DRC - Shifting from LP rewards toward market maker rebates

[max-holloway]

tl;dr

• Market makers are a critical component of the dYdX exchange, and to compensate them for the liquidity provision service that they provide, they receive incentives.
• Currently, top market makers receive incentives in two forms: a rebate on their trading fees, and LP rewards each epoch.
• We suggest that the rebate be increased from 0.5bps to 0.85bps, and that the per-epoch LP rewards be reduced by 50% from 1,150,685 DYDX/epoch to 575,342.50 DYDX/epoch.
• We expect that this reduction will lead to improved exchange liquidity at or near the the top of the book; it will be roughly economically neutral for LPs; and it will meaningfully reduce monthly DYDX emissions.

---

The dYdX LP rewards system has been the source of much debate, which has primarily revolved around modifications to the rewards formula. While there may be additional research to help us optimize the LP rewards formula, we do not focus on that here. Instead, we discuss changes to be made to the quantity in rewards itself.

With a current expenditure of 1,150,685 DYDX/epoch (~$2M USD), the LP rewards program constitutes a major source of emissions for the DYDX token. These rewards are distributed to market makers based on their Q score, a metric computed based on market maker’s uptime, depth, and spread. While this score has undergone changes, it does not strictly reward makers based on volume, which has led to unproductive liquidity: liquidity that earns rewards but has low likelihood of getting filled. Although the introduction of a volume term has helped move the exchange in the direction of more productive liquidity, I believe it makes sense from the community and production LPs’ perspective to move incentives away from the Q-score LP rewards formula and toward a market maker rebate.

Market maker rebates reimburse market makers for the liquidity that they provide that gets filled. Rebates are common across many crypto exchanges, including Binance, BitMex, and others. Currently, dYdX offers makers a 0.5bps (i.e., 0.005%) rebate. Similarly to the volume term in the current LP rewards program, makers increase their rebate when they take the other side of a taker’s order. However, dissimilarly from the LP rewards program, maker rebates come in the form of reduced maker fees, rather than in the form of a DYDX reward.

We suggest that the dYdX community move toward a lower LP rewards program and a higher maker rebate, for the following reasons.

1. The LP rewards program is costly to the dYdX community, as it is a source of high DYDX emissions. Lowering the LP rewards emissions lowers DYDX emissions.
2. The LP rewards program has historically had issues with market makers providing unproductive liquidity that earns rewards but does not get filled.

Consideration 1: this changes the shape of the orderbook

It is worth noting that this change will have implications on dYdX’s liquidity. For instance, there will likely be a greater liquidity at the top of the book (i.e., closer to best bid / best ask) as well as lower liquidity further from the mid price. We anticipate that very large trades will receive worse pricing after this change, as opposed to before. Conversely, most of the existing dYdX volume will receive better pricing as a result of this change, as evidenced by the graphs below.

Figure 1: MATIC market depth, as measured from number of basis points (bps) back from the mid price. The y-axis is depth, in units of MATIC.

Figure 2: MATIC fill depth, as measured from number of basis points (bps) back from the mid price.

For example, on the MATIC market, we see from Figure 1 that there is an increase in liquidity that spikes at 40bps away from mid price, which corresponds with the 40bps market cutoff for LP rewards. On the other hand, we see from Figure 2 that only a tiny portion of the MATIC volume hits 40 bps away from mid; in fact, only a tiny portion of MATIC volume eats through more than 10bps. Clearly, most of the volume on the exchange is concentrated close to the mid price, at least on the MATIC market.

We see a similar, although more nuanced relationship on ETH markets, where there is presumably a larger portion of organic depth.

Figure 3: ETH market depth, as measured from number of basis points (bps) back from the mid price. The y-axis is depth, in units of ETH.

Figure 4: ETH fill depth, as measured from number of basis points (bps) back from the mid price.

Consideration 2: why not do {tiered maker rebates, additional changes to the LP rewards formula, …}?

