Replace Fee Aggregator auction concept

[nymkappa]

Hi,

I think the Fee Aggregator auction system is broken by design and I want to share my reasoning, so I can get a chance to understand what I'm missing.

Why do we use auctions

In real life, auctions are a great way to value things that have no market value. For example in art, there is no market value for a specific piece. Therefore the most optimal way to value this object is to run an auction and let participants to decide of it.

BTP Fee Aggregator auction

In BTP Fee Aggregator, the auction item in question is the accumulated value of the tokens (generated BTP fees). Those tokens represent actual coins on other chains. Which mean that they do have a very well know market value for everyone.

When an auction can be started, the first bidder should in theory directly bid with the optimal bid, in order to cancel any interest from other potential participant. Most especially in current implementation, there is a minimal bid increase for each new bid. It's currently set to 10% but the value actually does not matter.

Example showing why it's not working

Let's use an example. There are XXX IRC31_PRA token in the FA, worth 100 ICX in total.

• The auction starts with the first bidder
• The first bidder bids 100 / 1.1 = ~91 ICX
• Other bidders now have to bid 91 ICX * 1.1 = 100.1 ICX in order to be the new highest bidder
  • They can buy the same amount of PRA tokens directly on the market for cheaper, so why would they bid in the FA
• Auction is now stuck, and no one will bid anymore

PRA market price movement scenarios

Based on this example, we can see that the FA auction is actually a "first come first serve" system. Now some of you may say that PRA price will move, so maybe it will be interesting for a new bidder to bid again. Let's see the possible scenarios here:

• PRA price does not move vs ICX: First bidder post the optimal bid and wins, no one else participate
• PRA price goes down vs ICX: First bidder post the optimal bid and wins, no one else participate, and the first bidder even looses money
• PRA price goes up vs ICX: One new bidder can participate, but will post the new optimal bid anyway...

So from those 3 scenarios, 2 of them breaks the auction as soon as it starts. I do think this is broken and I would like to offer an alternative, which will generate most likely more ICX for the CPS as well.

Lotterly, an alternative to auction

Using a lottery to sell all aggregated fees allows anyone to participate easily, with any amount they want. If we reuse the previous example, you may bet between 1 ICX to 100 ICX. The more you bet, the better your chances at winning. However, since there is no limitation on the amount, it should attract many people to try their luck. This should therefore collect more ICX overall which will benefit the ICON ecosystem (more ICX going into the CPS or being burned). It's a very straightforward system that everyone can understand and already used at least once in real life (auction rules needs to be explained). It's not difficult to implement too.

If you read all of this, thanks, and I hope to hear your feedback.

[ICONationDevTeam]

Good thinking, as you said there's no way the auction is the right design for a fixed price asset.

The lottery sounds okay, where should the information entropy / randomness of the lottery comes from in your opinion ? I'm afraid the result of the lottery may be manipulated.

[nymkappa]

Maybe by playing with Band Protocol we can come up with some kind of randomness? Price is unpredictable, if we hash it (to make it even more unpredictable) we can use the result as random seed maybe?

We can ask ICONbet how they generate they entropy

[ICONationDevTeam]

ICONBet uses the current block timestamp which is not secured but good enough for a dApp. I don't think such solution would be good enough for a protocol at the scale of BTP.

Ideally the randomness should come from a VRF, I think BandProtocol offers one so I agree it would be good if we can rely on this oracle.

[Skyvell]

Nice job spotting some of the flaws in the auction system!

Going to throw in another alternative solution which can add to the discussion:

Implement a buyback contract:

1. All fees generated by the by BTP are sent to a buyback contract.
2. Depending on which token the buyback contract receives, different logic will be triggered. If token x is received
   by the buyback contract, the contract will execute a buyorder for x/icx pair on an AMM (e.g. balanced).
3. When the buyback contract receives icx back from this trade, it is sent to CPS or burned.

The different incomming tokens to the buyback contract could be handled with tokenfallback method or a similar approach.

This system would create constant buying pressure on icx and result in a price increase. Value would be equally spread to all icx holders through this price increase.

Q: Not sure how hard this would be to implement using cross chain assets?

