Information about STIP/STIP Backfund
- Can you provide a link to your previous STIP proposal (round 1 or backfund)?
- How much, in the previous STIP proposal, did you request in ARB?
4.5M ARB
- What date did you start the incentive program and what date did it end?
Start date: December 29th 2023 00:00 UTC
End date: March 29th 2024 00:00 UTC
- Could you provide the links to the bi-weekly STIP performance reports and Openblocks Dashboard?
Bi-weekly STIP reports:
5th of January 2024
19th of January 2024
2nd of February 2024
16th of February 2024
1st of March 2024
15th of March 2024
29th of March 2024
Openblocks Dashboard
- Could you provide the KPI(s) that you deem relevant for your protocol, both in absolute terms and percentage change, month over month, for the first of each month starting from October 2023 until April 2024, including the extremes? If you don’t know what KPI might be relevant for you or how to properly define them, please refer to the following document:[Arbitrum DAO] OpenBlock Labs Incentive Onboarding Spec
Considering our retrospective is relatively extensive - and following our advisor’s feedback - we’re aiming to improve the readability of our proposal by including a TL;DR. Details will follow directly after:
Throughout the STIP-period, Gains Network witnessed substantial growth across various metrics, including daily trading volume, trade fees, number of traders, and number of trades. Notably, daily trading volume and trade fees experienced over 200% growth, while the number of traders increased by 66.67%. However, gTrade fell short of meeting the expected KPI for the number of traders, attributed partly to relatively high gas costs and minimum position size requirements for the majority of the campaign. Nevertheless, technical improvements such as gas cost reductions [EIP-4844] and (upcoming) updates [gTrade Credits, v8, and v9] aim to address these challenges and allow us to attract a more diverse trader base. Additionally, gTrade’s proactive approach in pair offerings and the introduction of multi-collateral trading signify our commitment to enhancing user experience and fostering growth within the Arbitrum ecosystem.
Sources: https://dune.com/gains/gtrade-stats-arbitrum-stip,
https://dune.com/gains/gtrade_stats
Throughout the STIP, 6,573 unique traders (addresses) conducted trades on gTrade, collectively resulting in a total of 214,492 trades. On average, this equates to approximately 33 trades per address. Including leverage, these trades amounted to a total volume of $12.86B. The fees paid for the settlement of these trades contributed to a protocol-wide (Arbitrum based) 3-month revenue of $8,541,646 which is shared among various stakeholders of the network. For example, with our LPs receiving 18.75% of all fees, this represents $1,601,558 in real yield for DAI, wETH, and USDC depositors.
The campaign also significantly contributed to many outliers for our protocol milestones. For example, an ATH for daily volume of $521.31M was achieved during the STIP. To contextualize the significance of this milestone it’s important to note that the protocol witnessed only 2 days of 200M+ and 0 days of 300M+ Arbitrum-based volume throughout its entire pre-STIP existence.
Generally, it’s imperative to note that the STIP has significantly bolstered participation within the Arbitrum ecosystem, consequently driving up sequencer fees. Metrics like ‘number of daily trades’ show that gTrade has been able to contribute in this regard. The above-mentioned growth trajectories translate into a more generic figure that covers all traffic the gTrade ecosystem has generated. For this, we analyzed the total gas spent by all stakeholders, which includes oracle requests, traders’ txs, depth updaters, LP commits, etc.
MoM Analysis
Comparing to pre-defined KPIs
Although we’ve mentioned numerous essential metrics above, we also proposed specific KPI goals in our STIP-application. Reviewing our initial STIP proposal, these pre-defined KPIs were established for a budget of 7M ARB. It is important to note that, in response to feedback from Abritrum delegates and the community, we lowered our ask to 4.5M which was the final amount granted. This represents a 35.7% reduction in the available budget and consequently necessitates a corresponding adjustment to the KPI growth expectations. The following numbers apply:
While we’ve seen a substantial increase in volume and the number of trades, it’s apparent that we fell short of meeting the expected KPI for the number of traders. This suggests that, during the STIP, gTrade attracted relatively large traders that trade more (often). The cause for this trend may stem from the fact that gas costs have remained relatively high for the majority of the campaign. And, to offset infrastructure costs associated with a fully on-chain environment, gTrade had to maintain a minimum position size of $15,000 for the majority of the period. However, with the implementation of EIP-4844 by Arbitrum, gas costs have been significantly reduced, allowing us to lower the required minimum position size by 66% (i.e. $5,000). We’re confident in our ability to attract traders of all sorts with our planned future efforts and diversify the traderbase.
