Inefficient Emissions and Incentive Optimization | Jake, Gauntlet | ETHTaipei 2023

subject for today is incentives also known as token emissions and how we can use data science to drive better growth strategies for D5 protocols defined symptoms are a foundational tool for all businesses and especially for protocol growth and this is a tool that should not be traded lightly my chat today aims to show how quantitative techniques use heavily in traditional finance and high frequency trading are often underutilized furthermore kotlin has seen them leverage while working with some of the biggest names in D5 to drive significant impact over the past year [Music] here's the high level plan to achieve that goal first I'll look back at how Dows typically spend money and how efficiency is hard to track and optimization is even harder second is to review the current state as to what we have seen from our strong signals working with our early adopters to show how adjusting strategies Based on data Sirens can drive significant impact lastly I'll cover what we're looking forward to in the near future [Music] first I'll introduce myself hi everybody my name is Jake and I'm currently a protocol manager at Gauntlet which is a very much a new title um before working in defying crypto I worked at as a growth analyst and a sales engineer in traditional Finance corporate finance and Tech startups like most of us who work in crypto I wear multiple hats as a protocol manager at Gauntlet I function as a Swiss army knife of a role this role includes sales engineering client management product marketing but most of the time I'm focused on solution fitting by working with our D5 partners and our translating the complex analysis coming from our research and data science teams into on-chain recommendations for those who don't know Gauntlet we service a financial modeling platform that uses ballot testing techniques from the algorithmic trading industry to inform on-chain protocol management we run hundreds of thousands of simulations a week simulating Market outcomes user behaviors and asset behaviors to help protocols combat the industry turbulence we currently work with some of the major players in D5 including Ave maker compound uniswap mutable Venus moonwell and others [Music] first it's important to line on how Dows are just a new iteration of business teams or corporations of course we're all here to push for decentralization and Innovation but it's also helpful to line up how the old framework aligns with the new to ensure that we're not starting from scratch and recreating the wheel most Dows exist to provide a service that benefits D5 while balancing maximizing profit shares of the new tokens terms are now baked into smart contracts luckily shareholder meetings are now forums or Twitter spaces Etc regardless thousand corporations alike must constantly review the revenues and costs and this context let's see how major players are spending on the cost of emissions or defined incentives specifically first we'll look at Curve finance a very popular decks when looking at Curve over the past few years you can see that they've spent around 3.5 billion dollars in total emission spending huge number if you look at the kind of the breakdown between their total economic emissions distribution around 50 of that is in liquidity mining or around 1.8 billion now switching over to a lending protocol like compound Finance we see that they have spent around 156 million dollars on token emissions spring liquidity on their lending protocol looking at their token distribution around 40 of this was spent on specifically liquidity mining incentives similar to zoma another particle that shaves on derivatives is dydx which is spent over 2.4 billion dollars in token emission spending and just on liquidity mining incentives around 50 of that around 1.2 billion simply put this is a huge huge balance sheet expense these numbers are not likely to go any down anywhere we're going to continue to see protocols spend a lot of money incentivizing liquidity on their platforms liquidity mining incentives are a complex subject but let's define them so we're on the same page about what they're defined as D5 protocols rely on liquidity being stored in the forms of pools to Foster usage and trading Protocols are paying to attract funds and users to Spur growth this is a common practice in traffic as well as people are incentivizing market makers luckily this is an optimization problem and our team at Gauntlet loves optimization problems and is Illustrated as the above the approach has a few assumptions one being that protocols have an objective usually to maximize profit to some degree the second being that they have a budget in mind this is a number that they're willing to spend on token emissions this is likely to change but this is a number that we can assume and the last that this budget is going to be distributed amongst X number of pools and this is also a number that can change written mathematically in the function below you can see that this is a function of expected profits subject to a total budget of individual pool budgets so summing up individual pools into total budget as a budget as a pro function of expected profit I'm now going to go into an example with a common Dex here are two pools on the Dex Trader Joe it's really interesting because these pools specifically show two sides of the spectrum the stable coin pool between usdt and usdc is operating at a loss shown here you can see that's losing 57 57 cents of Revenue per dollar in incentive spent so for every dollar that Trader Joe is using on token emissions this pool is losing 57 cents in Revenue which could be okay temporarily but it's an interesting data point on the other hand the usdc avax pool is operating in the green generating 30 cents in revenue of fees per dollar spent in revenue and again these strategies can change some pools are growth pools some pools are blue chip so it's okay these numbers are not the same they shouldn't always be green but it's very important to know these data points if I'm in the Trader Joe Dao it's extremely helpful to know what pools are performing well which Farms we should be putting more money into and how these things change over time so as a Dell member I need to focus on so many things I need to focus on feature development Community Management governance Partnerships investors user Innovation Market risk if vitalik is speaking at a conference if Elon is tweeting about Doge the list goes on so how do I solve the specific problem of rewards and budget optimization unfortunately and luckily this this number of teams is going down but a lot of teams go by gut feel which we strongly do not recommend most protocols use a form of quantitative heuristics which is usually using internal dashboards Excel spreadsheets which is a great start but as the companies are no longer Flying Blind but there's obviously a lot of room for growth solution you could possibly do is Outsource your your talents or insource your talents to a Solutions team to get a team that lives and breathes the data day by day like our Gauntlet applied research team which has officially born this quarter and is now working with big protocols like uniswap and immutable immutable being our first gaming protocol client and uniswap of course being a major Dex feel free to take a picture of our model architecture I'm not going to go in too much into the weeds to kind of go through at a high level if you look on the left side you can see that uh the flow of