The Fiat Peg Problem

Nagalim 2015-10-12 12:43:15 UTC #1

Nubits causes a great fluidity of digital cash, unburdened by many things very much including KYC/AML regulations. It is specifically this asymmetry between digital Fiat and cryptocurrency that generates an issue when pegging with fiat:

1 USD != 1 cryptoUSD

This is not really a solid fact, per say, but more an observation. Look at BTC-e codes, here we have a digital token that can be created for $1, yet wind up being worth more than $1. This is because their fluidity is highly desirable. As such, when we peg NBT to USD, we end up inflating the supply more than anticipated to keep the price down. This phenomenon is completely unprovable and conjecture.

However, when KYC/AML regulations are different for different currencies, the crypto price can vary at a different rate for one fiat than another fiat. By pegging to the fiat price rather than the crypto-fiat price, we could potentially enter a continual arbitrage loop in which we attempt to align entire global economies with our price feeds.

I believe I am seeing this on the NBT/CNY pair. Another view of this phenomenon is that of the BTC prices across exchanges:

Solutions

Track an adjusted price. Of course the question is what price?
Use a large tolerance. This is robust but shitty for Nu.

Note that asymmetric tolerances are equivalent with option 1). Also, this problem will look different under the parametric order book.

mhps 2015-10-12 13:00:23 UTC #2

Yes when we say electronic currency USD, we could mean different prices, depends on which issuer and on which exchange. For example this http://www.bestchange.com/perfectmoney-usd-to-btc-e.html shows you the price to buy btc-e USD with PerfectMoney USD. There is often a 1-5% difference. The USD on ccedk is already not $1, because of fees charged by banks or payment gateways. So when we say 1 NBT = $1 we actually mean 1 NBT = $1-on-an-average-cryptoexchange.

Nagalim 2015-10-12 13:06:57 UTC #3

Yes, this is true. The effect is not an issue when pegging nbt to USD because we can account for the difference with supply. The issue is pegging nbt to a different fiat, specifically cny.

huafei 2015-10-12 14:43:53 UTC #4

you can check BTC price of USD and CNY market with http://btckan.com/price/btc

mhps 2015-10-12 14:59:17 UTC #5

That was what I vaguely had in mind when I said

Nu has limited liquidity. There might be ways to corner LPs if you have a deep pocket, or a few seconds in advance to know the price on the feed, or know the LPs are obliged to rebalance…

Hyena 2015-10-12 20:33:18 UTC #6

We could also make the spread larger, that should take care of any exploitation possibilities. Let the free market fill in the gap between the spreads.

edit:
we can also make the spread a dynamic function of time. for example on weekends the spread would be smaller and during the week the spread is larger. that could potentially make exploitation more complicated.

Nagalim 2015-10-12 20:42:31 UTC #7

Yah, that’s solution number 2. It’s most likely the solution I will use, I’m thinking something like 1.5% offset for a 3% spread (2.6% SAF).

Dynamic spread is interesting, but I would suggest making it responsive to trade volumes rather than times of day. More trade volume increases the spread. That would be pretty complicated to implement though, it would require stuff happening on both server and client side.

tomjoad 2015-10-12 21:07:23 UTC #8

Isn’t the simple solution then to abandon NBT/CNY liquidity? Rather than introduce lots of complexity, NuShareholders should only compensate USD pairs. Force users to exchange their CNY for USD through a different channel if they want to cover a USD pair.

Nagalim 2015-10-12 21:30:28 UTC #9

That’s in the same vein as solution 2, which is to lower the quality of our product. I’d rather avoid reducing the use cases for nubits if possible. I’d also like to point to nuriver which is using a wide spread even with a USD peg because the local price of USD on cryptsy is significantly different from the global USD price. If we run away from this problem, we will find it everywhere we turn.

A solution resembling a reactive parametric order book, for example, could potentially solve this issue entirely. It would look like @Hyena’s dynamic spread but with asymmetry such that it is both a floating spread and price based on trade volume and market participation.

I could make an argument that the issue we’re seeing here is precisely what nubits were invented to solve, and that it is working beautifully.

cryptog 2015-10-13 03:09:58 UTC #10

Could your clarify? I think it is very important to spot any practical useful tacit use of NuBit. It seems that you have spotted one. So I am interested in getting to know what you have in mind.

Nagalim 2015-10-13 07:07:32 UTC #11

Basically, we are increasing the free flow of money globally for all currencies. We do this by offering local liquidity to ripple outwards in a kind of echoing effect of stability. People will of course use that service if we offer it cheap, easily to the point where we question abuse (which ends up looking like the one-sided peg that we’re holding on NBT/CNY). If we make the customer pay more for the service using high offset, we will be reducing our affect on the free flow of money in order to take a bigger slice of the pie.

