Update parameters for pool 3 (DAO: Community) to minimise risk

Summary

For pool 3:

1. Set liquidation incentive to 30%
2. Set collateral factors as follows: RGT 60% ETH 65% INV 50% ALCX 50% SFI 50% WBTC 60% USDC 65% SUSHI 60% YAM 50% MPH 50% LINK 65% BADGER 60% COVER 50% FTM 50% DAI 65% HEGIC 55% KP3R 50%
3. Set all 100% utilisation borrow rates to approx. 700%
4. Set 80% utilisation borrow rates as follows: RGT 50% ETH 30% INV 100% ALCX 100% SFI 100% WBTC 30% USDC 80% SUSHI 30% YAM 100% MPH 100% LINK 30% BADGER 30% COVER 100% FTM 50% DAI 80% HEGIC 50% KP3R 50%
5. Recommend supply caps based on liquidable amounts at 10% slippage (Some of these changes may require time for implementation due to ongoing developmental work, this proposal however binds community support for these changes.)

Background

Fuse is an exciting new product by Rari that allows you to borrow, long or short any asset against any other asset. To iterate quickly and obtain real market data, pools were created and parameters were set intuitively. Now that we are more mature we are looking to establish a more standardised and data-driven approach to risk.

Resource: A standard risk model for Fuse

Applying these principles to Fuse Pool 3 yields a number of desirable and potentially necessary changes to safeguard user interests.

Abstract

Changes are guided by the proposed risk model (linked above).

Change 1: Liq incentive to 30% As described in our model, protocol only needs to absorb shock, whereas liquidators need to absorb shock, slippage and gas. Therefore 8% incentive for liquidators is too low and leftover allowance for protocol (100% - 8% - cf) is unnecessarily high.

This is a high risk pool comprising of numerous volatile and illiquid assets - where all assets can be borrowed against each other. Therefore shock allowance will be decided by the most potentially volatile asset in the pool. We will allow 20% for shock.

Since some assets are illiquid, 10% has been allowed for slippage (or gas).

Change 2: Change CFs Keeping in mind a liq incentive of 30% and a slippage allowance of 10%, we can set CFs based on shock tolerances of individual assets. These get set in the 50-70% range due to 20% max shock tolerance and 30% liq incentive. Less volatile assets get higher CFs.

Change 3: Set 100% util rates to 700% Since all assets can be borrowed against each other, it does not make sense to set different wind-down periods for different assets. A wind-down period of 2 month may be ideal. This corresponds to a 100% util rate of around 700%

Change 4: Change 80% util rates 80% util rates are to be set based on the maximum borrowers will be willing to pay for the likely use cases.

Use case of USDC, DAI is borrowing and longing other assets. There may be high demand for this so 80% has been set.

Use case for most other assets is shorting. Higher interest rates are charged on more volatile assets due to higher potential profits and risks involved.

Suggested rates are experimental, and may be lowered in the future if sufficient borrow demand is not found.

Change 5: Recommend collateral caps at 10% slippage for those assets Since we have allowed a 10% slippage tolerance, we will also recommend collateral caps at 10% slippage.

Suppose LINK-ETH has liquidity primarily on Uniswap and this liquidity is 1,300,000 LINK against 22,300 ETH. Selling 65,000 LINK will cause 10% slippage, hence this Fuse pool will have a recommended supply cap of 65,000 LINK.

If this Fuse pool has more than 65,000 LINK deposited, the risk of this pool increases and appropriate action such as decreasing collateral factor or implementing a supply cap may be needed.

Motivation

DAO Fuse pools must ensure relative safety of both their lenders and borrowerers under extreme market conditions, while simultaneously being able to offer high risk products such as leverage and shorting on illiquid assets. We believes these parameters meaningfully capture this tradeoff.

At the same we do not wish to disturb the user experience of those who are already using these pools. Although liquidation incentive has increased, collateral factors have also been increased for many assets - leading to an overall better user experience.

An attempt has been made to ensure interest rates are competitive and maximise both lenders' yield and protocol revenue, without trampling over borrower interests. High rates at 100% utilisation may prove necessary from a risk perspective, especially in these early days.