An assessment, first hand experience, and recommendations from the Aave Offboarding

Sentence Summary

Discuss the effects of the current asset offboarding process and offer recommendations to improve future situations. Improvements would be aimed at reducing the potential financial impact to the vault user while ensuring timely execution of the asset offboarding as voted on by Maker Governance. The current assessment is based on the percentage of vaults going to liquidation as its benchmark.

Personal experience from myself, a user of Maker, will be provided as to the financial effects the current process has had.


With the recently approved executive proposal, Aave, an asset class that I personally minted DAI through a vault was offboarded. The reasoning behind the offboarding was listed on the Signal Request as:

The COMP, AAVE, and BAL vaults have very attractive rates (1% SF) compared to competitor platforms and have had ample time to build up a user base. However, no usage has materialized. The stability fees generated are quite poor (see below) and don’t come close to covering the Oracle gas costs. As such, I’m recommending the community offboard the COMP, AAVE, and BAL vaults.

Token Dai Minted Annual Stability Fees
COMP 13,722 137
BAL 171,470 1,715
AAVE 1,065,355 10,654

Oracle costs are commented at being $150,000 per year per Oracle, which is magnitudes cheaper than V1, which would have cost about $1.5M per year per Oracle.

The Signal Request passed a vote of yes to offboarding all collaterals, except MATIC. Subsequently, a Maker Governance proposal was listed on November 19th, passed on November 20th, and was executed on November 22, 2021, with the following effects:


As per this poll, the following parameters changes will occur to offboard AAVE-A if this executive proposal passes:

Lower the Liquidation Penalty (chop) to 0%.
Lower the Debt Ceiling (line) to 0 DAI.
Increase the Liquidation Ratio from 165% to 2100% over 30 days using dss-lerp.

Since the execution of the proposal, 14 Aave vaults have gone to liquidation as a result of the increasing liquidation ratio. These 14 liquidation events have collected 1,133,690 DAI.

Time Lapse

  • October 25th: NikKunkel initiates a Signal Request for Offboarding MATIC, COMP, AAVE, BAL with polls closing November 6, 2021. Offboarding off all collateral is voted yes, except MATIC.
  • November 19th: the executive proposal is released to Maker holders for voting
  • November 20th: the executive proposal passed by 10 supporters
  • November 22nd: the executive proposal becomes governing and the ratio begins increasing
  • November 22nd: four Aave vaults go to liquidation
  • November 23rd: five Aave vaults go to liquidation
  • November 24th: five Aave vaults go to liquidation


Liquidation Debt (DAI) Collateral sold (Aave) $ / Aave Total Collateral Remaining
29 9475 37.87 250.20 119.58 81.71
28 13040 52.12 250.19 163.46 111.34
27 24177 95.97 251.92 286.41 190.44
26 17591 70.08 251.01 200 129.92
25 203647 807.15 252.30 2279 1471.85
24 4650 18.31 253.96 48.7 30.39
23 15073 57.72 261.14 150.6 92.88
22 15587 58.9 264.63 143.95 85.05
21 120052 454.05 264.40 1072 617.95
20 237376 888.8 267.07 2107 1218.2
19 15678 59.15 265.05 119.4 60.25
18 273066 1030 265.11 2078 1048
17 172263 655.67 262.73 1185 529.33
16 12015 46.19 260.12 78.85 32.66
Total 1133690 Total 4331.98 Avg 258.56

With a total of 1,133,960 DAI being collected from the auctions and the Signal Request stating 1,065,355 DAI (as of Oct 25th), I think it’s safe to assume the 511 remaining DAI (shown on Makerburn) is minted from a single Aave vault. Based on the total DAI collected and not having an accurate snapshot of vault numbers at the time of proposal execution, I am making the assumption that 14 of 15 (93.33%) vaults have been liquidated.

From those 14 liquidations, the volume of tokens accounts for roughly 0.03% of the total Aave circulating supply. Maker’s Aave risk assessment highlighted that the token has relatively low DEX volume and has a higher severity of negative pullbacks. I wouldn’t doubt the recent volume had some negative effects on price, but won’t speculate further.

There was a comment of bringing these vaults back for the community once we get to L2s, but based on numbers you may have just scrubbed a good chunk of the Aave/DAI community through liquidations.