Other than that, gTrade is proud to have significantly outperformed its pre-defined KPIs and the team is excited to beat these numbers even more with future campaigns in the Arbitrum ecosystem. And, with all the learnings from the STIP, we’re confident in our ability to incentivize trading activity and liquidity provisioning even more efficiently.
Technical improvements during STIP
Multi-collateral
Upon frequent request, gTrade introduced multi-collateral trading during the STIP-period on January 23rd to allow for USDC and WETH as new collateral types. Previously, DAI had been the sole option since the deployment on Arbitrum. Data indicates a strong positive response from Arbitrum’s perpetual futures traders. In the first week of April, USDC already accounted for more than 31% of the protocol’s volume, while WETH was also experiencing substantial growth and averaged over 11% of volume in the same week.
Pair offering
At the time of writing (April 11th), gTrade boasts a total of 166 available pairs. At the start of STIP, this figure stood at 117, translating to a new pair listing approximately every two days. This growth underscores our proactive listing strategy, aimed at accommodating the diverse needs of perpetual futures traders. We believe a wide offering of assets with high Open Interests available is a pivotal factor in retaining Arbitrum-native traders within the ecosystem amidst a highly competitive industry.
Minimum Position Size Reduction:
Thanks to the EIP-4844 upgrade, gTrade’s has witnessed a significant reduction in gas costs. This development has enabled us to slash minimum position sizes by 66% across the board on Arbitrum, marking a significant milestone in our mission to welcome traders of varying profiles and sizes.
Technical improvements towards the STIP-bridge
Assuming STIP-bridge funds would be accessible somewhere mid-May, we’re excited to share a perspective on the protocol’s direction prior to this event. We firmly believe the rollout of our updates over the coming two months will bring the platform as close to a CEX-experience as possible and even improve on that experience on some fronts.
V8
As of April 18th, the 2-week timelocks for our v8 update have been initiated; meaning the deployment will be live once the backend & frontend are ready. It’s an important update that has been worked on for months and encompasses numerous improvements from a protocol composability- and UX/UI perspective:
Smart Contract Trading
With v8, Gains Network’s contracts are now fully open for Smart Contract Trading, allowing for integrations across various domains where our infrastructure can be leveraged to build products and strategies on top of our wide pair offering and deep liquidity. We’re actively engaging with Arbitrum-native parties to establish long-term partnerships.
Composability
Moreover, v8 significantly improves the composability of the gTrade platform. For the update, the entire codebase has been audited and fully documented with natspec. And, we’ve implemented diamond- and solid patterns, which allows for faster updates and reusability of any feature. For example, we’re able to list new collateral types with increased speed and efficiency; readying us to be adaptable in a fast-changing environment.
UX enhancements
Alongside the above-mentioned composability improvements, the update introduces various trading enhancements. These include, but are not limited to:
- Unlimited collateral size as far OI allows (previously max 250k);
- Unlimited amount of open trades per pair for any wallet (previously max 3);
- 20-30% gas reduction on all protocol-interactions;
- Single 1-Click Trading delegation for all collaterals (previously separate wallets per collateral).
