[BSIP42] Consider derailing feed price

[abit]

Update: USDT is now trading at around 0.991 USD. Data sources (direct trading pairs):
* https://www.kraken.com/charts
* https://bittrex.com/Market/Index?MarketName=USD-USDT

CoinMarketCap is reporting 0.9968 which seems a bit off. https://coinmarketcap.com/currencies/tether/

[abit]

Putting a Coinmarketcap chart here is only for reference. Afaik no witness is relying on coinmarketcap to produce price feeds. I don't disagree that coinmarketcap sometimes shows inaccurate data. However, from a macro perspective, data showing there is fine.

It's literally quoted as the reference chart and data source in BSIP42 documentation. For the reasons above I cannot agree it's 'fine'.

The chart is more or less showing the real trend, as I said a macro perspective. I think we may have to agree to disagree here. The mention of coinmarketcap in the documentation is in "rationale" section but not "specification" section. Witnesses are expected to make their own efforts/decisions to figure out what an accurate price is.

I guess you as a big bitUSD holder can't accept that 1 bitUSD value far less than 1 Fiat USD. But would you accept that 1 bitUSD value more than 1 Fiat USD? As a big bitUSD creator/borrower, of course I don't want that 1 bitUSD value too much more than 1 Fiat USD. In short, let's be fair, don't play double standards. If we don't fairly treat borrowers/creators, it's hard for bitUSD to have sufficient supply. BSIP42 is looking for fair.

The 'value' of a bitUSD may fluctuate day to day depending on market access to bitUSD- maybe there is a an excess, maybe a shortage and thats intended. As i hold bitUSD, I hold bitUSD, not real USD. So I acknowledge the asset may fluctuate in value vs an actual United States Dollar. If you think its underpriced then you can buy it, if you think its overpriced you can sell it. Sure as an individual stakeholder you are welcome to benchmark vs exchange 1, 2, 3 and so on. Whether it is fair or not is not relevant, its what the market is pricing it as, thats the price. If its differs too much from the price maybe I sell it or acquire more. So feed producers have a vested interest in ensuring its accurate.

The purpose of a price feed is for price feed producers to analyse this and any other variables they see necessary and then produce a price. I would hope there is more diversity in prices from feed producers in future.

IMHO bots WON'T work when there is a supply-side issue (either shortage or oversupply). Using trading bots to try to maintain the peg is the approach adopted by NuBits and IMHO has failed.
Actually there is NO instant opportunity when there is a premium or discount if you've tried running bots. There do have long-term opportunity but risks are much higher, E.G. margin calls.
A slight premium or discount is fine, but IMHO it's not acceptable if it's too big. Personally I think the threshold is +/-1%.

Absolutely not, the mechanism of arbitrage is entirely different to Nubits, that is how they managed their price feed, nothing to do with arbitrage.

IMHO why Nubits failed because their bots kept running at a loss, not due to inaccurate feed, but due to wrong market strategy (e.g. setting too little spread even when they can't afford). They introduced too high systematic risks.

A simple bot can do this:

Bot A: Huobi
Has balance of: USDT, BTS
Market: USDT/BTS

Bot B: DEX
Has balance of bitUSD, BTS
Market: bitUSD/BTS

If price on Huobi is < DEX. Bot A Buys BTS, and Bot B sells BTS. This is done instantly at same time.
If Cheaper on DEX, Bot B buys BTS and Bot A sells BTS.

Over time, one bot will have a excess in one asset and the other will have limited supply and the trader will need to make an intelligent decision on how to balance those assets again in the most efficient manner.

The trader makes a profit depending on what he bought and sold at after fees. This is an easy arbitrage that can be done and its not hard. Whether there is supply of bitUSD or not is irrelevant, that will be the reflected in the price, if there is not much bitUSD then like a properly functioning market the price for bitUSD should be higher.

OK, it turns out that you know how bots work, so we can discuss deeper.

Interestingly you've mentioned "make an intelligent decision", then said supply is irrelevant. How to make decision if there is no supply? In Chinese there is an idiom "巧妇难为无米之炊". When there is lack of supply, bots are simply unable to get into a position, thus unable to balance the market.

