On the 13th of December 2022, David Bouba shared a study that showed the performance of oracle-base liquidity pool strategies and how they outperformed those without them.
https://twitter.com/nullpackets/status/1602763544344133634
The research papers were based on automated market maker designs and liquidity pool payouts. The duration of the study was for one year and it was focused on bringing passive profitable MM strategies on chain.
They used a framework similar to the Markowitz model, which selects the most efficient portfolio by choosing securities that don’t align together, to reduce risk and improve returns.
The Markowitz model is also called the mean-variance model or standard deviation model. And some metrics for evaluating the liquidity pools were the presence of transparent and efficient products.
They also used a fully passive AMM design utilizing the Stochastic optimal control theory, which aims to design the time pathway to perform the desired task with soft constraints, limited risk of violation, and minimal cost.
And here is what they found out.
One, oracle-guided AMMs outperformed all oracle-free AMMs. Two, the quality of data provided by the oracles is important for the performance of the Oracle-guided AMMs. And finally, their model was strong against misspecification and mild oracle delays.
Using Markowitz’s finance theory, they define the maximum return they would expect from each level of risk. They used simulated pairs like ETH/USD, and they utilized the HODL strategy. Calculating the extent to which each strategy would perform on a risk-adjusted yield basis.
One way to drastically improve AMMs was to dynamically adjust the fees, depending on the market conditions. Another way was to introduce a means of perfect information which led to the introduction of oracles.
In such conditions of perfect information with little slippage, there was a significant positive performance. But that was all theory and it wouldn’t apply in real-life conditions.
So they simulated real-world projects like Uniswap and Curve Finance. The Oracle-guided AMMs still outperformed their counterparts under conditions like Arbitrage, lagged Oracles, and so on.
The optimal model performed well up to a point and then started performing poorly under high-risk tolerance parameters. This shows the importance of constantly updated prices on the chain.
Finally, they said some model misspecifications on variables like volatility, demand intensity, and drift to name a few. Under these conditions, the model still performed optimally.
This shows that the smart use of oracles greatly improves liquidity pools’ risk-and-return portfolios even under adverse conditions.
