Context
Lido spends >4M LDO each month on incentivizing liquidity for stAssets (see: Jul, Aug).
- In LDO terms, this is the largest expense the DAO has and at current prices represents >$10M/month.
As such, there’s a big opportunity in looking deeply at incentives: small improvements/optimizations could save the protocol millions per month (e.g. 10% would mean $1M saved per month at current prices). - Furthermore, Curve and Balancer pools on mainnet account for a majority (>60%) of incentives spend and as such those have been the first focuses of this research.
reWARDS v2 - Intro
The reWARDS Committee, led by @Jbeezy and @Kadmil has been vocal about the need to optimize the current program and there are two complimentary vectors we’re working on. At a high level:
-
How much liquidity does each of Lido’s stAssets need (and ultimately, how much does Lido have to incentivize LPs to attain those liquidity targets).
-
What’s are the optimal paths to reach those liquidity and incentivization targets (from a ROI point of view).
1. Liquidity Targets
Pool Composition
- For slippage minimizing purposes, even more impactful than raw TVL is the composition of (w)stETH/ETH pools.
From @chuck’s analysis, we can see that making the pools more balanced can have
LPs’ expectations
- LPs’ yield should be above stETH’s yield to compensate the additional risks relative to holding but not too much above otherwise we’d expect more and more stETH to be deposited into LP positions thereby unbalancing the pools further. (More analysis on this)
- Currently, a large % of the LPs’ rewards is in LDO, whose value denomidated in ETH (the principal) has and can be volatile. This can be a negative if we’re trying to keep the TVLs stable.
A possible next step here would be evaluate giving a part of the LP incentives in ETH or stETH (stable against the principal) (data), altough there are many downsides and seems unlikely.
Informed Liquidity Targets
- Lastly, we want to have informed targets for (w)stETH liquidity. For this, key inputs are: slippage tables (such as the above), liquidations’ data (actual and potential), amongst others.
@mcnut has been doing a great analysis on this and is also bulding a monitoring dashboard. - The goal here is obvious – to make stETH (and other stAssets in the future) as liquid as it needs to be: keeping all the benefits/attractiveness its liquidity brings while avoiding excessive LDO incentives spend after a certain point.
- As a next step, we also want to have elasticity models to ultimately get at what TVLs will do as incentives given vary.
.
2. Incentives Optimization
For target liquidity amounts, there are numerous (and growing) ways of getting the incentives to LPs. We have looked into most of the mains ones and will keep researching new methods/tools/protocols for this as they get built.
For both curve and balancer, we see three main ways of directing incentives:
- Giving LDO directly to the LPs;
- Bribing: paying for emissions-directing votes across the stacks (veCRV, vlCVX, veBAL, vlAURA, veSDT, etc)
- Owning: purchasing these governance tokens and constantly voting emissions to our pools with them.
Bribing:
Bribing markets such as Votium, Hidden Hand, Paladin and others follow market forces meaning their ROI potential for us can and will vary wildly each round.
As such we need to have real time analytics on them. Starting with this layout:
To inform each reWARDS budget, the committee will look at the high-level emissions/bribe metrics for ROI potential and increase or decrease .
Owning governance power:
Having the Lido DAO own and lock CRV, BAL, AURA or CVX (or others to be considered soon) would function as an initial larger reWARDS investment that would then pay off over time.
For this owning vs. renting decision, the main metric is amount of time until payoff (i.e. how many weeks until, say, 1 veCRV constantly voting leads to 1 CRV worth of emissions towards the stETH/ETH pool)
Next steps here would be to discount that back (accounting for opportunity cost) + account for price risk of being effectively long that position as a DAO.
.
Reporting:
As a next step, we also want to track, store and make available the performance of Lido’s rewards program over time (for optimization purposes as well as for transparency to the community).
What we’re thinking:
- ROI numbers, in both multiplier and absolute dollars, of LDO spent on bribes;
- Pools’ data including TVLs, TVLs variations, yields, yields variations, compositions, composition changes and slippage tables;
- Some light commentary on that round and/or month;
All of this as extra support to the reWARDS budgets JBeezy already publishes monthly.
(Also, let us know what other things you’d like to see in the reports!)
Roadmap:
We intend for this research to start having an impact from the september reWARDS budget onwards. This will/can include:
- Reducing Lido’s spend on bribes that have not been cost effective the previous rounds;
- Reducing overall incentives spend to account to 1. target 2. to account for recent price increases
- Possibly starting new bribing targets if the previously mentioned numbers make sense.
Mid-term, we intend to continue this work, expand it to other DEXes (e.g. Velodrome) and other ecosystems (e.g. L2s), produce the first aforementioned efficacy reports, explore new bribing platforms & targets and have the monitoring framework in place for overall liquidity needs.
Feedback:
One of the main goals of this post is to spark community engagement and feedback around reWARDS.
We’d love to hear ideas, proposals, new tools/protocols, further data!