The Arbitrum Liquidity Mining program has shown notable success over the past six months, particularly in terms of Total Value Locked and volume market share increase. The program, aimed at spurring long-term growth, has impressively boosted TVL by over 500% and captured an 18% rise in market share from competitors. These achievements highlight the effectiveness of the program in enhancing liquidity and attracting trading activity. As we move forward, a key focus will be on observing how these metrics evolve as incentive allocations are gradually reduced or phased out. This next phase is critical for understanding the program’s long-term sustainability and the potential establishment of a self-sustaining flywheel effect, where pools maintain high liquidity and trading volume even without ongoing incentives.
Liquidity Mining is a pivotal strategy that protocols use to spur long-term growth. In June 2023, the Uniswap DAO voted for Gauntlet to optimize ARB liquidity mining rewards on Uniswap V3 pools on Arbitrum to achieve three goals: increase volume, capture market share, and create a self-sustaining flywheel effect. In addition to these critical metrics, experimenting with liquidity mining on Arbitrum allows the DAO to gather valuable insights into the trade-offs of liquidity mining to inform the strategies of future programs. To ensure the community is aligned on one definition, a flywheel effect would describe the following chain of events:
- Liquidity mining incentives introduced.
- LPs add liquidity to the pool, which improves execution quality for traders.
- Traders route more of their swaps through the pool and thus pay more fees to LPs.
- Fees from traders further incentivize liquidity in the pool until a stable equilibrium between liquidity and returns (fees + incentives) is reached.
- Liquidity mining incentives are removed, resulting in some liquidity being removed due to the lost incentives. However, the new equilibrium between liquidity and fees is higher than before since the new fees keep some new liquidity in the pool, and that new liquidity keeps the trading volume flowing.
Reflecting on the strategic approach of gradually reducing incentives to effectively incentivize pools before testing for a flywheel effect, the preliminary results have been notably promising. This measured and intentional progression has not yet completed a full cycle with the removal of incentives from most pools, but it has already demonstrated significant progress. Key performance indicators show an impressive increase of over 500% in Total Value Locked (TVL), an 18% rise in volume market share taken from competitors, and a substantial increase in liquidity across all participating pools.
One new pattern we are tracking is the revival of pools that previously had little to no activity. As the program progresses, we are keen to further understand the endurance of these pools post-incentive removal. For our next cycle, we are removing incentives for the BTC.B pool and reducing others. By closely monitoring the ‘stickiness’ of incentives and the behavior of LPs in response to these changes, we aim to gather critical data. This data will enhance our understanding of LP elasticity and the robustness of the flywheel effect.
The metrics described in this section can be found in Gauntlet’s public Arbitrum Liquidity Mining dashboard.
In evaluating the success and efficiency of the program, it’s essential to delve into the KPIs that provide us with quantifiable measures of the program’s impact. TVL, volume market share, and LP fees are indicators of user responsiveness to incentives. They are crucial in determining if the requisite conditions are met for a flywheel to take effect. A consistent and significant improvement across these KPIs, particularly after the reduction or removal of incentives, would suggest that a flywheel effect is in motion. This effect would be characterized by a virtuous cycle where initial incentives lead to increased liquidity and trading activity, attracting more users and investment, thereby sustaining and potentially enhancing pool performance even without continued incentives.
Let’s break down each of these KPIs to understand their significance and the observations we’ve made so far. The numbers below describe comparisons to periods 30 days prior to first incentives on a per pool basis:
Total value locked: Since the start of the Arbitrum LM program, we’ve recorded increases to volume-weighted average daily TVL of 516% (compared to overall Uniswap TVL growth on Arbitrum was 64% during the same period). Increases have been observed for every pool, even pools previously cut due to lack of growth (USDT and RDPX) continue to see elevated levels of TVL despite incentives not being present for many months. For the currently incentivized pools, the greatest increase in TVL happened during the first cycle of incentives for each pool: subsequent rounds of incentives did increase TVL for most pools but not to the same magnitude. However, it’s important to note that while TVL is a vital metric for gauging LP participation, it doesn’t fully encapsulate the program’s success in creating a self-sustaining ecosystem. For a more holistic view, we also consider volume market share.
Volume market share: This metric is fundamental as it reflects the competitiveness of the pools within the broader market. Overall, the program has increased volume market share by approximately 18% (weighted by pool volume). In contrast, Uniswap’s overall volume market share on Arbitrum has remained stagnant (approximately down 1-2%). This growth is especially noteworthy in previously inactive pools (deadpools) that have been bootstrapped due to incentives. The remainder of the pools had varied improvements to market share: some have seen steady growth (MAGIC/ETH), while others have not seen notable changes (ETH/rETH). Volume market share is a key indicator of the program’s ability to attract and retain trading activity, especially in contrast to competing DEXes.
