Even the Anchor docs do not explain, and it seemingly has no sense to keep the $2000 collateral value hard limit required for partial liquidations.
If my total collateral value is $1900, the full amount gets liquidated, if it’s a hundred more ($2000), then only $400, as selling fifth of all collateral is enough to reach 80% LTV. Enabling partial liquidations without a limit would improve user experience imo.
I agree. I always wondered why the decision was made to set the limit at 2k. I thought it may have to do with tx fees or maybe liquidation bids?
Still I think work should be done to find a way to remove this limit
Agree. This has a disproportionate negative effect on those with the least ability to withstand the loss.
Had no idea this was a thing, guess I missed that in the docs, I agree that 2000 UST is a stretch but a lower bound will need to exist, otherwise we may end up with micro liquidations that are not worth to liquidate. But with the current fees, 5 UST should be more than reasonable.
The 5 UST limit solves the fee issue, it’s just a bit clumsy solution, isn’t it? Another option would be just to not execute liquidations below that threshold.
That could create an attack vector, automate the creation of millions of small loans and let them be liquidated, but because they’re so small they wouldn’t be, now the protocol is in default to it’s depositors.
The amount could be calculated live, if the remainder of a loan would be smaller than 5x (configurable value for governance) the current fees, then do a full liquidation.
I think that the limit could be lowered down to $500 of collateral value with no spam risks or attack vectors or making loans unworthy of liquidations.
When liquidations occur they the collateral is sold until a 75% borrow limit is reached. This requires paying fees and executing transactions. But at $500 collateral value a liquidation would still make economic sense to execute in my opinion.
With a collateral valued at $500 about $125 would be liquidated in a crash and the amount spent in fees would equate to about 0.4% of the liquidated value.
Maybe if $500 is too little, 1k could be considered a more worthy alternative
Yea that would work, and it can go even lower, but we will always need the lower bound.
Interesting - did not know of this inefficiency within the protocol.
Would support this readjustment towards a lower limit (500 UST or 1000 UST) barring no smart contract risk.
Agreed, this should work well.