Name of Project: Predictive Model for Borrowing Cost of Popular DeFi Protocols
Proposal in one sentence: Help DeFi investors make better decisions using quality information by presenting the best borrowing and lending opportunities in DeFi, as well as showing predictions of how that market will behave in the future.
Description of the project and what problem is it solving:
A bank is not a bank if it doesn’t take deposits and make loans; lending and borrowing are the most fundamental building blocks of finance. In the world of DeFi, lending protocols directly make up 44% of all TVL1 and indirectly touch nearly all other DeFi protocols.
Yet the rates for borrowing and lending are highly volatile, exemplified by the chart2 above: Worse, the factors by which these rates are derived are opaque to all but protocol insiders. We ask ourselves: How can DeFi builders construct upon these primitives without any idea of how they might behave in the future and why?
We believe that a smart application of data science will give us clarity. The resulting models would have several potential use cases in DeFi:
- Automated rebalancing strategy if the algorithm is predicting better rates in other lending protocol
- Reduce financial risk by withdrawing funds in case of a high borrowing cost projection
- Use favorable rates as a tool to get leverage for other DeFi investments
The scope of the project is as detailed in this document. Algovera have been running a series of hacking sessions to exhibit the process of developing this DeFi algorithm. The recordings can be found here. There is also a shared GitHub repo for the project. This grant funding would be used to further incentivize contributions to the project from data scientists.
- Publish borrowing cost dataset to the Ocean marketplace
- Develop first version of time series forecasting model for predicting borrowing cost
- Publish algorithm to the Ocean marketplace
Funding Requested: $1000