LP Fees: The increase in LP fees is a direct financial benefit to liquidity providers and a measure of the pools’ profitability. During the incentive phases, there has been a notable increase in LP fee generation from the pools involved. So far, the incentive initiative has resulted in an extra $596,000 in LP fees. This is approximately a 5x growth in average daily fees earned. However, this is in-line with the overall fees earned on Uniswap Arbitrum, highlighting this metric’s vulnerability to market conditions.
With the gradual withdrawal of incentives a decline in key performance indicators is expected. However, the goal is for the pools to stabilize at a higher baseline of volume and fees than pre-incentive levels, maintaining profitability and attractiveness to LPs without ongoing incentive expenditure.
In summary, these KPIs are critical in understanding the multifaceted impact of the LM program. Results so far are encouraging and we’re eager to monitor LP responsiveness in the coming months as we continue to adjust incentives.
Key Pool Highlights
Overperforming pool: MAGIC/ETH pool was especially successful, seeing volume and fee market share more than double in comparison to pre-incentive periods. TVL, volume and liquidity all are seeing dramatic and steady growth as well. To date, approximately $145K worth of incentives have been disbursed to this pool, yielding an additional $433K in LP fees, as well as sizable increases to volume and TVL. Results below:
Underperforming pool: USDC/USDT 0.01% pool did not fare as well. While TVL and liquidity both saw impressive increases, volume did not. Volume is essential to fueling a flywheel effect. Seeing a low reaction in volume to these incentives discouraged further incentive spend into this pool if the likelihood of a self-sustaining flywheel appearing would be low. Given the minimal growth in response to incentives, we opted to remove this pool early.
Unique finding: LM seems especially suited for stagnant or “dead” pools, almost instantly seeing surges of volume, TVL, and liquidity in these pools. For instance, BTC.B/ETH 0.3%, which, prior to incentives, had no activity or liquidity. Over the course of the incentive campaign, this pool reached a peak TVL of $1.68M and generated $97K worth of LP fees.
So far we’ve learned that liquidity mining is a potent tool when seeking to bootstrap liquidity and gain TVL very quickly, oftentimes resulting in sizable gains in market share and boosts in LP fee revenue. However, the sustainability of these elevated metrics is unclear. As we begin to remove incentives we will gain a better understanding of the speed at which TVL, volume and fees decay and where the new baseline for these metrics settle at. Only after a period of incentive cessation for responsive pools will we know whether a flywheel effect has taken place.
Insights + Patterns for LM
In this section we extract additional insights into pool responsiveness to incentives and greatest opportunity for market share gains using our swap simulation tooling.
How much volume is actually routed efficiently for specific pairs?
The main intent of liquidity mining is to increase liquidity through incentives, aiming to attract additional volume to these pools due to improved price execution. This increase in volume is expected to generate more fees for LPs, which in turn would attract further liquidity creative a positive feedback loop. However, it’s important to assess the current state of price execution in these pools. Specifically, we need to consider whether there is a soft limit beyond which increases in liquidity do not yield meaningful gains in price execution. Understanding this will help identify where the most significant opportunities for improvement exist.
Using our fee simulation tooling, we examined a handful of pools that were part of our original recommendations when the program launched in September and compared counterfactual swap destinations to examine what percent of trades were routed efficiently. The results can be viewed in the table above, describing on average what proportion of transactions for a given pair are executed at the most optimal destination. For example, most rDPX/ETH swaps tend to be routed optimally, where Uniswap only accounts for 0.5% of that optimally routed volume. With 92% of volume being routed optimally for rDPX/ETH, it suggests traders either may be sensitive to price execution for this pair and/or that aggregators are routing very efficiently to the best DEX destination. On the other hand, wstETH/ETH show low overall routing efficiently; possibly retailer traders eagerly swapping to liquid staking token of ETH without any DEX preferences or consideration for price execution.
How would routing efficiency change if incentivized pools had more liquidity?
Of the optimal transactions, a vast majority of these transactions had executed on competing platforms during periods prior to any incentives being introduced. Next, we want to see how much of that optimal volume on competitors has the potential to be re-routed if only our incentivized pools had better price execution (from increased liquidity).
We conducted an analysis of simulated swaps from September 1, 2023, to September 14, 2023, a timeframe preceding the initiation of any liquidity mining campaigns. During this analysis, we focused on pools earmarked for incentivization, significantly augmenting their liquidity. This augmentation was in direct proportion to the liquidity levels observed during periods when liquidity mining was actively underway.
