So these are some great points, thanks. We’ve just finished indexing every single bid execution since the new queue went live, so can speak to some of these with real data.
Under the ORCA-like system, which has not been tested through any major crash, we would need stakers to deposit MIM in queue contracts where the MIM would not earn interest.
This is a very fair point, and one we hear a lot from the Terra community today. The allure of a reliable, consistent, fixed APR is pretty strong compared to an unknown return from bidding on liquidations.
It’s only now that we’ve been live for ~ 6 weeks that we can start to pull some numbers together to make a comparison as naturally market downturns and the associated liquidations are unpredictable. What we’ve found is that a combination of the compounding effect of discounts (average ~8%), puts effective APR at around 80% for fairly conservative bids. When we also consider, that the general trend that dips recover (obviously not universally the case), these APRs start heading into the triple digits.
That is all to say that we’re pretty confident the returns from a sensible bidding strategy - which we plan to provide data, analytics and tools to support - will greatly exceed returns from the “safer” yield-bearing options.
It would mean that a few whales with millions of dollars, plus the Kujira/ORCA protocol, would receive the vast majority of liquidation proceeds.
This is definitely not the case, kind of the opposite really. Sure, whales with lots of dollars will still participate, and to your point below it’s really important that they do, but this approach allows everyone else to participate, too. The whales no longer have an advantage - if you bid at 5% and they bid at 5%, you both get a 5% discount. No more fastest-finger-first
For this proposal to work, Abracadabra would need hundreds of millions of dollars staked in the backstop liquidation queue contracts, otherwise the queue would fail (run out of money) during large market crashes.
This is a very fair point and something that 100% needs to be considered when the protocol is designed. The way we manage this today is with a bid_threshold parameter, which determines the total pool size below which there is no longer an activation period on bids. This activation period is a mechanic designed to prevent bots from sniping collateral at a slightly lower premium during market downturns (and works really well!), however as you rightly say, during a large crash it can mean the pools run out of funds.
When the pools fall below this threshold, bots are fully at play again and can implement their strategies as they do today - be it flash loans or whatever. It becomes a safety backstop to prevent insolvency in extreme situations, as opposed to the de-facto operation of the system.
Some comparison numbers - in 6 weeks, 121m UST has been used to repay collateral. 111m of that came from bids that waited for activation, whilst 10m came from bids that skipped the activation window. Some of these bids may have been from regular users bidding during the downturn, others may have been automated strategies. Regardless, the maximum pool that has been used is the 13% pool. No bids in any of the 14 - 30% pools have ever been touched - ie the system has always remained solvent.
There’s definitely a requirement for bots to provide this backstop, but in the more general day-to-day operation this opens the opportunity up to a much wider user base.
Also in the event of a major crash, there is a significant risk for ORCA queue participants. Those who purchased ETH collateral at a 5% discount could be rekt if ETH goes down significantly in a day. The collateral the stakers bought at a “discount” could end up underwater
This is true, plain and simple. And is the risk you take when purchasing any non-stable asset, whether it’s via a liqudiation or on an exchange. What we’re seeing is a narrative forming where users see bidding on liquidated collateral as similar to setting limit orders on an exchange - if effectively allows you to “buy the dip” (in fact it lets you buy the dip at a discount).
The simple fact is that during a market downturn, some people will experience loss. Whichever way you cut it, if you lock in your “profit” at the same moment that you buy liquidated collateral, somebody else has to buy it. It’s either you getting rekt or them. At least with a democratized approach like this queue, there is a broader range of strategies at play. Allowing folks to buy an asset that they are already accumulating, during a dip, at a discount, results in much more buying to hodl than immediately selling to lock in profit and allow a re-bid in the next block.
If the purpose of this proposal is to increase the amount the community gets during liquidations, then there is a parameter in the cauldron contracts right now that can be adjusted to increase the percent of the liquidation that goes to the treasury.
This is more about allowing the general defi user to access liquidations in a way that they weren’t before, and take profits that were previously exclusively the domain of whales and bots. If you remove that profit and just return it to the treasury, where is the incentive to provide the funds for liquidations? Why would I put up my money so that everyone else in the system gets a cut of those profits? I think the result of something like this is an incredibly insolvent system.
If the purpose of this proposal is to decrease the frequency or pain of liquidations for users, then this will not accomplish those goals.
I’m afraid it measurably does. Where previously borrowers were giving up 30% of their collateral to bots, the game theory at play with bid strategies has dropped that by over 70% to 7.9%. Borrowers getting less rekt, and 7,500 individual defi users helping to provide that capital and buy an asset at a discount.
I think the key thing here is that the profits that have been enjoyed by those running bots has not been stated clearly enough.
If I have a $1m pot, and a bot that’s liquidating at a 25% premium, and say $5m of loans become at risk over the course of a few blocks (and that’s a small event), then each block the bot buys the collateral and market-sells it to lock in the 25% profit, then in the space of 5 blocks - the bot owner is cleaning up 1 * 1.25 * 1.25 * 1.25 * 1.25 * 1.25.
The bot owner now has $3m where earlier that day they had $1m.
Do that a few times a year and you’re turning your $1m into $10m easy. That extra $9m doesn’t come out of thin air - it’s taken from the borrowers, and it’s taken from the market when the collateral is sold. What we’re doing here is taking that $9m, and giving everyone a chance to bid on it. And in doing so also minimizing the hit to the borrowers.
In my opinion, this is what democratizing finance is all about, it’s what defi is all about. Half that battle is designing and building financial systems that allow anyone to participate, and the other half of that battle is designing and building dapps and interfaces that allow anyone to use those systems. To use them safely, securely and with all the information they need to make good financial decisions.
Neither of those things are easy, or free to build, and we should be looking to distribute the responsibility for building and managing this infrastructure, rather than try and centralize it. Isn’t that why we’re all here in the first place?