[nymkappa]

That's a good idea. It may be a bit difficult to implement though. This also require maintenance every-time a new chain joins BTP network, a new pair would need to be added on the AMM. There is also liquidity issues involved. But it's something worth exploring, maybe for BTP 2.0 ? 🚀

[Skyvell]

Thanks for commenting on the feasability of the idea; I don't know enough about the BTP tech to judge it myself. Hopefully I will dive into it in the future. Yeah, pairs and liquidity on the AMM would be two more things to consider. Perhaps too much complexity for a first iteration. Anyway, hopefully the post added some fresh thoughts, and gave you BTP devs some ideas on current and future designs, which was the purpose of the post :)

[nymkappa]

I just found out about this project, leaving it here for later reference. Could be useful if BTP implement your suggestion https://equality.exchange/

[Yeol-yeol]

The more complex, the more errors.
The more complex, the more difficult it is for ordinary people to understand.
How about making it simple?

And auctions, lotteries, etc., these things don't seem to be good. If it were me, I would go to the casino.

Profits from BTP are divided into certain proportions and burned.

For example (proportion adjustable)
CPS:2 , Incinerate 3, Relayer 5

[nymkappa]

@Yeol-yeol since it's a new proposal, I've shared it in a new thread here

[nymkappa]

Suggestion from @Yeol-yeol

The more complex, the more difficult it is for ordinary people to understand.
How about making it simple?

And auctions, lotteries, etc., these things don't seem to be good. If it were me, I would go to the casino.

Profits from BTP are divided into certain proportions and burned.

For example (proportion adjustable)
CPS:2 , Incinerate 3, Relayer 5

[nymkappa]

@Yeol-yeol What do you mean by Profits exactly? If you are talking about BTP fees, then your solution is not applicable because som of those fees are IRC31_XXX tokens. We can't use them in the CPS, they need to be converted into ICX first through a system that needs to be defined (currently it's an auction). This is the topic of this discussion.

[Yeol-yeol]

Sorry. I don't know much about technology.

[BennyOptions]

Great to see this discussion. I originally designed the auction system in order to avoid using an oracle price-feed for such a crucial piece of our infrastructure, but as oracles have become more widely used and are currently securing billions of dollars of assets, I'd actually like to suggest using a simple oracle price feed.

My proposed solution would be:

• Fees are aggregated into this pool
• Fees are sold in exchange for ICX
• The selling price is determined from an oracle feed less a fixed discount (i.e. assets are sold at 10% discounts to ICX holders)

Pros:

• More reliable arbitrage opportunities, as the assets are sold instantly rather than a 12 hour auction
• You can buy any amount of the asset rather than the entire lot

Cons:

• We already developed the auction system
• Relies on a 3rd party service (oracle provider)

I actually think it would be worth considering to change this prior to launch. @jorisvial let's think about it and discuss next week. Although we already spent time writing the SCORE with the auction logic, I believe it will be even more difficult to change this post launch as people will already be using the product and there will need to be significant changes to a live frontend. From my experience, it's always harder to make big changes like this to a live product.

[EncryptedPedro]

If the asset price is always 10% cheaper vs ICX, no one would waste this arbitrage oportunity, so won't this make it so the pool is always empty and the assets are sold as soon as they arrive? Making it a first come first served system anyway? The faster bot to get it wins?

I might be missing something though, I'm not that educated in these matters..

Also, the Lottery seems like a good idea since it can burn 1000 ICX for an asset worth 100 ICX, for example.

However, let's say there are 10 players bidding on an asset for 90 ICX, who are aware that there are 10 players usually in previous lotteries.

A wise bet would be 90/10=9 ICX, so it would be a 10% chance for a 10x which is a 1 R/R ratio.

But turns out that a new player joined and he now bets 8.1818 ICX which breaks the 1 R/R ratio for the initial 10 players and only the 11th player has the 1 R/R ratio. This puts the first 10 players out of business and leave, which seems like a severely negative sum game which will never be rewarding for the majority. Assuming that the players can know about other players ofc, if that's not the case then the entire thing becomes even more gambly and would be an even bigger negative yield. I think, right?

This would absolutely fuck the players in turn for greatly benefitting the network, since it would always be a losing game because people would join and leave all the time, until it reaches a point where so many left that it is now profitable again to join, thus creating a savage way to perfect balance, until it falls out of balance again, and repeat cycle.

But wouldn't this make the losers (the majority) angry and express that anger towards ICX? Technically, this would be an anger generation machine.

So, what if in order to participate in the lottery, the "ticket" price would be a static 90% of the asset's value and a winner is picked at random, while the losers are refunded? Sort of an hybrid between a lottery and an auction.

