[Proposal] Adjusting liquidation parameter

Hello I hate to post this here as it’s off topic and I apologize, but can someone please tell me what I am doing wrong? I am trying to deposit and the “proceed” button stays dark and does not allow me to select it. My funds are right there in the wallet but it just won’t seem to let me deposit. Is there someone I can contact? Thanks much, again i’m sorry to invade this thread but no one will respond to my thread.

Fully agree with @ColonelSanders that lowering the max_premium_rate might be more detrimental to the solvency of the protocol than the potential benefit it could bring in terms of decreasing network congestion.

I get that this is meant to be a stop-gap measure before bidding liquidations are implemented, but the risk-reward doesn’t really make sense. Also, I question how effective a lower premium of 15% is going to be in reducing the spamming of liquidation transactions - wouldn’t the same number of liquidation transactions still get triggered in the event of a sudden crash (when price plunges further than what most people think is “safe” at 30% LTV for e.g.)? This proposal will only have the effect of bringing forward the point where arbitrageurs stop liquidating (i.e. when Luna-bLuna swaps spike to 1.15) - breaking something to solve for something else isn’t really an elegent solution.

I think we also run the risk of depositors losing faith in Anchor protocol and spamming withdraw transactions even before we reach the point where collateral value < value loaned out. The network will become even more congested, compounding all the other problems.

However, I do think the raising of the liquidation_threshold (while being unpopular) might be a good move if it really helps to reduce volume by 22%. However, as @ikarusinthesky mentions, this needs to have the appropriate disclosures to avoid a huge backlash afterwards.

If these were 2 separate proposals I would vote yes to raising liquidation_threshold and no to reducing max_premium_rate. But since the increased risk of insolvency far outweighs the benefits of reduced congestion, I’d think an overall “No” would be the prudent vote.

Furthermore, with more people having the experience of getting liquidated + the use of AnchorHODL script, we’ll naturally see less liquidations and hopefully less congestion too.

1 Like

The best way to defeat this proposal is to not vote and prevent Quorum from getting reached. I think this forum is the best way to voice the “No” opinion.

We have to reduce the max premium at least temporarily until the Anchor team is able to upgrade the liquidation logic into an auction.

Currently there is no market rate for liquidations, all liquidations occur at the max premium (30%). This extremely punitive rate is discouraging the necessary borrow volume that will ensure protocol solvency.

Once the liquidation contract is upgraded with the auction logic, we can discuss uncapping the max premium.

If you do not vote for the proposal or vote against, there is another risk of failing to attract enough borrowers and depleting the yield reserve. Right now the priority is to increase the value proposition for borrowers so Anchor can scale properly.

1 Like

@ZenDog Appreciate your comments & agree that encouraging more borrow volume is a very important goal for the protocol. What I do not agree with, is that lowering max premium is the best way to solve for this. We essentially introduce a much larger risk (of arbitrageurs not entering the market and risking the solvency of Anchor) in order to maybe boost borrow volume/reduce congestion.

I think the other proposal of trying to back-stop the yield reserves could be a less intrusive way of keeping the protocol well-functioning until bidding liquidations are in place.That being said, it’s a non-trivial matter using community funds for this and more discussion needs to take place on that topic.

I’m also slightly optimistic that we have the time to build out the bidding logic due to multiple changes in the environment since the selloff in May.

  • Launch of Harpoon (allowing retail to fill the shoes of bots if the bots can’t/won’t liquidate)
  • AnchorHODL script (biggest bLuna providers now have a way to avoid liquidation)
  • Reduction in risk-taking sentiment (I’m guessing people are generally more conservative with their borrow LTV)
  • Lack of a significant bLuna-Luna spike when Luna most recently dipped below $5, suggesting that the above 3 changes may already be having an impact

As such, I don’t think we’re in a dire enough state to warrant introducing such a “drastic” (and unproven) measure to try to improve Anchor. Dismantling this may be more difficult in the future as a rise in max_premium is always going to be viewed negatively.

The current problem is that the max premium is so large that it creates gaming opportunities for short sellers when markets are volatile on illiquid exchanges.