our operations we start with ingesting on-chain pool performance marking conditions and off-chain analytics this list can fed into our database and this is processed by our custom incentive and liquidity models and a service to Dows in the forms of ongoing reports recommendations to fine-tune parameters or budget optimizations budget adjustments themselves so that's how much you're spending on actual token emissions and then live dashboards to provide the community and core teams with live ongoing performance tracking to make sure that you're actually achieving your goals and if you're not how you can adjust otherwise what does this look like in collaboration really here's our approach the first thing we do is spend a lot of time with teams and communities to understand their pain and their visions of success which is very different for every protocol depending on their growth Cycles here are some questions that I would very much encourage you guys to ask your teams your communities and your Dows that you should that you should really like flesh out the big ones being what does success look like from a metrics perspective so how can you quantitatively track your success and how do you track and optimize for that a lot of questions go over here these are just the tip of the iceberg I'll go over now our approach or our solution so how do we actually service and relieve pain in these problems one thing we can do is we provide incentive program design specifically to make sure people are understanding that your mechanism is optimized for your protocol's growth we also do value distribution analysis and provide dashboards correlating to that to specifically understand that the metrics you guys are tracking are tracking towards the goals and the protocol growth numbers you look to achieve we also do user response modeling and ongoing optimization so as you guys are changing the incentive budget as you're in changing which incentives are going where you're understanding how effective those impacts are there's a lot of things we don't cover we leave the pros like zoma to think about Oracle pricing and pricing for options we don't do code audits but these are some of the things that we're interested in for a beaded out perspective beta dial is obviously a very new industry um so it'd be a curious next question would be like how is this going so far current solution is that so far so good here's a graph from one of our Partnerships that shows a five months of impact and how after working with our applied research team we saw a huge increase in efficiency this graph shows how volume is generated by incentive spent with this team we observe that while TVO is an important metric profitability and pool elasticity need to be really analyzed further and track deeply we were able to identify a portion of pools that were not returning a high investment for the protocol and guide incentives to pools that actually responded to the investment this is akin to Farmers who need to stop over watering one section of their crops and apply fertilizer and water to a different one this was especially helpful as an increasing profitability is super crucial during Market shrinkage times which you've seen over the past year so that this was the graph of the past example another example showing a little bit more of a recent data points so in Q4 of last year we were not working with this protocol in between in December we had some data gaps so please ignore that that dip and then in q1 Colin applied research started working with this protocol while Revenue per incentive spend used to be operate at around a 50 Cent average we were able to drive that number to about 2x in one quarter the main Insight that we were able to gather was how a portion of users were actually inelastic so this is a counter-intuitive data point finding that actually a big portion of users or a specific portion of users were not very responsive to changes in incentives which led us to hypothesize that reducing incentives for a specific group would thus increase profitability without decreasing volume and we saw some great results of that so far now of course these problems are Dynamic and they're obviously they're ongoing so this is just data points that we ingest to change how we change our strategies as these things change with Market changes here are some case studies that are available on our website from previous work that's led us to build this team and develop these models including work with sushi Swap and balancer here's some great partner feedback that I'd love to share I'm not going to go into the whole quote this is from one of my friends at immutable who leads the product and token team the main the main points when I drive home here is that immutable has been really happy to work with us and the results have been really strong we've been able to deliver sizable impact and ecosystem growth with tools of data science and wisdom of traffic we've already learned so much our models are just starting and just getting better and this is really just the beginning looking forward there's a lot to continue working on we are super excited about the industry shift towards concentrating liquidity and zkm and how that affects the liquidation problem the optimization and function models of those liquidation problems of those liquidity models need to be treated differently per sector it's taking rewards is a new field for us specifically and we're really excited to explore as many protocols ensure that the rewards that they're giving out for staking actually create a sustainable user base and people are not just farming mistaking we'll continue to provide research for the space as tarun our CEO and founder continues to write math and research papers for the D5 space frequently and another new product we're working on is treasury optimization if you are a protocol with a treasury and you don't want to see your safety module or Reserve suffer if the market is tanking neither do we we apply some of the learnings that we've got them from risk management and from our applied research team into optimizing your treasury research management apologies as I know I went through this pretty quickly and there's a lot of complex things here that I was trying to simplify in a very digestible way but here are some key takeaways before I kind of wrap up early for some questions Protocols are businesses that need to worry about revenue and spending protocols spend a ton of money on mining liquidity incentives and growth strategies and they need to really track these things solving this problem with simple heuristics or governance based votes is even harder a systematic quantitative approach is needed to maximize impact of their spending and that's what we do at Gauntlet we provide the framework for an informed and intelligent decision making for protocol teams and communities alike to to get on the same page on what quantitative metrics they should be focusing on and where their Investments should be better is used before voting on any major governance decisions all stakeholders should have a clear picture of what has been working what hasn't and what might work next if you'd like to partner with us work with us or just geek out on defy please do not hesitate to reach out to me I'll be in Taipei for a few days before heading back to New York and I'm very reachable via Twitter my DMs are nothing like vitaliks uh that's it for today do we have any questions no question [Music] okay great thank you so much thank you [Applause] [Music] so next up we have