I’ve been interested in and following Maker since 2017. I’m an active crypto user, don’t have Twitter, read Reddit, will check in on the forums and Governance page semi-frequently as I was a MKR holder at one point so I retain the habit. I’m proud to have a four digit ETH-A vault. I was on the Black Thursday conference call when there was a scramble for keepers. I remember contacting someone on the old MakerChat but ultimately didn’t have the technical skills to get myself set up. Instead I supported the way I could, which was minting DAI via ETH in uncertain times. Eventually, I switched to using Aave as my collateral, even prior to the SF decreases. Ultimately, I was proud to support both of my favourite projects, but now I get to identify myself as one of the 14 vaults to be liquidated.

How did we get here? Quite quickly and with little notice. I was away from my PC for the weekend followed by two days of sickness, yet had confirmed Defi Saver (DFS) settings before disconnecting. With DFS enabled and my phone alert notifications set for the required pricing, I wasn’t too worried as I had even checked forums for potential issues. Even today, if I were to use the search function of the Maker forum it wouldn’t be until the 30th result that the Signal Request for Offboarding Aave appears.

With minimized worry, I was away for the three days followed by one day of illness and another for sleep. Upon recovering on Wednesday morning, I went to check what the DeFi automation settings would be with the newly accrued SF to have the “Claim” button being shown. After confirming it wasn’t a UI glitch and speaking with Nikolaj of DFS through Discord, I learned that “the DFS cannot react to [offboarding processes] as the system cannot recognize this “next liquidation ratio” as it keeps going up.”

I can attribute a hard loss of 202 Aave or 53,964 DAI (as per the closing auction price) from what I deposited to what I received as a result of the liquidation process. The opportunity cost of that loss will only continue to rise, especially given Aave V3 is slated for updates at the end of the month. Given the opportunity of what I would deem a ‘reasonable time for users to react’, I would have shifted my vault from Aave to ETH or Matic (although the handling of Matic’s potential offboarding and what appears to be confusion of where the relationship stands gives me extreme doubt on that action).


#1 Forum search needs improvement
Even today when I search “Aave” I don’t find the offboarding discussions until the 30th result. Shame on me for not diving into the Signal Request sub forum and checking in on the Governance Dashboard every couple days to ensure I didn’t miss something, but in my defense, I thought I had done enough due diligence as a user especially as I was no longer a Maker holder.

#2 Use methods to decentive vault usage prior to complete offboarding
I believe other methods of “soft” shutdown should have taken place first prior to official offboarding of the asset. The Stability Fee was decreased to incentive DAI being minted, why was the same strategy not used in reverse to get vault owners to either close positions or switch vault types once it was deemed the asset was no longer a viable vault product. Even a week or two of higher stability fees paired with a communication campaign would most likely garner much better results than 14 out of 15 vaults being liquidated.

#3 Increased time between execution and introduction of increasing the liquidation ratio via dss-lerp
The Maker holders and users both need to be considered and treated as separate entities in the offboarding process. After an executive proposal receives voting approval, Maker holders receive a 48-hour window prior to execution. As a user, I believe a similar window of time should be granted to me starting at the execution of the proposal and is ultimately a delay on the liquidation ratio beginning to increase using dss-lerp. I believe MakerDAO owes it to the community to ensure there is a timely and orderly process for spinning down end-of-life products. Pair this time with recommendations from #2.

#4 Adequate notice prior to the introduction of increasing the liquidation ratio
If #2 and #3 were to be implemented, it would allow proper time for communication to be circulated regarding the upcoming offboarding. An example of a complete lack of information, there are no posts on Reddit regarding the delisting of Aave within the Aave or Maker subreddit. The only post noted in the Maker subreddit regarding these pending changes was simply titled “Executive Vote: November 19th” and was posted 45 minutes after the vote was passed.

#5 Pre-defined criteria and requirements for signaling offboarding of requests
From signal requests and comments, it appears the current business model is to have only vaults that are profitable right now. Through my reading of comments, I found that Oracle costs have decreased by a magnitude of 10 between V1 and V2. With that fact, plus ongoing discussion of the future of L2s would it not be best to look past the short term? It seems the criteria given for the offboarded collaterals was “not profitable” but what were the business goals originally? Is $150,000 annually the goal for all collaterals then?