This proposal is meant to be as simple as possible. Some other community members have made compelling cases for other changes, such as Colin Chan’s “Liquidity Providers’ Incentive Programme” and Callen’s “DRC - Introduce a Market Maker Rebate Program”. There are further optimizations that can be made to the maker rebates program, such as tiered incentives for top makers that commit to filling large amount of volume on the exchange (link), or changing the volume term in the LP rewards formula, or changing the uptime term in the LP rewards formula, or squaring the spread term in the LP rewards formula, etc. However, for the sake of minimizing governance complexity, I would like this change to focus on a simple decision: should we increase the rebate and decrease the LP rewards emissions, or not?

Recommendation

As a first step toward reducing LP rewards, we suggest lowering the LP rewards program by 50% (to 575,342.50 DYDX/epoch) and setting the maker rebate to 0.85bps (i.e., 0.0085%) on all markets; we chose these parameters to keep total LP incentives — from rebates and LP rewards — roughly the same as they are today. The saved LP rewards emissions would vest to the rewards treasury, which can then be utilized as dYdX governance sees fit.

Expected Impact

• Impact on LPs. Using estimates of the DYDX price and LP rewards-seeking behavior, we find that this change would be roughly economically neutral for LPs.
• Impact on dYdX community. This would lower DYDX emissions, i.e. dYdX community spending, by 575,342.50 DYDX per epoch. At current DYDX prices, that is a $1M+ saving per month to the community. This change is overall positive for the community.

• Impact on traders. Smaller traders will unilaterally benefit from this change whereas larger traders will see higher execution costs. Based on historical execution pricing and orderbook data, the volume-weighted execution cost across all traders will decrease, so long as the distribution of taker order sizes remains similar. This change is overall positive for traders.

[foxlabs]

Dear @max-holloway and the dYdX community,

We appreciate your detailed proposal regarding the shift from LP rewards towards market maker rebates. Your understanding of the dynamics at play within the dYdX exchange ecosystem is clearly reflected in the rationale and methodology of your recommendations.

We agree with the core proposal and intend to vote in favour of it. We also believe that the strategic restructuring of rewards as presented would be beneficial to the platform’s liquidity and efficiency.

However, while endorsing this proposal, we also want to bring forth a supplementary suggestion, that focuses on the affiliate system, which is already an integral part of the exchange.

To capitalise on the proposed shift in the reward structure, we suggest a subsequent proposal for enhancing the current affiliate system. Specifically, we propose that a portion of these newly restructured rewards be allocated to affiliates each time they bring on a new account that reaches a specific turnover volume. Concurrently, new accounts themselves could also receive a reward once they reach a defined turnover volume.

This twofold approach not only incentivises affiliates to attract new users but also motivates the new users to actively participate in the exchange. It essentially serves as a growth catalyst while also fostering a sense of community and mutual benefit.

We realise this would be a significant change and would necessitate its own discussion and vote. Hence, we are suggesting it as a second part or a separate phase to the initial proposal. If the community and stakeholders agree with our vision for enhancing the affiliate system, we can jointly work towards defining the specifics and timing of this proposal.

As always, we welcome further dialogue on this suggestion and are eager to hear the thoughts of the community.

Best Regards,

Kagan

[Nascor]

Thank you, Max, for the proposal. We should have reduced trading rewards more aggressively long ago, so I think this proposal is overdue. By now, I would argue that we should eliminate trading rewards completely. The majority of users trade on dYdX not due to trading rewards but because they have self-custody + access to deep liquidity and a reliable service, which no other DEX offers. There is no second-best DEX out there to trade perps, so “incentivizing” users makes no sense to me. The only reason why I would agree to have a 575k emission per epoch is because we are launching V4 soon, and it just looks better on paper. V4 is not yet battle-tested like V3, so yes, emissions still make sense until we reduce them to zero post-V4 launch.

[RealVovochka]

@Nascor Max is proposing to reduce LP rewards, not trading. I raised the problem about reducing rewards in general in other topic. Happy if you leave your comment there

[steve2gold]

About time. if it’s going to affect some important future metrics, then I think this should be looked into in details. Kudos to the author.

[Nascor]

awww my bad, got all excited

[Callen_Wintermute]

Hey Max,

thanks for the proposal!

