn more fees too and prop up liquidity demand there so instead of just staking gmx u can LP gmx and know there is a constant stream of volume. By utilizing uniswap/camelot/whatever other dex gmx is listed on we can drive more volume, aggregators can also hook up to gmx v2 to handle the swaps if the fees are worthwhile. So my idea would still be hooking it up to a aggregator and letting it look for the optimal swap path.

[brawlaphant]

I want to know how it works now, buying eth and distributing eth.

An aggregator makes sense if all fees are indeed buying gmx and we provide the option for gmx to convert to gas tokens.

This would assist in getting aggregators to plug into gm pools as well, could be a worthwhile work stream, the flywheel we need.

[gmsolq]

@0xjunwei

As spoken on telegram:

thinking that if u do gmx to eth users be losing 0.6% to fees due to swaps from originally whatever the amount is to gmx using uniswap rate of 0.3%, not factoring gmx internal swaps due to rates being different. then if user chooses to swap to eth after that, it would again cause another 0.3% fee

Hi junwei,

1. Various types of fees, whether swapped directly into ETH or GMX, inherently have potential rates involved.

2. Your current explanation may imply an assumption: you are assuming that all users prefer to ultimately swap into ETH. Therefore, compared to the original method, now there are two additional fee segments, one for swapping fees into GMX and another for swapping GMX into ETH. However, not all users prefer to ultimately swap into ETH. Based on long-term feedback from GMX, there is certainly a proportion of users who want to swap ETH into GMX. For these users, the improved method actually reduces the fees for swapping ETH to GMX. In conclusion, while the path increases for some users, it decreases for others.

3. So, which users are more important? In fact, we cannot and need not compare. We just need to ensure that the optimal path is chosen for any given scheme, which is undoubtedly necessary.

4. Technical implementation is also a factor. Indeed, before BB&D, the suggestion I proposed was precisely the one you mentioned. However, when compared to BB&D, we should immediately step out of this micro-level requirement to focus on the macro aspect: which approach can truly address the long-term undervaluation problem caused by the value retention deficiency of GMX? Clearly, it is BB&D rather than maintaining the current mechanism with just an added option for users to swap ETH to GMX. Of course, this means rewriting contracts, including necessary audits, which would take approximately 3-5 times more effort and time compared to the simpler modification (both plans have been estimated by GMX Core Contributors). My opinion is that since there is a more fundamental solution, we should not compromise on technical development and instead complete the simpler solution.

5. The goal of this proposal is not to solve a simple user demand for GMX but to reform the economic model and operational mechanism of GMX fundamentally. From this macro perspective, I hope you understand the reason for initiating this proposal and which issues are more critical in comparison.

6. As for the specific buyback plan, as I mentioned before, I still do not know the estimated implementation method from the GMX Core Contributor. However, using GMSOL with a similar mechanism as an example, we might even need to bridge back to Arbitrum for GMX buyback because, currently, GMX liquidity on Solana is insufficient. The specific method would be to set a relatively low threshold considering the comprehensive execution cost, and execute once this threshold is met to minimize the risk of frontrunning. On the other hand, if we collect fees weekly and perform a unified buyback and distribution, there indeed might be potential frontrunning issues. Solving this problem is not theoretically difficult, but it depends on different plans.

7. GMX has a significant advantage in that all assets constituting fees have a balanced pool within GMX. Therefore, there is no need to swap externally, and it can be resolved internally. The additional fees also go to GMX LPs, and some fees might even return to GMX itself, promoting the adoption of GMX.

I hope the above analysis can address the issues well. Once again, thank you for your valuable feedback. If you have further questions, I look forward to continuing the discussion with you.

[gmsolq]

@brawlaphant

An aggregator makes sense if all fees are indeed buying gmx and we provide the option for gmx to convert to gas tokens.

This would assist in getting aggregators to plug into gm pools as well, could be a worthwhile work stream, the flywheel we need.

I don’t think the swap will be a major issue, and I’m not sure if it will trigger such a strong chain reaction externally. Even if done weekly, the buyback amount might be only $500k-$2m each week, which is a very small volume, even for GMX. This trading volume is even lower than GMX’s daily trading volume on Binance Spot. So both the fees and the price impact can be ignored. The key point is the value retention and confidence boost for GMX, allowing us to see continuous (and of course, if it is net buying, it is a relatively high proportion of GMX’s current market cap) capital inflow into GMX. This is the core of this proposal.

[bgolil]

Fully support for buyback.

[Jonezee]

A well-thought-out proposal, with all the important implications considered. I see the advantages overall, and expect it will be worth it.

[FallingSands]

Excellent, this seems like a great approach. Using 100% of the GLP (V1) fees is a very interesting idea, will more than double the impact.

I agree with Q that all swaps should happen on GMX and benefit our LPs. Even if fees were slightly higher in the beginning, the increased volume/fees will draw more liquidity.