#6 Cooldown on Signal Requests for offboarding collateral
With my Aave position being closed and seeing Matic being voted no on the original signal, I was thinking of opening a Matic vault. I’m glad I didn’t as I found it somewhat surprising to see Matic up for the second round of offboarding voting, especially given what seems to be confusion between the teams in the comments. Given the current speed in which collateral can be offboarded once approved, I would say the trigger which activates that process should be used with great reservation; also continuous voting on a topic reduces confidence in the system and increases voter fatigue.

#7 Services
Much like the need for easier keeper systems, there needs to be an easier way for alerts to reach users that want them whether that comes from within the Maker system or community. I feel I had done my reasonable due diligence yet was caught flat-footed. Nikolaj from DFS stated, “We tried posting about this in our discord, and we did try sending an in-app notification, though neither of those is really optimal obviously”. I’m not technical enough to know the roadblocks, but a text via Twilio of “Hello vault owner. Guess what we just changed…” after preregistering for the service is something I’d pay for.


The process to offboard Aave as an end-of-life product has been handled very poorly. The speed of execution resulted in a lack of adequate time to effectively communicate to users through the available channels. Ultimately, the current process of offboarding has caused 14 of 15 Aave vault users to experience financial repercussions which could have been easily mitigated. I’d love to see this end with my financial losses from this being resolved but ensuring the lessons are learned is vital, and I’m able to provide some from the user’s perspective.

The final summarizing thought would be that Maker acted pennywise for its own expenses but pound-foolish with users’ funds.


How did you lose 202 Aave? Its just repaid your debt, and if you look market price right now you are in profit if you buy 202 Aave back.


Perhaps he’s discussing the taxable event the forced liquidation created? Otherwise, I also don’t see how he encumbered a loss.

How am I to buy back the 202 without the ability to incur debt?

Essentially forced to close a long position when at a loss, plus a taxable event.

Edit: Additional time would have allowed users to move positions to a similar risk asset within Maker, like Matic, or gain more room through going to ETH without incurring the forced liquidation/close

you were already in debt. Even if you would move it you would need to repay debt for that vault. So you can just open new vault, borrow 53,964 DAI and buy back AAVE, you will have same debt as before and all your AAVE or even more now that maket dropped, so its not hard loss as you said, since you were already 53,964 DAI in debt, you are now just debt free

1 Like

At the end of the day, I was able to take advantage of the dip to recover, partially as I wasn’t sleeping due to responding to replies here, but ultimately my financial losses being resolved weren’t the purpose of the post. If you’d like we can move on from that point and focus on discussing improving the offboarding process. I’ve moved my funds to the Aave platform for now, but I’m not going away and would like to move back eventually someday. I’ve been saved by Maker features before like the delayed Oracle feed vs Chainlink and do love the platform.

1 Like

First off I want to applaud the effort that went into making this post. It is not uncommon after a bad experience to give up on something in disgust. Kudos to @richards86 for taking the time.

I think we need to focus on the broader picture presented. The case presented shows that there is very little time from the time a signal request is created and until it effects end users. I believe Maker has a responsibility to give a fair warning when changes are made and especially when those changes is to the detriment of the user. I also think it’s a fair expectation of users to have that they should have some time to react and adjust their positions accordingly. If we don’t we run the risk of damaging Maker’s reputation and make conditions worse for future onboarding projects.

Given that this change was mainly a economic one I think we should take a lesson and allow ourselves more time the next time we plan similar changes. I think we need to accept that any onboarding will carry a certain risk that the given collateral type might be unprofitable. As is suggested (#5) I think it behooves us to have clear criteria that will lead to offboarding and a time table of when such an event will occur. This will help make us a more predictable actor and predictability builds trust. I am aware that taking this time may very well make us run at a loss for a while.

I think we should start by taking all the suggestions under serious consideration and my initial stance is that we should implement all of them.


I actually think it’s moderately important to deal with this point (in addition to the others, which are valid). Not specifically in relation to you, but as a general case. We really really don’t want users coming away feeling like they’ve taken a financial loss due to some arbitrary action taken by governance.