  1. Simply by waiting for cascading liquidations while actively shorting Luna on exchanges and piling on sell pressure through Luna acquired through liquidations.
  2. At worst case, possibly creating opportunities for oracle attacks leading to cascading liquidations. (Short Luna → move price down → create liquidations on Anchor → buy bLuna w/ 30% premium → instant burn to Luna → sell on exchange to move price down further)

While it is true that this could technically be done with any other crypto lending protocol that uses oracles, the sheer size of the default premium (30%) makes this attack vector more likely on Anchor.

The folks who understand this are the ones removing their borrow positions from Anchor. It is undeniable that the liquidation mechanism currently in place is an existential threat to Anchor.

The general delevering of Anchor in terms of total borrow and the reduction of average LTV of the remaining borrowers are the things that are making this attack more difficult, but this situation also isn’t sustainable due to insufficient cashflow.


The folks who understand this are the ones removing their borrow positions from Anchor. It is undeniable that the liquidation mechanism currently in place is an existential threat to Anchor.

Instead of lowering the max_premium_rate, why not just raise the maximum LTV allowed before liquidations occur to encourage more borrowing? Let’s propose raising the rate to 60% or 65%. Maybe consider even 70%!

The instant burn bLUNA → LUNA rate more than offset the discount liquidators got from the 30% premium for a significant amount of time during the crashening. That’s why we saw many liquidators stop liquidating.

We are attacking the wrong variable here. I think lowering the max_premium_rate to 15% is an enormous mistake that is an existential threat to Anchor. I would dare say that ANC should go to near 0 if this is passed. Increasing the maximum allowed LTV, on the other hand, will significantly increase borrowing on Anchor and I think it’s clear how that will impact the value of ANC.

I find myself agreeing with the decisions of TFL guys like EJ the vast majority of the time. This is one of the rare instances where I think they are way off base.

Increasing the LTV doesn’t solve the problem that oracle attacks are possible through short selling on exchanges.

This is a fairly important vulnerability. This isn’t just an issue with Anchor. It’s creates a vulnerability to Luna as a whole including jeopardizing the UST peg.

1 Like

I’m aligned with you that we’re attacking the wrong variable - this change may not solve for the congestion problem while potentially creating new problems. However, I hold a less extreme view where I doubt that Anchor will go to zero if this gets passed.

When bot liquidators drop off (bLuna:Luna premium > 15%), we still have Harpoon users (with a long term positive view on Luna) willing to continue liquidating at a 15% gain but choosing to hodl thereafter. Before Harpoon was launched, such users (myself included) could only wait for Terraswap rates to spike before getting a premium on Luna to bLuna swaps. Harpoon effectively becomes a shortcut for these market participants.

Timely article by Stanford of TFL talking about the liquidation dynamics. Of particular relevance

The maximum liquidation premium offered to liquidators should have never been set this high. A back-of-the-envelope calculation can be done for this parameter by defining an acceptable trade volume and then considering the typical spreads on the pair plus the additional tolerance during market stress (or equivalently the maximum allowable LUNA gain for liquidators). This value likely lies in the range of 7~12%. By reducing profitability, cascading liquidations due to the selloff of the underlying discounted collateral occur at a much lower velocity and additionally gives borrowers more time to react.

The data suggests bLuna providers are currently being punished excessively in the event of a liquidation. This is to the benefit of the liquidators who are being “paid” very well for their services.

I would be more receptive of a reduction of the Max_Premium number if it was done in increments of 5% (30% → 25% → 20%), with an assessment of the performance of terraswap rates & liquidations outstanding to understand the net effect. This is a price discovery process, but going too aggressive in reducing the premium could result in a large negative payoff for everyone.


I beg to differ on this - as the time runs down, people who want to vote may see that the response is overwhelmingly “Yes” and conclude that it’s a no-brainer and select “Yes” without much thought.

Hopefully some disagreement shown in the bar would encourage ANC voters to come to the forums and make an informed decision.

