Once upon a time on mainnet, there were six blind yield farmers, each highly skilled in their own farming techniques but unfamiliar with the full scope of optimization strategies.
They were guided to the mysterious entity known as an “elephant,” which possessed powerful tools for cryptocurrency. But, lacking complete understanding, the farmers each sought to explore this entity in their own way.
The First Farmer touched the elephant’s Liquid Lockers and said, "Aha! Stake DAO is just like the common liquidity pools I know. It merely is a yield farm."
The Second Farmer felt the elephant’s Vote Market and declared, "No, no, you are mistaken! The elephant is built for bribes to flow to stakers."
The Third Farmer explored the Strategies allocations and proclaimed, "You are both wrong! The elephant is like a decentralized team of quants. That's its essence!"
The Fourth Farmer stumbled upon the Factory and insisted, "How can you not see? The elephant’s true strength lies in its use of permissionless deployments."
The Fifth Farmer discovered the meta-governance and concluded, "None of you understand! Elephants are all about democracy, allowing participation across protocols."
The Sixth Farmer sensed the $SDT token and cried out, "All of you are wrong! An elephant is just a token! Let us trade it at the casino!"
They argued and debated, each one firmly entrenched in their understanding, unable to see that each had only grasped a part of the truth. The Elephant was known as Stake DAO, embodying all of these things and more, a complex and well-designed entity offering numerous benefits to those who understood its entirety.
It was then that a wise degen entered the scene. Hearing their dispute, he explained, "Each of you has touched only a part of Stake DAO. If you combine your knowledge and embrace the full picture, you will understand the true potential and innovation of this system."
The farmers listened, learned from each other, and finally embraced Stake DAO in full. Together, they thrived, reaping rewards they never thought possible, and the wise degen smiled, knowing that true wisdom lay in collaboration and comprehensive understanding.
A hot new video just dropped…
Yesterday the full information dropped. Yield farmers interested in free money should pay attention to Stake DAO’s new white paper on Onlyboost.
The concept is simple: there’s a mismatch between various protocols’ amount of veCRV and share of the TVL. The most popular, Convex, has 85% of the TVL but only 49% of Curve deposits.
Convex is so popular that the boost winds up getting diluted, maxing out at 1.8x instead of the maximum 2.5x.
In other words, Stake DAO and Yearn have an unused pile of veCRV just lying idle. Nature should theoretically abhor an inefficient market, but cryptocurrency has never been especially rational.
Stake DAO offers Onlyboost as the solution. No doubt Stake DAO hopes that correcting this imbalance will mean more TVL flowing into their platform, but don’t overlook it for this reason. The solution is novel and thoughtful, and the paper is well worth a read.
At the high level, one might imagine you can simply split LP deposits among platforms in whatever manner optimizes the boost, and then autocompound. The white paper notes that Yearn does something like this already.
However, when they crunch the numbers on this, some complexities emerge. For starters, if the gains are too low, then an extra layer of fees can simply erase any potential gains.
Stake DAO considers how to attack the problem in a fashion that could be deployed and managed automatically using smart contract logic. They start with the base case, which just looks at balancing deposits between Stake DAO and Convex with no regard for fees/incentives. The aim is just to optimize boosts.
Here is what it looks like:
For small deposits, the gains can be as high as 25%. However, once deposits hit a certain size, the effect tapers off gradually, occasionally turning negative. This is due to the fact that once boosts are equalized, the raw APR from Convex provides more incentives.
The attempts to optimize the formula at this point become more complex. For example, where Convex’s extra incentives are straightforward to calculate, Stake DAO’s are volatile and almost impossible to project. This logic might be tough to calculate onchain, so Onlyboost’s logic mostly has the effect of erring on the side of depositing into Convex at the edge cases.
Their ultimate solution looks as follows:
As you can see, the minimum here gets shifted left just a bit, because Onlyboost biases towards Convex around this inflection point and avoids the region where the boost can turn negative.
When they run tests on this over a variety of gauges, their modeling suggests that Onlyboost will deliver anywhere between 0-12% improvements, mostly falling between 7-10%.
The effects here are real but slim. The existence of the potential 0% increase means that Onlyboost cannot have fees if its architecture is to promise its LP returns would be net positive. Accordingly, the Onlyboost white paper assures the protocol will not charge fees atop any deposits that are sent to Convex. The upside to Stake DAO from Onlyboost would come only from any deposits that get shifted to Stake DAO using this algorithm.
The final consideration is how to architect the smart contract infrastructure governing Onlyboost in a gas efficient manner.
Since depositing into Convex and Stake DAO are gas prohibitive, Onlyboost uses a system in which users make a cheaper deposit into a vault. These deposits accrue until any user makes a call to transmit accumulated funds from the vault to the appropriate lockers.
The community response has been appropriately positive.
The protocol is expected to launch in September. We plan to try it out wherever it makes sense.
As markets continue their slump and users turn increasingly PvP, we have to imagine that solutions that passively increase farming yield will find easy product market fit.
For a bit more context, check out yesterday’s Leviathan livestream, in which Hubirb from the Stake DAO team joins the fray:
Loosely related, but if you are interested in some nuance on veCRV optimization, check out this thoughtful thread on gauge vote decay rates….