This is simply wrong. As mentioned above, nobody actually use coinmarketcap for price feeds. Also OpenLedger's operation has little impact on the prices so far.

As I said it's quoted in the actual BSIP, so yes I'd say it is used as a price reference. It's also referenced through this and other threads. I mentioned OL because as noted, people seem to reference he bitUSD price on CMC and the bitUSD markets of which CMC bases their 'price' of bitUSD on is derived by the bitUSD markets they have access to- which are OL markets (https://coinmarketcap.com/currencies/bitusd/#markets). I have no idea what openKRM or openXEM are, but if their prices spike, then it will influence the prices CMC uses to calculate bitUSD.

Therefore as I noted before, CMC cannot be used as a reference. You need multiple sources and it just so turns out we have people who are paid to do this and they're called price feed producers.

As of writing, on the chart, the main bitUSD markets are bitUSD/bitCNY and bitUSD/BTS which has around 90% of total volume, so OpenLedger only plays a minor role here.

There is no direct FiatUSD/bitCNY pair elsewhere to compare, so we end up with the only relatively easy method which is comparing BTS/bitUSD and BTS/USDT pairs (and maybe need to check USDT/FiatUSD rate as well which is not very volatile). Any way, I think we both know the data can be affected in many ways, so, anyone please make another tool to track the premium/discount more accurately? I've explained methods to measure it, if you disagree, would you like to give some constructive suggestions?

[binggo]

Absolutely not, the mechanism of arbitrage is entirely different to Nubits, that is how they managed their price feed, nothing to do with arbitrage. A simple bot can do this:

Bot A: Huobi
Has balance of: USDT, BTS
Market: USDT/BTS

Bot B: DEX
Has balance of bitUSD, BTS
Market: bitUSD/BTS

If price on Huobi is < DEX. Bot A Buys BTS, and Bot B sells BTS. This is done instantly at same time.
If Cheaper on DEX, Bot B buys BTS and Bot A sells BTS.

Over time, one bot will have a excess in one asset and the other will have limited supply and the trader will need to make an intelligent decision on how to balance those assets again in the most efficient manner.

The trader makes a profit depending on what he bought and sold at after fees. This is an easy arbitrage that can be done and its not hard. Whether there is supply of bitUSD or not is irrelevant, that will be the reflected in the price, if there is not much bitUSD then like a properly functioning market the price for bitUSD should be higher.

if all the Cex gone or disapear or no bts, what will the bots do?

The market will do the bot's work, don't need these any more.

The price Spread DEX and the CEX have a very simple cause: the margin call price----which was decided by the MSSR and the feed price.(don't talk about the force settlement and market sentiment here)

If you change the MSSR and the feed price anyone of them, you will find the change of the price spread.

[George_Bitspark]

Putting a Coinmarketcap chart here is only for reference. Afaik no witness is relying on coinmarketcap to produce price feeds. I don't disagree that coinmarketcap sometimes shows inaccurate data. However, from a macro perspective, data showing there is fine.

It's literally quoted as the reference chart and data source in BSIP42 documentation. For the reasons above I cannot agree it's 'fine'.

I guess you as a big bitUSD holder can't accept that 1 bitUSD value far less than 1 Fiat USD. But would you accept that 1 bitUSD value more than 1 Fiat USD? As a big bitUSD creator/borrower, of course I don't want that 1 bitUSD value too much more than 1 Fiat USD. In short, let's be fair, don't play double standards. If we don't fairly treat borrowers/creators, it's hard for bitUSD to have sufficient supply. BSIP42 is looking for fair.

The 'value' of a bitUSD may fluctuate day to day depending on market access to bitUSD- maybe there is a an excess, maybe a shortage and thats intended. As i hold bitUSD, I hold bitUSD, not real USD. So I acknowledge the asset may fluctuate in value vs an actual United States Dollar. If you think its underpriced then you can buy it, if you think its overpriced you can sell it. Sure as an individual stakeholder you are welcome to benchmark vs exchange 1, 2, 3 and so on. Whether it is fair or not is not relevant, its what the market is pricing it as, thats the price. If its differs too much from the price maybe I sell it or acquire more. So feed producers have a vested interest in ensuring its accurate.