We should drive as much volume through our pools as we can, and encourage GMX to host the deepest pool for our own asset.

[Buffzer]

@gmsolq

Options:

1. Maintain Current Model: Continue with “Buyback ETH and Distribute ETH.”
2. Adopt New Model: Implement “Buyback GMX & Distribute GMX” (with the option to convert GMX to ETH).

Thanks for this interesting and well-thought proposal

For the sake of completeness, i would suggest to add a third option for the vote:

3. Adjust Current Model: Continue with “Buyback ETH and Distribute ETH" and implement the option to convert ETH to GMX.

This way, we have an intermediary scenario where people would choose to convert their ETH yield to GMX once claiming/compounding.

[gmsolq]

@Buffzer

For the sake of completeness, i would suggest to add a third option for the vote:

3. Adjust Current Model: Continue with “Buyback ETH and Distribute ETH" and implement the option to convert ETH to GMX.

Confirm, the options would be 3:

1.Maintain Current Model: Continue with “Buyback ETH and Distribute ETH.”
2.Adjust Current Model: Continue with “Buyback ETH and Distribute ETH” and implement the option to convert ETH to GMX.
3.Adopt New Model: Implement “Buyback GMX & Distribute GMX” and implement the option to.

@gmsolq

• Technical implementation is also a factor. Indeed, before BB&D, the suggestion I proposed was precisely the one you mentioned. However, when compared to BB&D, we should immediately step out of this micro-level requirement to focus on the macro aspect: which approach can truly address the long-term undervaluation problem caused by the value retention deficiency of GMX? Clearly, it is BB&D rather than maintaining the current mechanism with just an added option for users to swap ETH to GMX. Of course, this means rewriting contracts, including necessary audits, which would take approximately 3-5 times more effort and time compared to the simpler modification (both plans have been estimated by GMX Core Contributors). My opinion is that since there is a more fundamental solution, we should not compromise on technical development and instead complete the simpler solution.
• The goal of this proposal is not to solve a simple user demand for GMX but to reform the economic model and operational mechanism of GMX fundamentally. From this macro perspective, I hope you understand the reason for initiating this proposal and which issues are more critical in comparison.

Providing more options is a good thing, allowing everyone to choose freely. However, I want to clearly state my position here: I oppose options 1 and 2, and only support option 3. Besides the reasons quoted above, there are additional points:

Currently, what GMX is doing is buying back ETH, storing the week’s fees in ETH, and then distributing it. My proposal is to buy back GMX and then distribute it, storing the week’s fees in GMX and then distributing it. It’s just a different form of storage.

I try to stand in opposition to my proposal, which means I have to argue why buying back ETH is better than buying back GMX. For the actual user experience, if ETH performs better than GMX within a week, then for that week, buyback ETH > buyback GMX; if ETH performs worse than GMX within a week, then for that week, buyback ETH < buyback GMX. However, this relative better or worse is actually a very subtle difference in APR. For example, if GMX outperforms ETH by 5% this week, with the current method, the APR is 10%. But even if switched to buyback GMX, the APR would be around 10.5%, which is not a significant difference.

So where is the significant difference? It lies in the continuous buy pressure and value retention. Let me give a 100% better example regarding the issue of zombie accounts. Currently, there are still many addresses that have exited GMX relying on esGMX to continuously earn rewards, and these addresses have even forgotten to claim these rewards. Distributing in ETH and AVAX means helping reduce their circulating supply, while switching to distributing in GMX means helping more circulating GMX go dormant.

Even if all users choose to convert GMX to ETH (which is impossible), the example above alone already proves that the benefit of switching from buyback ETH to buyback GMX is strictly greater than zero. Therefore, in the case of freely choosing between GMX, ETH, and AVAX, it is very difficult for me to convince myself why I should oppose buyback GMX.

In summary, while having multiple options is beneficial, I firmly believe that option 3 is the best path forward for GMX.

[the4hodlmen]

I invest a significant of my assets into GMX and use fee as a source to pay my bill for years. ETH is although volatile, still to me manageable (through hedging). If this were to be converted into GMX, now I think I can no longer rely on the reliable of this source. It is too volatile and the buying pressure from fees is a drop in ocean compared to market. Not sure if you can give some example that buying pressure help with price for any token. This is also equivalent to company does buy back stock and redistribute in stead of paying dividend which I have never heard of (also dont know why). I accum ETH for my financial freedom goal (as I believe in it) and GMX currently align with my strategy in plain and simple way, which I also guess why it has been success until now.

[gmsolq]

@the4hodlmen

I invest a significant of my assets into GMX and use fee as a source to pay my bill for years. ETH is although volatile, still to me manageable (through hedging). If this were to be converted into GMX, now I think I can no longer rely on the reliable of this source.