1 Like

That is a very scary proposition to have the fate of Anchor Protocol depend upon the work of something developed by Harpoon Protocol. That is an unacceptable risk. Also, there were plenty of loans available for liquidation that sat idle for long stretches of time during the crashening and the Harpoon guys collectively either didn’t want to perform the liquidation or did not have the funds to do so.

I’m not sure if it’s a good assumption that bot liquidators as a whole are not LUNA bulls. I’m quite confident there’s a mix of pure liquidators and LUNA bulls.

Let’s assume this 15% max_premium_rate gets passed. Upon the launch of the auction mechanism, will this get removed? While I still find the 15% max_premium_rate to be a bad idea, setting an expiration date for the 15% max_premium_rate would alleviate some of the long-term risks of passing this measure.

1 Like

please watch the video, maybe this can help

What sort of fix could there be for Network congestion? As this is only a few months old, the last black swan event happened in 2017, so this “temp fix” doesnt seem needed to me, as the bidding/ques will solve it, tho i do support the 15% premium, just not the ~2k fully liquidated, as to me once harpoon is released, alot of smaller wallets would love to particapate in liquidations, not everyone has 1.5k+ just sitting there so its shortening the market, but i do belive a fix on network congestion is a better option, as we are still early, so if there is a problem now… What about when all the Dapps release with “Anchor Protocol” and people start digging for the source, our congestion is gonna be alot higher then currently? This is the issue we should be sorting out, not a band-aid fix on attempting to lower it, as we will be in this discussion again for organic reasons.

Lowering liquidation premium itself is a temporary solution - the biggest problem here is, currently we are using FIFO for liquidation bidding, so even there are enough liquidation incentives, liquidator bids with max_premium so that they can get more profit while it cost more loss for borrowers.

As we upgrade to orderbook based bidding system, the liquidation will be processed by premium. Please understand this is a temporary solution before we upgrade to orderbook based bidding system.

1 Like

Also taking Aave as an example, most of the cases, their liquidation bonus(which is liquidation premium in Anchor) is around 10~5% in general.

1 Like

Yup, understand that it’s a temporary solution and I’m really looking forward to the bidding system which would truly be an elegant solution to this problem. However, as @ColonelSanders points out, will the limit be raised again after the order book is implemented? That seems to be a good idea as it would allow the market to move more freely, and reduce the risk of liquidators dropping-off.

As mentioned above, I’m actually less resistant to the proposal now after reading through all the thoughtful replies. I’m also more optimistic of the ecosystem as it seems to have become less fragile after the change in circumstances since the dump in mid May.

However, on the balance, it’s still a “No” for me because we could potentially introduce an increased risk of ruin if this is shifted too quickly. While other protocols may provide some kind of precedence for what level is “fair”, each protocol’s dynamics are different, and what seems like a minor tweak can be amplified by 2nd or 3rd order effects.

To me, this feels like we’re trying to steer a car that’s slowly veering off lane by swerving hard in the opposite direction. Taking a slow measured approach might be better since we’re not facing any imminent danger now.

Seems like there are a lot of productive discussion going on here. Just want to point out several things.

  1. Problem of current max_premium_rate is that current liquidation algorithm takes FIFO(First in first out) while process bids. So even if there are two liquidators bidding 10%, 15% premium, if liquidator who bids with 15% premium is faster than liquidator who bids with 10%, protocol process bids with 15% premium.

  2. 15% liquidation is not a harsh number. Even if we decrease max_premium_rate to The liquidators are able to take profit - the oracle price feeds happens every 7~15 seconds. Also, considering the Aave’s case (their oracle price feeding is slower than Anchor oracle feeding) their max_premium_rate is 15% for the assets that has lowest liquidity.

  3. Many community members point out that lowering threshold value will harms small borrowers for network congestion, but its actually not true. Even if loan is under Liquidation_threshold, we only liquidate collaterals upto current loan_amount. So the collaterals are not totally liquidated, it liquidate upto current loan_amount of user. Also, as we decrease liquidation premium to 15%, the liquidation loss for small borrowers is even lower than current situation.