The purpose of a price feed is for price feed producers to analyse this and any other variables they see necessary and then produce a price. I would hope there is more diversity in prices from feed producers in future.

IMHO bots WON'T work when there is a supply-side issue (either shortage or oversupply). Using trading bots to try to maintain the peg is the approach adopted by NuBits and IMHO has failed.
Actually there is NO instant opportunity when there is a premium or discount if you've tried running bots. There do have long-term opportunity but risks are much higher, E.G. margin calls.
A slight premium or discount is fine, but IMHO it's not acceptable if it's too big. Personally I think the threshold is +/-1%.

Absolutely not, the mechanism of arbitrage is entirely different to Nubits, that is how they managed their price feed, nothing to do with arbitrage. A simple bot can do this:

Bot A: Huobi
Has balance of: USDT, BTS
Market: USDT/BTS

Bot B: DEX
Has balance of bitUSD, BTS
Market: bitUSD/BTS

If price on Huobi is < DEX. Bot A Buys BTS, and Bot B sells BTS. This is done instantly at same time.
If Cheaper on DEX, Bot B buys BTS and Bot A sells BTS.

Over time, one bot will have a excess in one asset and the other will have limited supply and the trader will need to make an intelligent decision on how to balance those assets again in the most efficient manner.

The trader makes a profit depending on what he bought and sold at after fees. This is an easy arbitrage that can be done and its not hard. Whether there is supply of bitUSD or not is irrelevant, that will be the reflected in the price, if there is not much bitUSD then like a properly functioning market the price for bitUSD should be higher.

This is simply wrong. As mentioned above, nobody actually use coinmarketcap for price feeds. Also OpenLedger's operation has little impact on the prices so far.

As I said it's quoted in the actual BSIP, so yes I'd say it is used as a price reference. It's also referenced through this and other threads. I mentioned OL because as noted, people seem to reference he bitUSD price on CMC and the bitUSD markets of which CMC bases their 'price' of bitUSD on is derived by the bitUSD markets they have access to- which are OL markets (https://coinmarketcap.com/currencies/bitusd/#markets). I have no idea what openKRM or openXEM are, but if their prices spike, then it will influence the prices CMC uses to calculate bitUSD.

Therefore as I noted before, CMC cannot be used as a reference. You need multiple sources and it just so turns out we have people who are paid to do this and they're called price feed producers.

[sschiessl]

I believe one of the main concerns is lack of readily available information in the approach, which the community desires (IMO). I would also like to see a report in the matter (please don't mention that it's all on chain ...). Without even discussing the content of BSIP42 this is enough for me to think we should take a moment, properly analyze how it affected bitCNY, reflect on and then decide how to move forward.

For example, provide measures over past weeks since BSIP42 is active (and possibly before that timeframe for a comparison):

• old price feed
• effective values of MCR, MSSR and FSO (consider differences in the feeds)
• effective offboarding ratios when
  
  • selling bitCNY for BTS (and BTS for CNY)
  • when settling CNY for BTS (and selling BTS for CNY)
  • offboarding directly with bitCNY (aka the peg)

Those measure are interesting for the common user. I believe you would see that there is a jump when BSIP42 was activated,
and it will show that the values are normalizing over time (which would show that the market accepts the peg, and what the consequences are).
The effectiveness of the BSIP 42 can directly be judged with those measure, furthermore it will be clearer where the premiums are paid now.

All those values measures might be clear to the initiators of BSIP42, but not to the common eye.

[abit]

Ok firstly, as a holder of about 860k BitUSD these are my thoughts on this:

1. This whole discussion is based on a faulty premise using coinmarketcap chart data. Does the OP and others realize how coinmarketcap calculate their price data? Its based on a weighted avg of the exchanges they list and markets trading that asset in addition to an external USD/BTC price feed, markets that are low volume or strange pairings can affect the 'coinmarketcap price' by quite alot when people buy $20 of BTS for $5/BTS it reports a price of $5 for that market. Obviously that should not be taken seriously as a price of BTS as only $20 was traded however it is folded into the calculation of BTS on Coinmarketcap.