You have described your situation and thoughts in great detail, which is appreciated. For your situation, this change might not result in a noticeable difference. You only need to check “Convert GMX to ETH” the first time you claim or compound. After that, you will always receive ETH, and you won’t need to hold any additional GMX.

Currently, the method is to buy back ETH, which means converting all fees collected during the week into ETH and then distributing it linearly over the next week. Real yield is inherently volatile, and no matter what method is used, it cannot achieve risk neutrality with guaranteed returns. The difference between buying back GMX and buying back ETH is very subtle in terms of the risk you bear. The risk you take is mainly determined by your choice of distribution, which is flexible; it can be in ETH or GMX, and it is all done within a single transaction.

@the4hodlmen

It is too volatile and the buying pressure from fees is a drop in ocean compared to market. Not sure if you can give some example that buying pressure help with price for any token. This is also equivalent to company does buy back stock and redistribute in stead of paying dividend which I have never heard of (also dont know why).

@gmsolq

Data Insights

Average $GMX price in the last 30 days: $27.19

Total fees generated by GMX V1 (annualized/30d): $17,902,288
Estimated 100% of fees from GMX V1 for buyback $GMX (annualized): $17,902,288
Total fees generated by GMX V2 (annualized/30d): $69,370,908
Estimated 27% of fees from GMX V2 for buyback $GMX (annualized): $18,730,145
Estimated total fees for buyback $GMX (annualized): $36,632,433
Average daily purchase amount: $100,362

CEX, DEX $GMX Balance (current): 1,798,405 $GMX
Equivalent $GMX for buyback (annualized): 1,347,215 $GMX
Estimated average daily buying pressure: 3,691 $GMX

You can check the data and directly correct the viewpoint; you can never say that an annual buyback of 74.9116% of the current circulating market value is “a drop in the ocean.” You mentioned stocks, so let’s take $AAPL as an example. You can compare the current market value of 3.44T USD with an annual buyback amount of 2.57T USD.

@the4hodlmen

I accum ETH for my financial freedom goal (as I believe in it) and GMX currently align with my strategy in plain and simple way, which I also guess why it has been success until now.

@gmsolq

Claim Rewards (After):

Claim GMX Rewards
Claim esGMX Rewards
Convert GMX to ETH

I fully understand and support your goal, and this proposal will not affect you in achieving it. To summarize, you only need to check “Convert GMX to ETH” the first time you claim after the upgrade. This will then become your default option, and you won’t bear any additional unpredictable GMX risk exposure. This adjustment will be seamless for you. I hope this answers your question.

[brunopenandes13]

Could you please clarify where you are getting the buybacks being 74% of circulating supply from? equiv annualized buybacks from your proposal at current price is c. 1.34m tokens
Coin market cap has circulating supply as c. 9.5m tokens

[gmsolq]

@brunopenandes13

Could you please clarify where you are getting the buybacks being 74% of circulating supply from?

https://dune.com/saulius/gmx-analytics

[xdev_10]

Thank you for this proposal

For the buyback mechanism, would like to propose the following process:

• Fees are swapped to ETH or AVAX, using GMX V1 / V2 swaps as it is currently
• The portion of ETH / AVAX allocated for buyback is deposited into a Buyback contract
• The Buyback contract will allocate 1/ of the deposited ETH / AVAX to buy GMX every for 7 days
• The Buyback price will be based on the GMX Chainlink oracle price (available on Arbitrum and Avalanche)
• There will be premium to buyback the GMX, which gradually increases from 0% to 5% over the

An example of the interval value can be 1 hour, which would split the purchases into 168 buys over 7 days, if the size of a purchase per interval is greater than e.g. 20k USD, the interval may be decreased for smaller but more frequent buys

This is to reduce the risk of front-running / price manipulation from a single purchase

The GMX bought back can then be distributed in the following week

With this format it also means that during the transition from rewards in ETH / AVAX to rewards in GMX, there will be one week where GMX tokens are being bought back and there are no tokens to distribute, for this week, GMX matching the estimated buyback amount could be minted for distribution, or if this one week gap is fine, then this mint may not be needed.

There is an alternative option to use CoW Swap TWAP to buyback GMX, but developing the above buyback mechanism should not take too long, so could be done to reduce external dependencies. CoW Swap is also not at the moment available on Avalanche.

[gmsolq]

@xdev_10

There is an alternative option to use CoW Swap TWAP to buyback GMX, but developing the above buyback mechanism should not take too long, so could be done to reduce external dependencies. CoW Swap is also not at the moment available on Avalanche.

Thank you Xdev for your detailed implementation description, which I find very reasonable. One point I would like to emphasize is that compared to relying on external implementations, completing the entire mechanism within GMX can significantly reduce the risk of attacks, making it worthwhile to undertake some additional development. And I am glad to know “it should not take too long”.