[Signal Request] Adjust minimum bid increase(beg) and lot size (bump) on flap-auctions

Overview:

Over the past month we have seen the burn kick back into action. This has given us some data as to the efficiency of the burn auctions and how the system handles burns over some time. Examining the data it is shown that there is some inefficiencies when it comes to how we buy back MKR. For the auctions in November/December 2021 the average price of bought MKR was 3,62% over market price and the median was 3,47%. Due to a few outliers the median seems to give the more correct picture.

The auctions are by design meant to go over market to compensate keepers for gas cost and the price risk they face by locking in their MKR in a 30 min window. This is achieved by having a minimum bid increase (beg) of 4%.

Comparing this to the data we can see that most auctions are rewarding the keepers close to the current beg. There might be room to change the level of this reward. One way to do so is to lower the beg.

Lowering the beg makes it less attractive to take part in auctions due to gas cost. To compensate we can increase lot size which makes auctions more gas efficient.

Lowering the beg was also mentioned as a possible future change in this signal request

Proposed Changes

I would propose to lower the beg to retain more value to be burned. In order to offset some of the disadvantage I would at the same time propose to increase the lot size.

Expected pros and cons of decreasing the beg:

Pros:

  • More value retained from the auctions to be burned
  • More granularity in bids

Cons:

  • Greater risk and less reward to take part in auctions
  • Risk of auctions having to run multiple times due to lack of kick
Change the beg parameter
  • 2,5% (-1,5%)
  • 3,0% (-1,0%)
  • 3,5% (-0,5%)
  • No change
  • Abstain

0 voters

Expected pros and cons of increasing the bump

Pros:

  • More gas efficient which makes auctions more attractive for keepers
  • Less auctions overall makes them easier to track and could increase competition for them

Cons:

  • Higher lot raises the bar for individual actors to participate
  • Fewer overall auctions could lead to less monitoring available auctions
Change lot to
  • 40.000 DAI (+10.000)
  • 50.000 DAI (+20.000)
  • 60.000 DAI (+30.000)
  • 75.000 DAI (+45.000)
  • No change
  • Abstain

0 voters

Next steps
This poll will close on 2021-12-30T23:00:00Z.

9 Likes

Thanks for the Signal Request Justin.

As a reminder to voters, you can vote for more than one option, so vote for all of the options that you would deem acceptable. Please bear the following in mind when casting your votes:

  • Only options that receive more than 50% of the non-abstain votes will proceed on-chain.
  • If more than one option receives more than 50% of the non-abstain votes we will run an instant-runoff vote on-chain.
  • All on-chain votes will have “No change” as an option.
  • If none of the options, or only “No change” receives more than 50% of the non-abstain votes there will be no on-chain vote.
2 Likes

Congrats on your first Signal Request @JustinCase! :slight_smile:

On a more general view: I would like to see us adopting Liq2.0 for MKR buy back as soon as possible. It hurts those auctions are running so inefficient and are prone to really large discounts by the current mechanics on multiple running flaps.

To you points

  • The costs to participate (gas for the auction, recycling the DAI to MKR) are already quite high and the 30 minute ttl (which was already higher!) makes it necessary to provide quite a big discount. I think lowering the beg is risky, we might loose keepers by doing so and we already don’t have a lot
  • In the current situation I think raising the bump makes sense. Always a bit sad as it locks out small participants, but it will make the auctions a bit more efficient
6 Likes

Yes! Do we know what a realistic time frame is here, and is it a priority?

@JustinCase - Since there allready are limited numbers of keepers participating in these auctions it would be great to see a rough estimate of the average net return (after gas) for keepers in November/Desember just to compare this with the risk they are taking by locking the MKR price for 30 min.

I guess @Derek is the best to ask here. Curious as well.

1 Like

As it stands we’re heading towards increasing the lot while keeping beg at current levels. Increasing lot without also decreasing beg will not do much to address the current slippage, only save keepers gas while denying smaller players the oppertunity to participate.

1 Like

I took it upon myself to collect some gas data for flapper auctions since nov 15th.

Kick transactions 290
Gas used total 20.90982399088353
Avg gas per kick tx 0.07210284134787424
Median gas per kick tx 0.077688644013544

Deal transactions 158
Gas used total 3.697884819272131
Avg gas per deal tx 0.023404334299190702
Median gas per deal tx 0.009165345

Many deals are batched which is why there are fewer deals than kicks and why the median for deals is so much lower than the average.

Median gas cost to kick + deal is 0,0869 eth or ~$350 at current prices.