V9
Over the past months, our team has also been working on the next update in parallel, set to go live by the end of May: v9. This update will finally introduce a long-awaited feature called “Partials”, enabling traders to scale into positions and partially take profits/losses. While recognizing this is a critical feature for a large cohort of traders, our infrastructure and codebase needed modifications for the technical implementation. Now, with v8 in place and our entire codebase being refactored, the team is able to push similar UX-improvements exponentially faster.
gTrade Credits
In an attempt to retain as many traders as possible post-STIP, Gains Network introduced gTrade Credits; a new fee discount loyalty program designed to save more on fees as you trade. Throughout the last month of the STIP-program, traders were already accumulating credits to enjoy discounts as soon as ARB rebates stopped. This system is developed to become an integral part of our trading platform and will be active throughout the STIP-Bridge period as well. These native platform discounts leave room for us to spend ARB rewards on other activities, enlarging the impact of the grant for other purposes.
Onboarding
With the start of the Bridge campaign, gTrade will have a new onboarding feature integrated through a partnership with Li.Fi. This will simplify the process for new traders to bridge & deposit collaterals and gas funds from other ecosystems to start trading on our platform. Coinciding with a targeted marketing campaign and gamification aspects, we aim to onboard a number of new traders that should stay on Arbitrum after experiencing everything the ecosystem has to offer.
- [Optional] Any lessons learned from the previous STIP round?
For the same reasons as with Point 5, we’re including a TL;DR here followed by an in-depth analysis.
For the STIP-campaign, Gains Network implemented innovative incentive streams through a points system, rewarding traders for different behaviors: Fees Paid, Absolute PnL, Loyalty, and Relative PnL. Rewards were paid out per week (i.e. epoch) where various combinations of allocations were introduced for each category.
While this contributed to the gamification aspects of the protocol, it also attracted sybil attempts; particularly in the Relative PnL category. Actors tried to game the system through delta-neutral positions in an attempt to extract $ARB from the reward pools. Efforts have been made to mitigate this adequately, including the removal of the category. Albeit of less significance, similar attempts were observed in the other categories, prompting adjustments such as trade frequency requirements and enhanced point-accumulation criteria. These learnings are of important value to iterate on the designs of our points system for future campaigns.
Lastly, analyses for category-effectiveness and its allocations have been conducted. Findings indicate that the points system has effectively contributed to the expansion of gTrade’s market share while recognizing that various other factors could be of influence. Continuous efforts are focused on refining reward mechanisms and enhancing research practices.
Learnings on Sybils
At Gains Network we spearheaded innovative ways of incentives through various categories of trading gamification, leveraging our points system to track and reward our traders. The following behavior categories were rewarded: 1) Fees paid, 2) Absolute PnL, 3) Loyalty, and 4) Relative PnL (PnL%).
We encourage you to read our Medium post that served as an introduction to our STIP campaigns and highlights the mechanics of each category. To start off, gTrade introduced eligibility for ARB rewards in all categories at the same time. While this provided traders with ample tools to qualify for rewards, it unfortunately also attracted sybilors attempting to game the system. While these sybilors were mostly unprofitable, the ARB rewards claimed by those addresses were a net-negative to organic traders. On a positive note, this provided us with important learnings for the rest of the STIP-period and solutions for future campaigns which we’ll outline below.
Relative PnL (PnL%) learnings
The most significant insights were derived from the Relative PnL category where the majority of inorganic farming attempts were made. Towards the end of each epoch, as it became clear how much ARB was to-be-distributed among all eligible addresses, sybilors tried to game this category by opening low-collateral, very high-leverage, ‘delta neutral’ positions on multiple addresses. The rationale here is that at least one side of a short-term position should perform well, even if the other side gets liquidated. Before the STIP, many actors have preceded them with similar attempts but due to our risk management practices and the inherent volatility of the market, this strategy tends to yield negative expected value on average. However, with ARB incentives in place, and assuming the profitable side of the position yields a high % PnL, the addresses would be positioned somewhere in the leaderboard and become eligible for a chunk of rewards in this category. And, with these addresses becoming consecutively eligible for other categories, their potential losses on the delta-neutral positions would be compensated by the rewards for their fees and “loyalty”. The attractiveness of these extra layers of potential profits enticed one or more actors to execute this strategy on a relatively huge number of addresses. While this seemingly led to a steep growth in our active users, it must be clarified that this was partly inorganic.