This would make many people want to participate in a positive yield game, which would yield less as more players join but in the end, it will always be positive yield, until it's no longer worth the electricity spent having the bot running, which would then balance out as players leave.

But, what prevents someone from just buying 1000 tickets (1000 addresses) (would require 1000 x 90 ICX = 90,000 ICX), the odds would always be in his favor and he wouldn't be punished for "cheating" since he would be refunded on the losing tickets. However, if this strategy became popular, it would create indefinable demand (or lack of supply*) for ICX and the game would then be Pay 2 Win, where the whales always win while fighting each other, which I'm not bothered by honestly.

• the lack of supply would be like staking, since those 1000 addresses of 90 ICX would need to be constantly bidding while the bot is running, kinda "locking" the ICX.

Now, imagine if the ICON Foundation provides a public, legit, easy to use bot and the fact that a guaranteed 10% profit by using this bot to bid in BTP fees would become mainstream, it would make it easy for participation in the bidding war and thus bringing more whales to use the 1000 tickets "cheating" strategy, bringing insane demand for ICX, since it is a guaranteed positive yield game and greed is king in crypto.

It would be even better if people could bid with staked ICX, making it way more appealing because they wouldn't be missing out on staking rewards. But I think this wouldn't be possible at all unfortunately, unless there's a work around, by leveraging sICX or something, dunno.

Either way, this would capitalize on people's greed (riskless profits) for the benefit of the network.

I'm not even sure if the above suggestions are possible to execute at all but I just figured I'd put my perspective out there anyway.

PS: I spent 3h writing this and didn't even notice.

[nymkappa]

I'm gonna try to read this more carefully and give a more complete answer but I just wanted to comment about the "angry machine" idea you've pointed out.

Many government runs official lotteries. Odds are usually very very light (1 on millions) yet people keep playing. It's extremely lucrative for government because of course they don't refund loosing tickets. The main difference is that odds are fixed vs the lottery I've proposed.

So you've mentioned this "floating odds" problem. I believe it can simply be fixed by allowing user to bet more ICX before the final draw.

[BennyOptions]

"If the asset price is always 10% cheaper vs ICX, no one would waste this arbitrage oportunity, so won't this make it so the pool is always empty and the assets are sold as soon as they arrive? Making it a first come first served system anyway? The faster bot to get it wins?"

Indeed that is the case, but I think perhaps we are trying to solve two different problems. I do not mind if bots run the show. That's not a bad thing to me personally, it means it will be a smooth and efficient market, with bots/traders buying ICX from exchange to be prepared to purchase assets at a discount from our chain.

I was more-so trying to remove the unnecessary complexity I introduced with the auction system as a way to avoid using oracles. Now that oracles are more widely accepted, I'd gladly simplify the process/development.

[gitdiff75]

In financial systems, most bots run the show when it comes to arbitraging. It is a vital part of the ecosystem. I'd like to see more documentation on the bot creation part so that more people can participate in it. It'll be healthy to see more bots in general from different people - would help to ensure the network stability in case one region fails or one participant with many bots no longer wish to participate in this activity.

[shikamarumind]

The main reasoning behind bots in financial systems is to take advantage of arbitrage opportunities, which in theory allows the market to be as efficient as possible, eliminating asset mispricing. Arbitrage bots are essential in this system due to an "order book" structure on exchanges. Through the use of bots, advantage comes from not only available capital but also speed of transactions.

Scott's proposal below is for a syndicate book alternative. People can still use bots to issue IOI's in this new system, but speed is no longer a key element in this structure. Arbitrage will still occur, but there isn't a technological bottleneck for participants.

[BennyOptions]

I have been thinking about a way to achieve:

1.) Effective arbitrage opportunities
2.) Access to effective arbitrage opportunities for all rather than only bots

The current auction system does not achieve either one of the aforementioned goals. It is serviceable, but far from ideal, and I'd like to change to something like this prior to launch if reasonable.

I believe the below solution will achieve these goals @jorisvial. It is very similar to a syndicate book from my previous role in banking, so I'll call it the BTP Syndicate for now. I'll walk through an example then explain.

Imagine there is 100 ETH from fees in the FA SCORE. And imagine the current market price of ICX/ETH is 10 ICX per ETH. The first step is collecting "indications of interest" over a set period of time, let's say, 24 hours.