I think there is some confusion since the common PoV here is that if you’re liquidated at ~market price with no penalty, then that shouldn’t be counted as a financial loss. However, assuming someone has levered up on their position, I don’t think that is strictly true, since you might be forcing it closed at a loss for the user.

Is this what happened in your situation? If you’re not willing to share that’s completely fine, or if you want to share privately feel free to reach out to me on discord.


However, assuming someone has levered up on their position, I don’t think that is strictly true, since you might be forcing it closed at a loss for the user.

That would be what happened, as you explain. I can certainly pull more extensive records after I go down for a bit of a rest, but essentially it feels like I just had a “WSB maneuver” performed on my portfolio of buying high and selling low.

Liquidation closed me at $265 with the closest loss being Boosts at $276 and $299. For me, through happenstance, it would appear I should be made okay but there are certainly other vault users who might not have the same outcome.

1 Like

I agree with almost all your recommendations.

In general when services are being shut down or change substantially there needs to be a migration period before the changes are enacted onchain that allows integrators and users to react. This is something I pointed out in the Vox MIP as well and something we did during the transition from SCD to MCD.

I think a fix to especially #2, #3, #4 could be to implement e.g. a one month delay from when the offboarding was ratified to when the actual offboarding begins. In order to disincentivize new Vaults the debt ceiling should be set to 0 immediately, and then after a month the gradual LR increase should be engaged. That should provide a clear deadline and give users and front end providers some extra time to react.

However for that to work we also really need someone to take over the official MakerDAO twitter and blogs to ensure deadlines are communicated well in advance and as wide as possible as we saw it with e.g. the migration from SCD to MCD.

For #5 this is something we will dive into in CES to try to define a clearer offboarding process. It’s very important to have this clarity for all stakeholders to make sure expectations are aligned between users, collateral providers, and the Maker Community.


and then after a month the gradual LR increase should be engaged.

Yes! While the proposal states 30 days for the rate to rise; essentially all vaults providing any substantial DAI were liquidated in 3.

Agree with @lollike that suggestions 2, 3, and 4 could be improved upon. Curious if you feel carpet posting on reddit is the best approach? Just to start the gears on communicating offboarding in the future it would be great to hear your perspective on meetings users where they look for updates.

And this is a cool feature request:

Okay, I did a breakdown of my vault and would be happy to provide further detail to someone in private, but it starts with liquidation 18. This resulted in 1030 Aave being sold off to cover a DAI debt of 273,066. A neat fact I just learned, I was the largest Aave vault holder to be liquidated.

Action Purchase Price Total Aave Purchased Loss @ Liquidation
Bought $ 303.21 82.41 $ 3,139.82
Bought $ 307.90 162.39 $ 6,948.67
Bought $ 387.90 64.45 $ 7,913.82
Bought $ 319.37 156.56 $ 8,494.95
Bought $ 309.00 161.81 $ 7,101.84
Bought $ 314.52 158.97 $ 7,854.71
Bought $ 297.92 329.89 $ 10,823.69
Bought $ 322.60 154.99 $ 8,910.38
Bought $ 299.98 250.01 $ 8,717.85
Bought $ 276.14 191.49 $ 2,112.13
Bought $ 384.76 167.35 $ 20,023.43
Average $ 328.84 1030 Aave sold at liquidation
979.64 Aave costs accounted for
50.36 Aave to apply average cost to
$ 57,543.19 Loss from known Aave purchases
$ 3,209.44 Loss from applying average cost to 50.36 tokens
$ 60,752.63 Incurred loss (229 Aave tokens at liquidation price)

I recorded the in and out history of my vault, removing duplicates such as boosting then selling the same amount for a profit to pay down DAI debt. After removing the duplicates, I’m easily able to trace back the purchase of 979.64 Aave tokens with a purchased value of 57,543 DAI. Without a deeper audit, I’ll just assume the remaining 50.36 tokens were purchased at my average purchase price of $328. This would add 3,209 to the 57,543 for a total of 60,752 worth of realized loss due to the forced liquidation. A little more than my napkin math last night as I didn’t go as far back and would have ended up calculating a lower average purchase price.