At the peak, ‘minimum position size orders’ (i.e. low collateral) brought gTrade only 1% of the total volume, while they accounted for half the bot triggers and half of total transactions:
In an effort to mitigate this, we introduced an eligibility requirement where addresses must trade more than 2 days and execute at least 5 trades within the epoch. While it seemed to be effective, bots were still farming the campaign and there was uncertainty on whether they would be profitable by the end of the epoch. So, instead of waiting, we moved over to the next solution by removing the Relative PnL category which caused the number of fake traders to finally decrease.
Even though the overall overall PnL for these actors was negative, the ARB they have managed to extract from the campaign before we addressed the issues would have been better utilized on organic behavior. For future campaigns, it’s clear that we need to integrate attestation solutions (e.g. Proof-of-Humanity, zk-KYC) for us to reimplement this type of gamification category. We’re actively exploring suitable solutions that are minimally intrusive while ensuring sybilors are kept at bay.
Absolute PnL learnings
For this category, sybilors tried to execute on a similar strategy as we’ve outlined above for the PnL% category. They would open delta neutral positions on various accounts and become eligible for ARB rewards on the winning account; potentially offsetting the fees paid on both addresses. Again, in combination with rewards in the other categories, this extra layer of potential profits enticed some actors to try and game the system. The number of attempts here was not significant as this strategy required more capital and naturally yielded lower payouts as it’s harder to beat the performance of top traders. Moreover, we were quite fast to identify this behavior and imposed a required minimum of 3 trades executed during 2 different days to be eligible for ARB rewards in this category.
Loyalty learnings
At the start of the campaign, we designed the eligibility criteria such that traders would accumulate points based on the # of days they traded within the epoch. While this design functions effectively when considered in isolation, it also served as an extra incentive to farm the Absolute PnL rewards as we’ve outlined above. Hence, we implemented an enhancement that eliminated the potential issues, such that traders also accumulate more points as they trade on consecutive days.
Fees paid learnings
In isolation, this category isn’t farmable if you don’t incentivize beyond 100% of fees paid. However, it lowers the cost basis to farm other categories which is why it’s of critical importance to consider all incentive streams and how they are interconnected.
Learnings on category effectiveness
One of the purposes of the points system was to increase our ability to experiment and learn. Therefore we’d like to share one high-level analysis regarding the effectiveness and impact of each category. In an attempt to identify the relative strength for each category (or a combination of multiple), we’re comparing gTrade to peer-protocols in the same vertical: PerpDexes. To account for market volatility, which naturally contributes to volumes on perpdexes, we normalized the impact for each epoch on gTrade’s volumes by comparing it to these protocols’ numbers; specifically for fees.
At times, we tweaked the distribution per behavior where one or more categories were incentivized more significantly. The following applies:
As we deprecated the Relative PnL category for Epoch 8 we continued to incentivize as many categories as possible for that week. Even though Distribution II incentivizes the same categories as in Distribution IV, the proportions are different. Additionally, since it only covers one week of data, we exclude this from our analysis as it provides too little information to draw any conclusions. Consequently, our conclusions will be based on Distribution types I, III, and IV.
For conciseness and to protect proprietary data, we’re outlining a comparison with just one peer protocol here.
Distribution I - 4 categories
For these weeks [7 in total] we saw a market share of 40.22% on average.
The maximum market share was realized in Epoch 6, with 69.61%. It’s important to note that this was NOT the Epoch where we’ve experienced the sybil attacks on the Relative PnL category.
Distribution III - Fees only
For these weeks [3 in total] we saw a market share of 38.54% on average.
Distribution IV - 3 categories
For these weeks [2 in total] we saw a market share of 28.9% on average.
While I and II resulted in similar market share percentages for this specific peer comparison, the numbers suggest that our points system, along with its gamification aspects and diverse set of incentives, has proven to be a successful strategy to grow gTrade’s market share (in terms of fees). We’ve seen similar results for the multi-faceted peer analysis and are thus motivated to continue refining our rewards mechanisms with specific adjustments in place.