According to simulation results, liquidity increases to levels similar to those observed during incentives would’ve notably increased Uniswap’s price execution and consequently volume.
A point to consider is the limited scope of the data: analyzing only a two-week period prior may not provide a comprehensive enough view for drawing generalizable conclusions. Nevertheless, the focus was on examining dates immediately preceding the liquidity mining campaign, with the rationale being to ensure the findings are more relevant to the current market environment.
How much volume from competitors could we theoretically re-route?
We began experimenting with different liquidity counterfactuals to understand where diminishing returns start to become evident. This is particularly important in cases where incentivized pools are in direct competition with pools that offer a lower LP fee, suggesting that for most trades even in extreme liquidity scenarios the incentivized pool would not have better price execution than competing pools with lesser fees, even though they have less liquidity.
To understand how much volume was up for grabs, we limited our focus on “re-routable volume”, that is, volume that was optimally routed to competitors that incentivized pools could potentially steal if they reach sufficient liquidity. The results are visualized below, where the x-axis represents the liquidity increase scenario (e.g., multiply historical liquidity by 10x) and the y-axis represents proportion of volume that could be optimally re-routed to incentivized pools from optimal competitor volume.
The findings reveal at what liquidity multiplier diminishing returns start to occur and the presence of plateaus in re-routable volume. As alluded to earlier, the LP fees may be the key distinction. The rDPX (1%), rETH (0.05%) and wstETH (0.01%) pools in particular have higher fees than their primary competitors (0.3%, 0.02% and 0.008% respectively). These lower LP fees create a soft limit for which liquidity increases can result in sufficient price execution improvements to re-route trades away from competitors.
With MAGIC/ETH’s case, fee competitiveness is less of an issue as the incentivized pool’s LP fee is par with the pool’s primary competitor. The main gains are a result of pre-existing price execution inefficiencies among these pools (only 73% of volume is routed to the best destination, where that number is typically 80-90% for other pools, except wstETH which is the worst at 44%, which also explains wstETH’s stark re-routing percentages).
We plan to further investigate the feasibility of incorporating these simulations into our pool recommendation methodology.
Importance to the future of liquidity mining
The insights drawn from this analysis are crucial in refining our future liquidity mining recommendations, and additionally highlighting the value of integrating swap simulation tooling we’ve previously developed. By understanding the dynamics of volume routing and impact of increased liquidity, we can make more informed decisions about which pools to incentivize. This approach would not only optimize allocation of incentive spend but also enhances the overall efficiency of future liquidity mining programs.
As alluded to the previous section, we are aiming to re-apply some of the swap simulation tooling developed for other analyses to our pool recommendations. Understanding which pools are able to capture the most volume from competitors by exceeding price execution could prove especially valuable as the incentivized pool would become “the best” swap destination for a given pair, motivating traders to migrate their trades for the sake of price execution efficiency. This additional volume could spur the flywheel effect sooner and result in less incentive spend required to achieve market share dominance.
However, the sample size of candidate pools is already small (at least, in Uniswap Arbitrum’s case) and significant optimizations to pool selection criteria may not be overly performant over simply targeting pools with weak market share and liquidity.
Conclusion and Next Steps for H2 of Arbitrum LM
As we approach the second half of the Arbitrum LM initiative, it’s an opportune moment to reflect on the program’s substantial achievements and outline our strategy for its ongoing development. The program has successfully met several critical benchmarks, with noticeable increases in TVL, volume market share, and LP fee generation. These achievements underscore the program’s effectiveness in enhancing liquidity and trader engagement across various pools.
Looking forward, our strategy will shift towards a combination of gradual reduction as well as more optimized choices of incentive allocations. This phase is necessary for acquiring insightful data on the persistence and adaptability of LPs in response to changing incentives. This data will be used in assessing the program’s long-term viability and in fine-tuning our strategies to maximize efficiency and impact.
In terms of methodological advancements, we are exploring the integration of our sophisticated suite of simulation tools from prior analyses. This initiative aims to identify pools with potential for increased volume capture and to determine the required boosts in liquidity needed to achieve dominance in specific trading pairs. The feasibility and implementation of this enhanced methodology are currently under thorough examination.
The upcoming period will be crucial in evaluating the true efficacy of liquidity mining as a catalyst for sustained trading and LP activity. It will provide us with a clearer understanding of the program’s overall success. Our focus will be on maintaining the momentum gained while strategically managing incentive distributions to ensure long-term stability and growth of the pools.
In conclusion, as we transition into the latter half of the program, our objective remains to solidify Uniswap Arbitrum’s position as a leading platform for liquidity and trading activity, leveraging the insights and successes of the past months to guide our future endeavors.