So effectively we’ve been paying keepers ~2.3% for the risk of locking up their MKR for 30 minutes. That’s one hell of an APY.

If we increase auction size to 50k while keeping beg, this subsidy increases to ~2.77%.

9 Likes

thanks a lot @makerburn for this analysis - what’s missing in the calculation is the recycle cost DAI->MKR - but nevertheless: do you thing go down to 3% beg and 50k bump is a reasonable path forward?

1 Like

I think 3%/50k is better than the current state, and a good short term improvement. There is of course some risk in kicking, but >2% seems like a lot.

Also, I agree with you wholeheartedly that Liq2.0 style auctions would be way better than flap auctions!

1 Like

I completely agree. I should have made that clear in the signal request itself. I view the parameter changes as merely a temporary fix as my understanding is that the more permanent fix (liq 2.0) is some time into the future.

I can see your concern but I don’t share it to the same degree with respect to keepers. As long as there is profit to be made, economic actors will be there to pick up. I also think there is a fundamental difference between the burn auctions and the liquidation auctions in that there isn’t really a big downside to failed auctions. As long as it’s not a permanent failure, a failed auction merely means a delayed auction. The surplus buffer gets to stay a little over limit a little longer before eventually getting bought and burned

I think @makerburn gave a better answer to this question than I am able to give.

3 Likes

Yeah, we’ve been paying close attention to the flap auctions. I second your point that it would be good to get data on the exact number of keepers participating, from previous observations it looked fairly limited, but we should confirm/deny that to determine the level of competition.

As a ballpark, Liq2.0 for flap auctions would take ~ at least a quarter pending audit availability.

5 Likes

Here is the list of keepers that has kicked (successfully) since november:

0x008Ca3a9C52e0F0d9Ee94d310D20d67399d44f6C: 151 kicks
0x489E4054ecDf7E211e790Fa6680A6892b3C76768: 72 kicks
0x067FB6b6ff920e3ed9eeE13808009BC2a6d5d736: 34 kicks
0x00173A1479EFe831EAB2b86e89990a46307746aF: 30 kicks
0xB6D7031957B8c83A614B8FD282444b19Bc8b2f2a: 1 kick
0xa8162ebe6170D30219e0e251FCA451d454C19F98: 1 kick
0x1FfF8e6045092656B7580C27775E2B3a99e9542C: 1 kick

3 Likes

@makerburn I’m afraid I disagree completely with many of the assumptions that form the basis of your argument.

  1. As @ultraschuppi points out, in your estimates of profitability you’ve forgotten about the leg that takes the keeper from DAI back to MKR… they don’t just accumulate it infinitely and never sell it :joy: . That’s a critical omission because hedging is the most important and usually the most expensive part - gas cost (sometimes very substantial) + 0.3% uniswap fee + market impact of the trade depending on its size (again, often very substantial depending on liquidity). Furthermore, this expense is sometimes DUPLICATED on the occasions when a keeper gets outbid due to the market moving, because they then have to unwind the trade they just did and sell the MKR they just bought back into DAI in order to not have any risk on.
    Here is a typical example of a keeper using this strategy - buying MKR and kicking an auction in one transaction.
    https://etherscan.io/tx/0x9a961115be5ef39c5d9c3498e4d64433e5fde6073c4bd59d7923dd60ba8ceb9f
    As you can see that alone already costs 800 dollars.
    But depending on the keeper this obviously varies and is often not visible, as it’s done in separate transactions.

Once you factor all that in, your figure of 2.3% becomes not even 1%. Reducing the beg from 4% to 3% would make it 0, or negative. You don’t have to take my word for this however, because…

  1. There is at this point essentially only 1 regular keeper. That’s because a couple of the entries in your list of keepers are actually the same keeper. For example, 0x008C… and 0x0017… call the same contract. Same goes for 0xa816… and 0x489E… . So basically, there are only 3 people that have ever kicked more than 1 auction. Of them, one stopped kicking regularly a week ago, and the other stopped kicking regularly a month ago. Clearly this would not have happened if you were correct in your assertion that flaps are highly profitable.

This then leads to the situation we had on the 4th of December, where some auctions went for 30% offside precisely BECAUSE there are not enough interested keepers with eyes permanently on this market. So basically, issues are arising because the role of keeper is already too unattractive as a function of the effort and market risk required, yet astoundingly people on this forum are permanently convinced that the way to fix them is to make it EVEN LESS attractive.

