This document outlines an initial analysis and recommendations for optimizing the Borrow and Supply Rates within the Radiant Protocol. The goal is to adjust the current rates to target rates according to market dynamics and suppliers’ and borrowers’ activity.
Over the past few months, there has been noticeable interest rate volatility within various markets on Radiant, particularly stablecoin-related markets. Given its impact, the RDNT liquidity incentive scheme is a very important factor in the context of Radiant interest rate parameterization. Currently, these incentives paid to borrowers drive borrow rates, and thus utilization rates, to very high levels due to borrower subsidization in native RDNT yield and/or minimizing effective interest paid, especially when leveraged. This can be seen through the graphs below, depicting a 10-day moving average of borrow APYs through the various markets over the last 90 days, whereby stablecoin rates generally resided above 15%-20% to as much as 50%-60% on Arbitrum for long periods. This was additionally the case within the WETH market on Ethereum, oscillating between 10% and 30%. Thus, to mitigate any theoretical worst-case scenario whereby the utilization rate is deemed too high to liquidate positions, pure interest rate parameterization necessitates substantially more conservative values than that of the competitive market.
High Reserve Factor = Harder to Dynamically Incentivize Lenders:
Radiant employs a 50% reserve factor on all markets, distributing the relevant yields to RDNT lockers, thereby driving utility for the underlying reward token distributed. However, this strategy falls short in the context of utilization/interest rate reversion on its own, as supply interest is effectively cut in half. Thus, the relevant interest rate parameterization is much more a function of RDNT price growth and the distribution of RDNT rewards than that of interest rate growth, especially concerning our targeted Uoptimal rate. This further holds when referring to edge cases whereby there exists borrowing demand for a given asset, however, the relative reversion of the utilization rate is contingent on borrowers repaying (disincentive) much more than that of supplier deposits (incentive). In the context of stablecoins, the overall effect on utilization rates can be seen below:
By proxy, the borrow APY in markets whereby the utilization rate hovers around Uoptimal is contingent on the underlying price of RDNT. This is especially true in the context of leveraged yield farming, as users realize the amplification in both supply and borrow yield, leading to a net outcome that requires high borrow rates to converge to a net unprofitable state.
As such, to adequately parameterize these markets, it is imperative that we attempt to converge the effective net APY in the range below Uoptimal. This can be achieved through raising Uoptimal whilst simultaneously raising slope1. By requiring more capital borrowed to reach Uoptimal whilst increasing interest rates, we diminish the net reward realized with respect to the increase in utilization rate, due to the dilutive effect as more borrowers come in. On the contrary, supply rates will stabilize at higher levels, further increasing liquidity and thus converging below our Uoptimal target. Otherwise, these oscillations will persist due to the amplification when above Uoptimal.
High Liquidation Bonuses:
The 15% LB requires large amounts of debt paid off to revert a given position to a solvent state.
Large concentration of leveraged looping: In tandem with the aforementioned liquidation bonus, the unwinding of very large subsidized market-exposed positions in the context of liquidations requires a larger relative utilization rate buffer.