Secondly the USD price displayed on coinmarket cap is derived from the relevant BTC/USD price- a subjective metric for us when determining a real price as it relies on one price feed, Coinmarketcap. For example, if the weighted avg 'coinmarketcap price' on Bitcoin is $6500 (and it its subject to the same problems on a larger scale as above) then when calculating a BTS/USD price it takes the BTS/BTC * BTC/USD in order to determine a final dollar price of BTS. Thats two trades, including a totally subjective metric of the coinmarketcap BTC price which determines what they price BTS as. The bitcoin/USD price is entirely out of our control so when looking at a coinmarketcap price for BTS you should take this as a massive grain of salt as its not going to be accurate. Its basically relying on an external feed called coinmarketcap rather than our existing decentralized list of price feed producers on the BTS DEX.

Putting a Coinmarketcap chart here is only for reference. Afaik no witness is relying on coinmarketcap to produce price feeds. I don't disagree that coinmarketcap sometimes shows inaccurate data. However, from a macro perspective, data showing there is fine.

To check how much a bitUSD worth, a simple way is to compare the trading price on BTS/bitUSD market in the DEX (https://wallet.bitshares.org/#/market/BTS_USD) and the price on another BTS/USDT market E.G. https://www.huobi.com/bts_usdt/exchange/ when USDT is around 1 fiat USD. As of writing, majority of witnesses is applying BSIP42 on bitUSD, thus the trading prices on those two markets are similar -- that means bitUSD is pegged well, and coinmarketcap shows exactly the peg.

I guess you as a big bitUSD holder can't accept that 1 bitUSD value far less than 1 Fiat USD. But would you accept that 1 bitUSD value more than 1 Fiat USD? As a big bitUSD creator/borrower, of course I don't want that 1 bitUSD value too much more than 1 Fiat USD. In short, let's be fair, don't play double standards. If we don't fairly treat borrowers/creators, it's hard for bitUSD to have sufficient supply. BSIP42 is looking for fair.

2. Having a price premium or discount is a good thing and incentivises arbitrage and is only natural, its a feature not a bug.

If there is a profitable premium to trade between BTS on the DEX vs the CEX in bear or bull  market then we should be promoting that as an opportunity not trying to 'adjust' things via central planning so that there is no opportunity. It is an exciting opportunity to create bots or trade manually in these markets at these times and if there is a premium, fantastic advertise it and the market will ensure its not there for much longer. Part of our roadmap is adding bots to do exactly this (as we currently do similar things in bitcoin markets on CEXs) I dont see how this is a problem. Yep I get how there can be a liquidity short squeeze, thats the market giving you price signals on their valuation of a decentralized asset. 

IMHO bots WON'T work when there is a supply-side issue (either shortage or oversupply). Using trading bots to try to maintain the peg is the approach adopted by NuBits and IMHO has failed.

Actually there is NO instant opportunity when there is a premium or discount if you've tried running bots. There do have long-term opportunity but risks are much higher, E.G. margin calls.

A slight premium or discount is fine, but IMHO it's not acceptable if it's too big. Personally I think the threshold is +/-1%.

3. BSIP42 does not have Sparks support and we would prefer the following alternatives:

* Better price feeds means voting out price feeds which the market consider inaccurate
* If anything, bitUSD could likely look at adjusting MCR if necessary by +/- 5% and measure the results. If nothing tangible then it can be reverted.
* Parties concerns about price premiums and discounts should actively promote these opportunities to trading community on social channels and we as a community should look towards building tools that can easily take advantage of situations like this, for example bots that can exploit these opportunities. Expanding DEXbot is an ideal use case here.   

edit: Looking at the bitUSD markets on Coinmarketcap their USD price is derived from its almost entirely from Openledgers markets anyway with only about 3/20 markets having sufficient volume to be taken somewhat seriously, the others can as per above skew the accurate price of BitUSD or BTS. If OL have some issue with one of their altcoin markets then this will massively affect the price calc for BTS/bitUSD and feed through to inaccurate prices on CMC. This happens often when Openledger may have withdraw issues for a coin for have delays in issuing it like openEOS etc.