This rippling stability effect is the global beauty of nubits: allowing people to exchange stable currency without dealing with local regulations or economic imbalances. We just need to make sure we’re charging properly for it.

For example, I put my operator fee up for sale at 7 CNY/NBT. It has yet to be touched even though all custodial funds get eaten up the moment they get dropped into the sell wall. It’s not like CNY is worthless, the specific prices we use for the fiat peg matter a whole lot.

masterOfDisaster 2015-10-13 07:54:28 UTC #12

Nu needs to find the balance with a spread that is small enough to effectively have Nu pegged and that can be afforded.
That’s still not sorted out.

And I still believe the USD peg is way more important than the BTC peg - at least as long as Nu has only NBT.
For other fiat pegs it will be similar; EU-NBT will need to keep a tight peg to EUR, but not to NBT/EUR.
The BTC pairs might need a wider spread to be sustainable with sufficient liquidity.

I’ve lost approximately 1.5% yesterday on the bter BTC/NBT pair. If that were to continue, I’d withdraw my funds from the pool. I still waitand hope that others withdraw liquidity and increase the compensation rate for the remaining liquidity
Fixed compensation is nice.

cryptog 2015-10-13 07:55:36 UTC #13

Would it be cheaper than using non decentralized digital FIAT such as OK Pay?

Up until which point can we increase the spread and still be offering a good value proposition, which is a liquid crypto FIAT, is one important question, I believe.

Could you clarify what it means?

Nagalim 2015-10-13 08:01:05 UTC #14

What’s the fee on OK Pay? We’re only charging a 0.4% fee.

I’m just saying that no one wants to buy my NBT for a price of 7 CNY each, but everyone wants to buy them at 6.35 CNY each.

cryptog 2015-10-13 08:04:08 UTC #15

You have lost money because you are offering liquidity at Tier1 which is subject to all sorts of arbitrage risk. But you are compensated for that contribution.
I am wondering if it is possible that soon we would get to a situation in which Nu would not need to compensate any providers on Tier1 just because offering liquidity at Tier1 would give you enough incentives for doing so. Could there be that there would a natural incentive, which would give birth to sufficient competition so much so that it would be profitable to even offer a tight spread?

Nagalim 2015-10-13 08:06:30 UTC #16

The question comes down to why people would bother submitting their liquidity to the network and how does Nu decide who is trustworthy for liquidity information. Currently, those things happen because Nu pays operator fees and so on. We could probably get the process super efficient and cheap, but I doubt we could ever truly reduce it to zero.

cryptog 2015-10-13 09:57:16 UTC #17

0.5% it seems according to this.

1USD=6.34 CNY right now.

You are right. I forgot about that one.

masterOfDisaster 2015-10-13 13:30:10 UTC #18

I’m fully aware of the risks and know why I’ve lost money.
What I tried to say is: with a (too) tight spread compensation needs to be higher than with a less tight spread.
With a sufficiently wide spread one could even make money instead of losing money.
That’s why compensation is required - to compensate for a spread that is too tight to make money from it.

I still wonder why Nu affords high costs for pegging NBT tight to the USD price of BTC.
I know it’s perceived as marketing effort.

With additional fiat products on the road map I dare to ask when and how the BTC pegging will be phased out - that’s one option.
Another option is pegging EU-NBT to BTC/EUR as well as CH-NBT to BTC/CNY.
Keeping only US-NBT pegged to BTC/USD makes no sense in terms of product consistency.

cryptog 2015-10-14 01:06:07 UTC #19

So you are saying that it should be ok even if the peg is a bit looser? If yes, up until which point?

dysconnect 2015-10-14 05:27:12 UTC #20

Personally I agree to put most of our weight on NBT/fiat but there are still many reasons to keep BTC buy walls. There have been discussions and not all community members/shareholders can make up their minds on shrinking the BTC walls; so the parametric order book is the only way out at the moment. That said there aren’t good compensation schemes that are easy to implement and fit well with the parametric order book, which is why I tried to bring up running two instances of NuPool with different spread/compensation parameters. I’m also guilty for dropping out from that conversation but it’s because I know that a large body of shareholders won’t take it as seriously as I do until the last moment.

The best compromise that can both improve costs and decentrality is to offer ALP on Tether/NBT or BitUSD/NBT, which is risky in another way that we’d tying ourselves with competitors. So we talked about similar relationships with BKC; BCE has an incentive to keep BKC at just about 1 USD where people will buy them at 1 USD and BCE can maximize profits. If NBT is the most price-stable medium to purchasing BKC and for reselling unused BKC that creates NBT demand. Liquidity provision would seem unnatural here but a strong impression that 1 BKC = 1 NBT will help us in a similar way that fiat could.