I can certainly tell you I was happy to learn that the Maker liquidation penalty was not applied, but there was already one built-in for most vault users that were liquidated given recent market conditions. I’m happy to provide further specifics if questioned via PM.


As I was about to reply here, I did notice some discussion is taking place over on Reddit right now. Thanks to PaperImperium for starting that post.

Overall, I don’t believe carpet posting on Reddit is the solution, but a pinned post in the Maker forum with either pertinent information in the title or at least posted in the body. If it’s information of utmost importance, it will most likely get picked up from there and regurgitated through content creators which is why I like @lollike proposal of a month as it gives time for the media cycle to get the message out.

The post I mentioned with the title “Executive Vote: November 19th” is simply a link to the Executive Governance page, nothing else. Every click added before a user gets somewhere will result in fewer users reaching the intended material. A standardized method of posting like:

Titled: Executive Vote: November 19th

  • The WBTC-A Stability Fee will be increased from 2.5% to 4.0%.
  • Local Liquidation Limits (ilk.hole) for ETH-A, ETH-B, ETH-C, WBTC-A, WSTETH-A will be increased.
  • P1-DROP (Peoples Company Series 1) covenant will be changed.
  • BAL-A, COMP-A, AAVE-A will be offboarded.
  • Changes to LRC-A, BAT-A, ZRX-A, UNIV2LINKETH-A for collateral offboarding.
  • WBTC-B will be onboarded as a new vault type.

Would you like more information? Read here.

I think any coordinated communication strategy through the major channels would be an improvement, but focusing on initial basic engagement on their platform of preference is key. As to which platforms, I’d start with whatever ones are listed on MakerDao’s website so Twitter, Reddit, and Telegram then add-on services like Defi Savers notice to users. Regardless, I do believe the best strategy will include a direct-to-user channel. The Ethereum Push Notification Service at first glance is what I had in mind when I made my Twilio comment.

Edit: If something like the EPNS was to be put in place, community content managers and front-end services (like DFS, Makerburn) should be able to subscribe to those notifications in addition to users. That way all parties receive the same message and can begin their tasks to prepare for the pending change. i.e.

  • Content managers post standardized statements / information notifiying users in their preferred platform
  • Front-end services can begin a notification plan for any affected users
  • Users receive notification immediately and can create alternative plans of action

Quick look at the chain of actions would be something like:

  • Signal request approved for offboarding
  • Governance passes vote
  • Governance is enacted, 1-month timer begins, debt ceiling of asset set to 0.
  • EPNS activated and notifies affected users + front-end services + content creators
  • Timed soft measures to shutdown begin with subtle SF rate increases
  • Scaled LR rate increases takes effect and perhaps more subtle between vault minimum and 400 then scaled after that?
  • Outstanding vaults are pushed to liquidation

I agree to the point made by OP, particularly on the buffer period between execution and increasing liquidation ratio. It seems like Maker “can’t wait to offboard the collateral”, while in fact, a buffer of few days to 1 weeks will be of minimal impact to Maker, and sufficient to avoid this. I believe having this buffer period is a good trade off to users’ experience and and also users’ sentiment to Maker.

From the OP’s description, the cdp that is under OP’s care is a “high quality” cdp, meaning well taken care of. Had it been a few days of buffer, I believe many users including OP would have noticed the change in min. collateralisation ratio and acted accordingly, hence avoiding the unnecessary liquidation.

Moving on, for future collateral offboarding, a buffer period will be nice to have, and also some official announcement made on this to get the message to end users

do you think maker should get one core unit to do governance communications

1 Like

Perhaps not to that extent. It only requires admin of twitter/reddit/discord to make an annoucement, and this offboarding is not happening frequently also

perhaps a core unit that produce content and makes the announcement on that channels?
that would help?


1 Like

Hey, I produce reviews and summaries for Governance Communications posted on Forums, Reddit, and Twitter on a very recurring basis. However, this situation shows areas for improvement, which I will look into and act on recommendation #4. Regarding recommendation #1, using search results may result in older, more popular posts to overshadow newer posts. That is why we have weekly reviews like the Maker Relay to help aggregate all the weekly information within forums and governance into one post. But we are always open to feedback and ideas.