Ranked liquidations

The first level is the open bidding process, during which bidder will submit a series of volumes $\{Q_i\}$ that they are willing to take on from the liquidated account, as well as a percentage of the resulting change in LMR that they request as reward. The open competitive process should ensure that the reward parameter is optimized.

The other cornerstone for this procedure is a scoring function used for liquidation bids, which summarizes the reward parameter and the market impact. For a bid requesting volumes $\{Q_i\}$ of $n$ assets for a reward parameter $d$, the score is

$$\mathrm{score}(\{Q_i\},d) = \frac{1}{n}\left(w(1-d) + (1-w)\frac{\sum_i Q_i \times\mathrm{pSlippage}i\left((Q_i(Q{i,max}-Q_i))^{1/2}\right)}{\sum_i Q_i}\right)$$

where $\mathrm{pSlippage}_i(Q)=\phi Q^\beta/\mathrm{price}_i$ is the estimated percentual price slippage (using governance set parameters).

An important feature is of this scoring function is that it disincentives over-liquidations by multiplying by $1/n$ (which in practice stratifies bids according to the number of requested assets), as well as maxing out the slippage benefit at half the available volume.