This is simply wrong. As mentioned above, nobody actually use coinmarketcap for price feeds. Also OpenLedger's operation has little impact on the prices so far.

[kimchi-king]

@George_Bitspark Thank you for providing your fantastic input. I couldn't agree more with everything you shared.

[xeroc]

This!

[George_Bitspark]

Ok firstly, as a holder of about 860k BitUSD these are my thoughts on this:

1. This whole discussion is based on a faulty premise using coinmarketcap chart data. Does the OP and others realize how coinmarketcap calculate their price data? Its based on a weighted avg of the exchanges they list and markets trading that asset in addition to an external USD/BTC price feed, markets that are low volume or strange pairings can affect the 'coinmarketcap price' by quite alot when people buy $20 of BTS for $5/BTS it reports a price of $5 for that market. Obviously that should not be taken seriously as a price of BTS as only $20 was traded however it is folded into the calculation of BTS on Coinmarketcap.

Secondly the USD price displayed on coinmarket cap is derived from the relevant BTC/USD price- a subjective metric for us when determining a real price as it relies on one price feed, Coinmarketcap. For example, if the weighted avg 'coinmarketcap price' on Bitcoin is $6500 (and it its subject to the same problems on a larger scale as above) then when calculating a BTS/USD price it takes the BTS/BTC * BTC/USD in order to determine a final dollar price of BTS. Thats two trades, including a totally subjective metric of the coinmarketcap BTC price which determines what they price BTS as. The bitcoin/USD price is entirely out of our control so when looking at a coinmarketcap price for BTS you should take this as a massive grain of salt as its not going to be accurate. Its basically relying on an external feed called coinmarketcap rather than our existing decentralized list of price feed producers on the BTS DEX.

2. Having a price premium or discount is a good thing and incentivises arbitrage and is only natural, its a feature not a bug.

If there is a profitable premium to trade between BTS on the DEX vs the CEX in bear or bull  market then we should be promoting that as an opportunity not trying to 'adjust' things via central planning so that there is no opportunity. It is an exciting opportunity to create bots or trade manually in these markets at these times and if there is a premium, fantastic advertise it and the market will ensure its not there for much longer. Part of our roadmap is adding bots to do exactly this (as we currently do similar things in bitcoin markets on CEXs) I dont see how this is a problem. Yep I get how there can be a liquidity short squeeze, thats the market giving you price signals on their valuation of a decentralized asset. 

3. BSIP42 does not have Sparks support and we would prefer the following alternatives:

* Better price feeds means voting out price feeds which the market consider inaccurate
* If anything, bitUSD could likely look at adjusting MCR if necessary by +/- 5% and measure the results. If nothing tangible then it can be reverted.
* Parties concerns about price premiums and discounts should actively promote these opportunities to trading community on social channels and we as a community should look towards building tools that can easily take advantage of situations like this, for example bots that can exploit these opportunities. Expanding DEXbot is an ideal use case here.   


edit: Looking at the bitUSD markets on Coinmarketcap their USD price is derived from its almost entirely from Openledgers markets anyway with only about 3/20 markets having sufficient volume to be taken somewhat seriously, the others can as per above skew the accurate price of BitUSD or BTS. If OL have some issue with one of their altcoin markets then this will massively affect the price calc for BTS/bitUSD and feed through to inaccurate prices on CMC. This happens often when Openledger may have withdraw issues for a coin for have delays in issuing it like openEOS etc.

[gghi]

        The goal is to kill this premium.   When there is no premium, I think settlement offset can be set to zero.

no, you haven't considered the scenario that smartcoin has discount.

example: when BTS's price is 5bitCNY in DEX, however, bitCNY has 2% discount and because of the negative feedback, the feed price is 4.8.   what will happen if you set settlement offset to 0?

   yes,You're right.

