Dynamic liquidation fees

[crispheaney]

Currently, liquidation fees are static and applied regardless of the users health.

As outlined in this paper, once a users health breaches some margin ratio, a static liquidation fee can push the user into bankruptcy. To avoid bankruptcy, fee < margin ratio

Right now the liquidation fee is composed of a liquidator fee and an insurance fund fee. I think it makes sense to scale down the if fee based on the user margin ratio to avoid bankruptcies. For example, if fee = min(static if fee, max(0, margin ratio - liquidator fee)).

If this were the case, the only time there would be a bankruptcy is if margin ratio < liquidator fee.

Note that this largely just affects accounting. When if fee is charged and it causes a bankruptcy, the if pays for the bankruptcy immediately but ultimately makes it back a portion of it in revenue.

[Banditx0x]

If the IF's fees are reduced to avoid bankruptcy, but the IF would have paid for the bankruptcy anyway, wouldn't the end result be the same? I'm not sure if thats what the final paragraph is saying

[crispheaney]

The only real issue with having the bankruptcy is that the if stakers have to wait for the revenue pool to make it's way back to the if. This could potentially hurt people finishing a withdraw during that period.

Less importantly, bankruptcies are public and may hurt people's confidence in risk engine

[0xbigz]

this has been completed

this liquidation dynamic discussion continued here: https://github.com/orgs/drift-labs/discussions/855