And, while analyses like these are of significant value, we recognize that there are other factors that influence such metrics. We are committed to enhancing our in-house research practices to ensure the continuous and efficient allocation of incentive budgets.
New Plans for STIP Bridge
- How much are you requesting for this STIP Bridge proposal?
Gains Network is requesting 2.25 million ARB tokens for the STIP-bridge. This allocation will facilitate the sustained expansion of the protocol, catering to Arbitrum’s extensive community of perpetual futures traders.
- Do you plan to use the incentives in the same ways* as highlighted in Section 3 of the STIP proposal? [Y/N]
No
- [Only if answered “no” to the previous question] How will the incentive distribution change in terms of mechanisms and products?
For the STIP campaign we proposed and acted on the following breakdown: 85% to trading incentives and 15% to LP incentives.
For the STIP bridge addendum, we’re opting for the following breakdown of funds:
Trading incentives:
For trading incentives, we propose to continue with our points system and the incentivization of Loyalty, Absolute PnL (i.e. trading competitions) and Fee rebates. After the implementation of attestation solutions that fit our platform, there’s a chance of us re-introducing the PnL% category.
Furthermore, we see peer protocols align on a 75% max fee rebate. We will adhere to this amount.
Liquidity incentives:
Our original proposal outlined the following for liquidity incentives:
New proposal for liquidity incentives:
During the STIP, Gains Network successfully launched its gUSDC & gETH vaults. The collateral type of choice for many traders became gUSDC. We are opting to further entrench the gUSDC vault into the arbitrum ecosystem as gDAI currently enjoys this to a larger extent.
We are also opting to move our protocol owned liquidity to Camelot and work together with arbitrums leading decentralized spot exchange to deepen the liquidity on relevant markets.
We are opting for camelot to show our commitment for collaboration with Arbitrum native projects.
Hence, the new breakdown will be:
New category introduction: Partner projects
Allocation: 112,500 ARB (5%)
As mentioned in “Technical Improvements towards STIP-bridge”; our v8 Smart contract trading update will be released out of timelocks on the 3th of May. This update will significantly simplify the ability for third parties to integrate the gTrade smart contracts and expand their native product offering. In order to incentivize third parties to build on our network we will commit 5% of our STIP bridge addendum to these parties bolstering our commitment on supporting the wider Arbitrum ecosystem.
We’re introducing a forum where parties can apply for parts of these incentives. A committee will review all applicants and allocate funds accordingly.
We invite all types of possible integrators to apply. Some examples of potential integrators are:
Copy trading platforms, Aggregators, SocialFi platforms, Terminals, and alike.
- Could you provide the addresses involved in the STIP Bridge initiative (multisig to receive funds, contracts for distribution, and any other relevant contract involved), and highlight if they changed compared to the previous STIP proposal?
Funding Address: 0xc5fCA2c19c5Ca269a10e15ee4A800ed82F53787D
- Could you share any feedback or suggestions on what could be improved in future incentive programs, what were the pain points and what was your general evaluation of the experience?
We believe that the governance process set forth for the campaign is well-structured and provides ample opportunity for protocols to showcase their creativity and innovation. However, reflecting on the first round of the campaign, we feel that there could have been a budget for marketing efforts as the STIP campaign relied on its participants for marketing. We feel that this might not be sufficient for the size of the campaign.
We suggest that, for future campaigns, the DAO could consider allocating a separate budget specifically for marketing initiatives for projects to apply for. This would empower each applicant to more effectively promote their projects to a broader audience, ultimately enhancing participation and engagement. While there may be logistical challenges in coordinating such a budgeting approach, we believe it holds potential to amplify the impact and reach of future campaigns.
Additionally, we suggest future campaigns could foster collaboration among peer protocols by allocating incentive programs to specific verticals. This approach, akin to the Gaming Catalyst Program and Fund, would encourage collective innovation within targeted sectors.