# Aave Lending curve findings

Aave's lending pools currently use two different models, Variable and Stable Rate, with each of those models using different parameters depending on the tokens pairs, these params are decided in a way that optimises the lending pools utilisation.

We will be looking at the Variable Rate pools as it resembles the Sovryn lending pools the most.

## Variable Rate Lending

The interest rate follows the model:


And the deposit APY follows:


These are the universal formulas for Aave's Variable Rate pools.
However, they have many different parameters:

These variables are:

* Uoptimal : the Optimal % of Utilisation that a pool is looking to achieve
* Base : The APY% at 0% Utilisation on the lending curve, in other words y for x=0.
* Slope 1 : The Rate multiplier when utilisation is under optimal utilisation
* Slope 2 : The Rate multiplier when utilisation is over or equal to optimal utilisation

The model parameters are calibrated based on a token's/currencies' risk profile.
Aave distinguishes assets that are used predominantly as collateral and classes them as volatile assets, they always need available liquidity in order to enable liquidations. That is why for assets in this category require a low Optimal Utilisation. Furthermore, Aave place an importance on the liquidity of these assets on their own platform, noting that more illiquid assets should have lower Optimal Utilisation Ratios than highly liquid assets.

These Rates are all variable and should be recalibrated as the market evolves, bringing us to another point: lowering the Optimal Utilisation of assets with high yield earning liquidity mining and staking programs, therefor increasing the borrow costs for these assets. The reasoning being that users are incentivised to borrow these assets in order take advantage of the high yield opportunities in DeFi.

Finally, it is imperative to also consider the current rates of other lending protocols and stay competitive with our borrowing and lending rates.

## Deposit APY without any stable lending

Aave uses a formula that determines the APY of asset suppliers using the Utilisation of both the Variable and Stable Rate pools of an asset.

As we are looking solely at Variable Rate Lending, we can ignore the Stable Rate related values.

## Lending pools on Sovryn

We will be looking at the rBTC and XUSD Lending pools on Sovryn as they are currently the most important pools on our platform.

* XUSD: 13+ Million XUSD Borrowed and 4 million + XUSD Available in the pool. Over $17,000,000 worth of XUSD. A low liquidity stablecoin that has a low risk of losing peg as it is easily tradeable for and is backed by the leading DeFi stablecoins such as USDT, USDC and DAI.

* rBTC: 13+ rBTC Borrowed and 65 rBTC Available in the pool. Almost $5,000,000 worth of rBTC. A low liquidity Bitcoin DeFi coin, so subject to the same price movements as the largest crypto asset on the market, but vulnerable to higher slippage and price impacts due to on-chain trading.

Two different comparisons are available for these assets:

- A comparison with lending pools with similar liquidity and utilisation
- A comparison with another Stablecoin and a wrapped Bitcoin token

## Liquidity and Utilisation

Looking at utilisation on Aave's pools, the BAT pool is eerily similar to our rBTC pool. With just over $5 million in the pools and a low utilisation rate of 10%.
It's parameters are the same of that of BAL, CRV, ENJ, LINK, MANA, MKR, REN, xSUSHI, YFI and ZRX.

* U(optimal) = 45%
* Base = 0
* Slope1 = 7%
* Slope2 = 300%

As for the XUSD, most of the stablecoin lending pools have much higher liquidity to be able to compare, with the exception of USDP (Unit Protocol's stablecoin which has struggled to remain over peg) but I strongly believe this token has a much higher risk profile than XUSD due to it being minted using many different DeFi coins as collateral.

## Comparison by "type" of asset

If we look overall at the different stablecoins on Aave, we can see that they all have very similar parameters:

The higher liquidity and lower risk stables have a higher Optimal Utilisation Rate and a lower Slope2 multiplier. But they all have the same base and Slope1 parameters.
It would appear that Aave is more than happy with the 80-90% U(Optimal) range and 60-100% Slope2 range as they have used these values for all of their stablecoin pools. If I was to recommend parameters following Aave's example, I would suggest that we use the 80%-0%-4%-100% parameters. Perhaps even increase our slope2 or decrease the Optimal Utilisation Rate to account to our current high utilisation on the XUSD pool.

As for the rBTC pool, we would be copying their wBTC pool that has parameters of 65%-0%-7%-100%
Although I do not recommend this as the wBTC pool currently has over $2 Billion USD of wBTC in it (over 32k wBTC). The difference in liquidity here is too important and should not be neglected just because they are 'both some kind of BTC'.

## Parameter Rebalancing

I do not believe that Aave's rebalancing style suits our platform's profile. Aave is a relatively old and battle tested protocol with one of the highest TVLs in all DeFi. It may make sense to them not to rebalance too often, but Sovryn is a much newer protocol that is building faster and making substantial changes to the entire RSK ecosystem with every large product release. This requires a more dynamic and proactive approach to the rebalancing. We also do not know how similar to ETH DeFi the Sovryn ecosystem acts, nor how connected it is to other chains (ie. how many users bridge back and forth between RSK and other chains in pursuit of the highest yields/ how deep the Bitcoin DeFi moat is). Although these suggested parameters may seem like the best options based on Aave's data, it may be that we will need to quickly rebalance if the problems we are looking to fix are not resolved, and fine tuning most definitely will be required.