Lets take some examples:

Borrower ‘A’
loan_amount = 1000UST
collateral_amount = 2000 bLuna
collateral_price = 1UST

And if price drops to 0.95 UST, the position value(collateral_value - loan_amount) for each param will be:

liquidation threshold
2000UST 500UST
liquidation_premium 15% 723.5294118 820
30% 496.2406015 660

So if the proposal passes, borrower with small positions suffer from less liquidation loss compare to current conditions. (position value after liquidation 660 UST → 723 UST)

  1. This solutions are temporary solutions. Soon we change current liquidation system into orderbook based liquidation system so that we can avoid FIFO while having large range of liquidation premium that liquidators can get.
1 Like

thank you for taking the time to adress these things.
It should be obvious that not all of the collateral is liquidated, but only enough to pay off the loan plus the premium.
The difference is that 100% of the loan is liquidated and not only a part of it as it happens above the threshold. At least that is how it is described in the docs on Anchor, if I read it correctly.

Regarding your example:

I feel like with liquidation threshold = 500 UST the position value after liquidation should definitely be higher because only about 240 UST plus 30% premium should have been liquidated to reach an LTV of 40% as partial liquidations with collateral value higher than 500 UST (here: 2000 bluna each 0,95 UST = 1900 UST) are liquidated to a safe risk ration of 0,8 (risk ratio being liability / borrow limit. borrow limit 50% of 1900 UST = 950 UST → safe risk ration 950 UST * 0,8 = 760 UST, difference to loan amount = 240 UST).

I don’t know the value of “executionFee” and “terraTax” to calculate the “feeDeductor” according to the docs of anchor, but coming from a position value after liquidation of 660 UST (= 694,73 bLuna at 0,95 UST) in your example the liquidated collateral was more than 1300 bLuna being 65% of the collateral amount with a value of 1,235 UST. This seems way to much for a partial liquidation?
If the difference between a partial liquidation at 500 UST and a full liquidation at 2000 UST ist just about 9% of the total collateral (660 UST to 496 UST)… well okay

so if you don’t mind to share your calculation, I’d be grateful. I’d love to understand it even if it’s subject to change! Anchor is the core project of so many dApps to come and I feel like the Terra Ecosystem relies heavily on it. thanks in advance!

We put up a good fight and lost. Should we have a vote to raise the Quorum to make it harder for this to happen in the future?

Just to chime in, I fully agree with ColonelSanders but maybe for a slightly more nuanced reasoning while also empathizing with borrowers facing a harsh 30% penalty.

The primary difference with anchor and other lending platforms that offer much less liquidation premium is the fact that anchor isn’t collateralized by Luna, but is entirely collateralized by a very illiquid bLuna. No lending platform on eth would allow bluna to be used as a collateral for 100M+ loans because the risk of using a illiquid token as collateral is extremely high since the only market for bluna is the terraswap Luna:bLuna pool with a combined USD value of 3.5M.

I’m not sure how this isn’t a huge concern for TFL that 100m+ in loans are collateralized by 1 token with 1.75M in total liquidity… on a x*y=k amm. Go to uniswap and find any random token with less than 5m in liquidity and see if anyone would be willing to take that as a collateral.

The solution is actually fairly simple, massively increase bLuna liquidity (say by 10x) and reduce premiums. That way borrowers aren’t harshly penalized while liquidators can profitably liquidate loans even with less premium.

The entire defi loan model is built on the fact that liquidators will be motivated by immediately profiting from a liquidation but that isn’t true for a significant amount of loans in Anchor. At this exact moment, when the bluna pool is close to 1:1, it isn’t profitable to liquidate a 2M loan because the swap would put you at a loss. It is not wise to expect liquidators to buy collateral that can’t be immediately sold for a profit.

TFL’s reaction to this entire thing has been absurd. Removing liquidity from the bluna pool (which they already did) and reducing the premium will ensure that less than 2% of anchors loan’s can be profitable liquidated before any rational liquidator will stop if market conditions are bad enough and puts significant risk to anchor’s solvency.