We support this proposal and agree that these changes are net beneficial for the protocol and strike a good balance between the two reward mechanisms. Ultimately, LPs should be able to quote better based on fixed rebates instead of accounting for a fluctuating DYDX price.

[she.trades]

Overall, positive.
However, I still have a question.
@max-holloway, how would you respond to those who claim that such a change in LP rewards supports concentration of liquity providers by removing smaller ones and bolstering larger ones?

Finally, a less diversified and more concentrated LP base represents a risk for the DYDX protocol.

[0xCLR]

Hey @max-holloway great proposal, I like it for a number of reasons

• Simple is better than complex.
• It naturally scales the cost of liquidity with the amount of trading activity.
• It incentivises competition amongst market makers.

I have a few questions on the implementation:

• Firstly, I’m not really familiar with the rebate program. Is this referring to this snapshot? Or something else?
• If so, is the idea to shift to a fixed rebate for all eligible market makers? Personally I like fixed for its fairness and simplicity.
• Is this proposal meant to affect v3, or just v4?

Again, full support for the proposal as I understand it – increasing the rebate 0.35bps and halving LP emissions in v4.

The expected take rate on the exchange is around 2.5bps. So adding 35bps to a market maker rebate, and assuming MM’s make up 70% of limit orders this represents around 0.25bps of that leaving 2.25bps for the protocol. This is much healthier as the cost of liquidity scales with actual usage and fees. The reduced emissions should help drive value to the token long term driving a lot of value to the DAO.

[fig]

Max,

Great to see you again here.

Xenophon Labs has been diligent in its process, research, and investment into protocols such as dYdX. We have been impressed by their quality of work in Aave and elsewhere.

This proposal seems to introduce key benefits into dYdX’s operations:

• increased rebates, by .35 bps
• decreased DYDX emissions; lower spending / increased rewards treasury
• positive experience for smaller traders, more predictable rates

We’re supportive of this sort of change and perspective during the advent of V4.

My questions remain around these cells:

Forgive my ignorance, what is this showing? Is this the resulting loss in revenue due to the change?

If so, I’d be keen to understand Trading’s perspective on this.

[max-holloway]

Thank you everyone for your thoughtful responses. I’ll address some of the questions that popped up.

---

@she.trades

@max-holloway, how would you respond to those who claim that such a change in LP rewards supports concentration of liquity providers by removing smaller ones and bolstering larger ones?

The current LP rewards formula has a cutoff at .25% of the epoch’s maker volume, which is somewhere around $75M in volume per month. In comparison, the cutoff for the rebate is $200M in volume per month. This change will be negative for LPs who provide between $75M and $200M in volume, but it will have no effect on the small LPs that do less than $75M. It is worth noting that this is not a tiered rebate program, so it is an equal-fee game between all MMs that do >= $200M in monthly volume.

---

@0xCLR

• Firstly, I’m not really familiar with the rebate program. Is this referring to this snapshot? Or something else?
• If so, is the idea to shift to a fixed rebate for all eligible market makers? Personally I like fixed for its fairness and simplicity.
• Is this proposal meant to affect v3, or just v4?

• This is in reference to dYdX Trading’s market maker rebate program. This proposal is meant to affect v3 only, as it is in reference to the dYdX v3 market maker rebate program.
• Yes, this proposal would just be for a fixed rebate. I agree with you, the fixed rebate is simple and fair. I leave it up to the community in future proposals to determine if tiered rebates’ benefits are worthy enough to move away from fixed rebates.
• Great question, I should have clarified. This proposal is strictly meant for dYdX V3. Perhaps something like this will be relevant in the v4 world, but let’s stick with v3 for now to keep voters’ decision simple.

---

@fig

My questions remain around these cells:

Forgive my ignorance, what is this showing? Is this the resulting loss in revenue due to the change?

That is correct, this is my estimate of the impact that the change will have on dYdX LPs and dYdX Trading Inc. Since the market maker rebate is effectively a lower trading fee for providing liquidity, a higher rebate means less fee revenue for dYdX Trading Inc.

[she.trades]

Thank you very much, @max-holloway, for your prompt and clear response.
We will support the proposal as soon as it is made public.