• Bob deposits 50,000 ICX as an "indication of interest"
• Alice deposits 10,000 ICX as an "indication of interest"
• Charlie deposits 40,000 ICX as an "indication of interest"

The total indications of interest at the end of the 24 hour period are 100,000 ICX.

• Bob represents 50% of the IOIs
• Alice represents 10% of the IOIs
• Charlie represents 40% of the IOIs

At the end of the 24-hour period, the price of ICX/ETH is still 10 ICX per ETH. This means the BTP Syndicate is ~10x oversubscribed. From here, the BTP Syndicate will allocate the 100 ETH to Bob, Alice and Charlie, pro-rata, based on their IOIs.

• Bob receives 50 ETH
• Alice receives 10 ETH
• Charlie receives 40 ETH

The price that is paid will be a simple calculation. At the end of the 24-hour period, we will pull an oracle price from a BAND oracle and discount it by 10%.

• Bob receives 50 ETH for 450 ICX (10 ICX per ETH * 0.9), the rest of his ICX is returned to him
• Alice receives 10 ETH for 90 ICX (10 ICX per ETH * 0.9), the rest of her ICX is returned to her
• Charlies receives 40 ETH for 360 ICX (10 ICX per ETH * 0.9), the rest of his ICX is returned to him

This may sound complicated, but from a UX perspective it will be quite easy. You will visit the BTP Dashboard, select an asset, then submit how much ICX you'd like to spend on this asset, where your purchase price will be guaranteed to be 10% off market price at time of purchase.

If the BTP Syndicate is oversubscribed for an asset, all those that submitted IOIs get their pro-rata portion based on their IOI. If it is not oversubscribed, you just pay exactly what you indicated with the 10% discount off market price.

[BennyOptions]

Point 2 was not about whales vs smaller holders, but rather, developers/bots vs non-technical ICX holders.

I have no issue with whales. Discouraging whales is equivalent to discouraging large ICX holders - of course the goal of all ICX holders is to encourage people to buy and use more ICX.

Additionally, like you said, all this would do is further narrow the group that can take best advantage, because whales with development skills will split their ICX across multiple wallets and run a bot to maximize their IOIs. They would be able to capture more than whales with no development skills.

[sulaaardit]

Great thread- learning loads!

Just wanted to add my part on how to make these arbitrage opportunities more fairer.

How about dividing the FA score -for example 50% for developers/bots and 50% for non technical holders. This way you can force browser only transactions for non technical holders which will ultimately stop the bots. The ratio can be voted in by the community. 60/40 for example.

[shikamarumind]

Whether or not a user is able to take advantage of the BTP discounted assets for arbitrage, ICX holders will always benefit. The ICX used to purchase these assets will still be sent to the CPS and eventually burned, no matter who purchases the assets.

"Fair" can be a relative term, and technically a disproportionate allocation towards whales is still "fair" in my opinion because they took a larger risk of owning such a vast amount of ICX. This risk allows for them to take better advantage of the arbitrage opportunity, and they are rewarded proportionally for this risk. I think normal-users can still be creative with strategies if they want a better portion of the allocation.

[nymkappa]

I like the idea overall because the rules are clear and easy to understand. The UI would be easy to craft as well. That being said, this limit the amount of transaction and ICX collected compared to a lottery system.

[Primo009]

The btp syndicate is the way to go in my opinion previous auction model will just benefit bot participant long term and eliminate non tech savvy icx participants.

Just a few though on the syndicate model hoping it can help:

1/ As stated earlier it will be an endless loop of the most capital (ICX whales) constantly making the most of this model. Participant with low icx amount will not even deem worth it to participate as what they will receive will not cover the opportunity cost of committing (indicating their interest) capital for 24h.

2/ In order for this to be a real opportunity for every ICX holder I think there should be disincentives in place to avoid whales always getting the cake and the cherry on top as the 10% discount price is already significant

3/ Disincentive method that I can think of right now are:

• having a longer commit "indication of interest" (kind of a soft lock period) time-line relative to the auctions frequency so that the same whales cannot jump from one auction to the next one with the same capital
• creating different time-line auctions from longer ones to shorter ones to have whales consider opportunity cost

I don't know what is the scope of design for this model but this btp auction can really be a rallying tool for the icon community so let's try to make it as much so that it empowers every iconist as it is already highly beneficial to whales