Apologies for having taken so long to get this written up that it is coming at the last minute.
I have been hesitant to get back into this discussion since it has been so contested. I also knew I wouldn’t be able to resist spending many hours on a more detailed analysis, so here it is. I’d just like it to be over so we can move on, but some valid points have been made that it was not only those who were liquidated that suffered a loss. And those of us who avoided it might feel they are getting the short end of the stick after having gone the extra distance to keep the platform functioning.
I have one other observation I’d like to make that may further stir things up, but it should be pointed out. When we experienced reverse rebalancing it added collateral as well as debt to our positions, so by the time people were liquidated they had significantly more collateral than they did before the rebalancing started. So, 1/3 of the final liquidated amount would also be significantly more than 1/3 of the originally held collateral. Let’s follow through an example position. I recall one person saying they had a liquidation price of 0.40 bnUSD at the beginning and they were liquidated at about 0.60 bnUSD / sICX. If we take a starting state with
100k sICX collateral, that gives about
27k bnUSD loan
Rebalancing prices ranged from 0.88 bnUSD / sICX to 0.60 bnUSD / sICX, but let’s assume a worse price from the range was applied for all sICX added and take the sICX price to be 0.80 bnUSD / sICX, which is on the high side of the likely average.
If we take the amount of sICX added to the position as a variable x, and the liquidation price to be 0.60 bnUSD / sICX, then
(2/3) * (coll@liq) * 0.60 bnUSD / sICX = loan@liq
or,
(2/3) * (100k + x) * 0.60 bnUSD / sICX = 27k bnUSD + x * 0.80 bnUSD / sICX
0.4 * (100k + x) = 27k + 0.80 * x
13k = 0.40 * x
x = 32.5k sICX added to the collateral position by the time of liquidation.
By the time of liquidation they would have had 32.5% more collateral than at the start. While I don’t have exact numbers for this, it does seem close to what I actually saw happening. The 0.4 → 0.6 liquidation price move is close to what happened. It could have been more.
The point is that due to reverse rebalancing some of the people who were liquidated will be getting back more than 1/3 of the collateral they had before the price started to collapse. That is not the case for the first to be liquidated, but the first happened at a price closer to 0.85 bnUSD / sICX, before reverse rebalancing had any effect.
Let’s continue with the example to see what the state would be if collateral equal to 1/3 of the amount liquidated were returned.
Return 1/3 scenario
Collateral liquidated = 132.5k sICX
Amount to return = 132.5k / 3 = 44.17k sICX = 44.17% of original collateral
Value of return at the time of liquidation = 44.17 * 0.6 = 26.5 bnUSD
Value that the account would have had at the liquidation price if reverse rebalancing had not happened = collateral value - debt = 60k - 27k = 33k bnUSD
Returned value % = 100% * 26.5 / 33 = 80.3%
Honestly, that is not what I was expecting! The account liquidated after experiencing rebalancing is getting back more than 1/3 of the amount of sICX it started with, but it is at a loss wrt its starting sICX value, while an account that was liquidated at the highest price and suffered no rebalancing would receive all of their lost sICX in return. A flat return of 1/3 across all liquidated parties does not seem fair, but sorting it out to the point of perfect fairness appears more difficult the more we look at it.
This illustrates how differently we all experienced the event and how difficult it is to come up with a tailored solution for each account. The biggest factor that makes it difficult to devise a blanket solution is the variety of paths we all took in our battle to avoid liquidation, successful or not. We see in the above example that across the liquidated accounts we could easily have a range from 0% to -20% impact on account value with a simple 1/3 return solution. The same is no doubt true if we were to compare across non-liquidated cases. An account that did not have to pay off any debt simply ended up buying more sICX at a low price and will not suffer any effective loss in the end, even without any payout, while another that had to sell a lot of assets at a low price to pay off debt could end up sustaining a significant loss.
The complexity of a proposed solution will determine how much work it will take the community devs to implement it. Let’s go for a proposal that tries to account for the varying impact of rebalancing on the liquidated accounts. Roughly, those accounts liquidated later experienced more rebalancing and more loss. Without accounting for exact timing, the amount of rebalancing, and many other factors unique to each case it is impossible to calculate this exactly.
If we approximate the effects of rebalancing as linear over the time period from first liquidation to final liquidation, I’d suggest that we scale the return accordingly so that the first liquidation receives a return of 26% of liquidated collateral and the final liquidation receives 33.33% of liquidated collateral. With the timestamp of each liquidation event this should be a relatively easy adjustment to make.
For those of us who avoided liquidation it is even more difficult to devise a simple and fair way to make up for the excessive impact that reverse rebalancing had. Originally, I did not even think it would be necessary since the system was operating as designed even if it was putting us under pressure due to a short squeeze that many had never experienced before. For these, I’d suggest a nominal goodwill gesture proportional to the amount of debt held, followed by an increased incentive for borrowers going forward. For the goodwill gesture we could use the remainder of the sICX in the emergency reserve, which will be about 500k sICX. That will result in about a 5% sICX return based on the debt held on 1/20. As an encouragement for future borrowers I’d suggest we reduce the origination fee to 0.5% and award a bonus of 100k BALN from the reserve to borrowers proportionally for three months.
Liquidated Accounts: 174
Amount of sICX liquidated: 5,862,934.76 (note, this is not net of debt, this is the total amount of collateral liquidated)
Amount of sICX available in the Emergency Reserve: 2,268,364.66
Amount of BALN available in the Emergency Reserve: 720,881.53
Based on the above reasoning, here is my revised proposal:
Title: Return all sICX held in the Emergency Reserve, along with 3 months of doubled BALN rewards to borrowers, redirected from the DAOfund, and reduce the origination fee to 0.5%.
Eligibility to claim sICX: All those holding a Balanced debt position on Jan 20th 2022
Amount of sICX and BALN to use: All sICX held in the emergency reserve, plus about 300k BALN, to be apportioned as follows: Between 26% and 33.33% of seized collateral will be awarded to borrowers liquidated between Jan 20th and Jan 24th 2022 scaled by timestamp linearly between the time of the first liquidation and the time of the last liquidation. The remaining sICX, expected to be about 500k, will be awarded to non-liquidated borrowers proportional to debt on Jan 20th, 2022. Additionally, 3 months of doubled BALN rewards to borrowers redirected from the DAOfund.
Note: It was pointed out to me that the 300k BALN award from the emergency reserve is effectively the same as redirecting 12.5% of rewards from the DAOfund emission to double the ongoing borrower incentive, so I have updated that part of the proposal accordingly.