[max-holloway]

Given that there is mostly positive feedback on this proposal, I will put forward a snapshot vote for it on Monday, July 17.

[Jordi]

I have spoken to the relevant parties about the motivations behind this proposal. In private we have explained that this is likely a significant mistake and wanted to at least give a voice to this in public also. I understand there is general support to reduce LP rewards going into v4 and to try to get more top of book liquidity. Unfortunately this massive and abrupt change will have some predictable negative side effects and it is not a case of being equivalent between rebates and rewards.

In terms of the community saving $1mn- there could simply have been a repurchase of the tokens given out instead of increasing the rebate. Also to motivate top of book liquidity there can simply be a squaring of the ‘distance from mid’ term in the formula like some of the newer competitors have done to great effect.

Simply- The obvious thing that will happen is a hollowing out of alts markets liquidity. As reward quoting goes down, uptime and depth stop being incentivized and it goes for opportunistic top of book quoting during quiet times only. This can have a negative flywheel where traders get bad slippage and eventually seek their liquidity elsewhere, reducing liquidity further as there isnt enough volume to merit quoting and getting picked off occasionally.

There is a lot more to say about this but given this seems to have been predetermined to pass the vote the intent is mainly to give at least some public voice to the obvious deterioration in alts markets that will result. There are several better solutions we see and have discussed without rebuttal. Another one would be using governance to direct more LP rewards for some books instead of others based on factors the community cares about incentivizing. We also expect much more consolidation of volume into BTC and ETH now which is less a differentiator compared to other options.

We hope things work out well somehow and will continue to help discuss thoughts on these incentive plans going forward. Appreciate all the insightful and constructive talks we have had with the relevant counterparties despite the outcome being in what we feel might very well be in the wrong direction.

[cryptosdave]

Nice one. What are the most significant things to note

[jake]

My suggestion is: 1. DYDX V4 as an independent application chain should use $DYDX as the gas fee 2. Continue to reduce the emissions of transaction mining, and then use the agreement fee repurchase every month to destroy the transaction mining emissions of the month. Transaction mining has less and less impact on protocol growth, the community has had enough of $DYDX’s inflation and death spiral.

[jake]

I also hope that Antonio, as the founder, can listen to the voice of the community more, instead of posting some strange words on Twitter. There are no administrators in the discord communities.

[jake]

snx launched a decentralized perpetual contract platform to fully compete with dydx, and all proceeds are used to empower snx
What do you think?

[foxlabs]

Hey @Jordi, It’s good to see a variety of perspectives being expressed on this proposal, especially those highlighting potential drawbacks.

Some recent proposals have been hurried through without comprehensive risk assessments. This is a concern in light of the negative impacts observed following the reduction of trading rewards. However, the present proposal, with its applicability confined to v3, may serve as a constructive trial to guide decisions as we move towards v4.

As we transition, refining our governance procedures becomes an absolute necessity. A straightforward solution might be to set a specific timeframe for proposal discussions prior to them being voted on by validators. This approach would grant sufficient time for in-depth risk evaluations, exploration of potential adverse effects, and more profound community dialogues.

Moreover, adopting a council structure could be highly beneficial. A governance council could pave the way for more systematic decision-making, while a risk council would facilitate detailed risk assessments. Additionally, instituting a grants council is essential not only for ensuring automation and cost-efficiency but also for expanding the scope of decision-making processes.

[0xcchan]

Hey @Jordi , nice to meet you here. While I agree of the potential implications from a shallower alt liquidity, a key question we need to ask is : Does the exchange actually require that amount of depth to begin with?

I’ve previously shared a research from the grant recently completed - Research Post, and apart from markets such as SOL which sees occasional sizeable trades, it appears that the exchange doesnt need to constantly incentivise depths and can consider moving towards a more sustainable and mature LP structure, aligned with other CEXs. Alternatives can also include tightening the maxSpread and setting an absolute number as opposed to having relative squaring of distances wrt to mid.

I agree on having more directed LP rewards to certain markets, which I alluded to as a recommendation in the grant for tiering of markets and their respective allocation of rewards.

I look forward to any feedback and discuss on the research as well.