On this note, for completeness I should add that,
3) it is a complete fallacy that increasing the bump to 50k would be positive for keepers. The average keeper already faces a very unattractive ratio of variance to edge - making just 100 or 200 dollars on average on the auctions they win, yet standing to lose 1k or more on the (admittedly rarer) ones where the market goes against them, in which case they lose either directly if they have risk on, or through gas fees if another participant dimes them and their kick was for nothing. (Because as I explained on several posts and community calls in the past, there are far more participants willing to be parasitic and dime a keeper, than there are willing to risk their own money and kick. So no surprise there is not much interest on the keeper front). Increasing the size of a single auction would simply make this ratio even more stressful. And yes there would indeed be economies of scale on gas, but as mentioned in point 1) this would be easily offset by the less efficient hedging of such a large clip of MKR, which is not at all liquid.
So increasing bump and decreasing beg is NOT a balanced action, it would in fact be more of a double whammy for keepers. More risk for less edge :stuck_out_tongue:

I don’t deliberately set out to come across combative. Yes it IS tiring that we are discussing the same misconceptions from scratch for like the 4th time in the last couple of years. But that’s part and parcel of governance in a DAO. I also don’t mind that we are in a situation where important decisions could be pushed by people that have never tried to participate in a flap auction, yet set out to analyse them, and therefore may end up doing more harm than good despite meaning well. Again, that is part and parcel. What IS frustrating though, is that as I say I remember being on a community call over a year ago now and talking about this exact topic. Back then I was still a semi regular keeper, and I warned that the incentives of the system were misaligned in such a way that the sweet spot for a keeper to be profitable was vanishingly small, and we would eventually lose most keepers. It is therefore very frustrating that this is exactly what has come to pass, yet people are STILL discussing measures that will make the situation worse still.

To finish on a more positive note… where I agree with everyone in this thread, is that the solution is LIQ2. LIQ2 will address the issue at its root, in a way that endless tinkering with the current parameters never will.

12 Likes

Thank you for a very enlightening post @dmitri.z. As one of the people who did not have the same level of exposure to past discussions of the same type your post has really helped me see a different perspective. I fear with the rapid developments in crypto we’re still going to be discussing some topics over again as time marches on. Hopefully we can teach and guide each other as a community.

My main takeaway from this discussion is that the introduction of Liq2.0 for the burn auctions should be the priority. With current parameters we stand to burn quite a bit of MKR and should they improve we’re likely to burn even more. It seems even more urgent given that so much apparently hinges on a single keeper. That being said I want to offer some counter arguments when it comes to the lack of keepers.

In regards to your point 1.; that describes one possible strategy. It is still possible to do a trade without the hedge in the case where an actor wants to do a permanent switch from MKR to DAI or from MKR to another asset. This ties in that there is no urgency as such for MakerDAO to conclude the burn auctions as opposed to the liquidation auctions. This should provide at least a little market incentive to participate in auctions as an alternative to do straight up swaps on other platforms.

In regards to point 2.; Another factor to consider is that burn auctions have been inactive for a long time and just recently got started with some regularity. Setting up and monitoring is another cost that comes into play. This also creates a barrier of entry for other actors as they need to be able to compete to get to kick. With auctions at 30k and the dust limit of ETH-B vaults at 40k and ETH-A vaults at 15k, there should be a fair amount of overlap between vault holders and actors with the financial capacity to partake in the auctions, so it is possible there are other reasons than profitability holding back the availability of keepers.

4 Likes

Good points @dmitri.z . After digesting your post I think anything that would reduce keeper incentives to watch this market would be short-sighted. There has been more than one occasion where small MKR holders and active community members have had to step in to save an auction that was about to be “stolen” when there were many flaps in parallel.

I think while we wait for Liq 2, the current set up is a reasonable price to pay for orderliness.

2 Likes

Thanks for jumping in with this post. I often have deja-vu on certain topics too.

What worries me is that there are so few keepers in the game. So do you think it’s conceivable that if there is only one keeper left in an auction, it will take it with minimal or 0 bids?

Fortunately, if they bid really low (or zero) , it doesn’t take much MKR to outbid them, which buys some time for a bigger holder to correct the situation.

That was attempted a few times last year, but myself and some other community members were able to stop it from going through.

2 Likes

If there are >30 parallel liquidations, it will happen again.

1 Like

If this is a real threat, it would be good to hear the opinion of the core-unit member responsible for auctions.

Largest keeper seems to be getting MKR from binance, so at least for them Uniswap gas cost doesn’t seem to be relevant…

https://etherscan.io/address/0x008ca3a9c52e0f0d9ee94d310d20d67399d44f6c#tokentxns