[bitcrab]

        The goal is to kill this premium.   When there is no premium, I think settlement offset can be set to zero.

no, you haven't considered the scenario that smartcoin has discount.

example: when BTS's price is 5bitCNY in DEX, however, bitCNY has 2% discount and because of the negative feedback, the feed price is 4.8.   what will happen if you set settlement offset to 0?

[gghi]

        The goal is to kill this premium.   When there is no premium, I think settlement offset can be set to zero.

[abit]

Poll workers created:

1.14.118 Poll - BSIP42 - Adjust price feed dynamically

1.14.119 Poll - BSIP42 - NO adjustment to price feed

[R]

I really think Bitshares needs to find a way to collateralise with bitcoin instead of BTS. (don't worry BTS will still have value because its need to pay fees)

You can just create another (private) MPA e.g. YOUR.USD backed by YOUR.BTC.

You can't do this through the UI yet, but it's in the works: https://github.com/bitshares/bitshares-ui/issues/1537

Yes we could create bitUSD backed by OPEN.BTC but as a centralised IOU people won't trust it. Hence the need for a trustless BTC / sidechain.

ACCT/trustless BTC (and other cryptos) is a good proposal & on the core roadmap, but people are also seemingly willing to trust entities like USDT which may be using cryptocurrency as backing collateral - if an UIA had the marketcap of USDT with semi-proven backing collateral (1up on USDT lol) then that'd bring a lot of value to the BTS DEX (fees at least).

What would be cool would be the ability to back an MPA with one of many identically weighted EBA (<gateway>.BTC), or perhaps an MPA backed by multiple differently weighted cryptos configured something like the following?

Code: [Select]

"reference_asset": {"name": "Multipass.USD", "amount": 1, "feed_price": bitUSD.median_feed_price}
"allowed_backing": {
  "BTC": {"allowed_tokens": [<gateway(s)>.BTC, bitBTC], "feed_price": (reference_asset["feed_price"]/bitBTC.median_feed_price)},
  "ETH": {"allowed_tokens": [<gateway(s)>.ETH, bitETH], "feed_price": (reference_asset["feed_price"]/bitETH.median_feed_price)},
  "USD": {"allowed_tokens": [<gateway(s)>.USD, bitUSD], "feed_price": (reference_asset["feed_price"]/bitUSD.median_feed_price)},
  "BTS": {"allowed_tokens": [BTS], "feed_price": (1/reference_asset["feed_price"])},
}

It'd be pretty neat to be able to settle an USD & get a portfolio of crypto worth $1 in return 👍

[bitcrab]

Chinese community has discussed this topic a lot, let me express my understanding on this new solution:

in current design of smartcoin system, feed price is defined as "real BTS market price in the pegged asset", borrowing, margin call, forcesettlement , black swan all has a logic base that "smartcoin need to be backed by enough BTS value calculated from the real market price".

this design really guarantee that the smartcoin are backed enough, but it's always too harsh, the typical scenario is when BTS is in downtrend and bitCNY are kept being squeezed out, bitCNY always get a high premium with serious shortage. this make smartcoin not a robust enough stable coin  solution.

the new feed price solution(I call it negative feedback feed price)redefine the feed price as "Guide Price".

BTS is always under/over evaluated  in the market, the smartcoin premium/discount is the indicator on how much BTS is under/over evaluated. witness need to calculated a Guide Price based from the market price and the smartcoin premium, the higher the smartcoin premium, the higher the Guide Price will be over market price.

increasing Guide Price will increase smartcoin supply, and then the smartcoin premium will go down, when the premium reach zero, the Guide Price will be as same as market price.

while in bull market and smartcoin has a discount, then Guide Price will be lower than market price, a negative feed back process in another direction will start.

here one key point is how to design the algorithm on negative feed back, negative feed back is a traditional control system problem and has some mature algorithms like PID, these can be taken as reference however more detail need to be considered in a financial system, this also need work and sufficient discussion.

to face the challenge on smartcoin, several ways are suggested, to make MCR dynamic, to adjust MSSR etc. ,the "negative feedback feed price" solution has some advantages than other solutions, it introduce negative feed back idea and is totally feasible in technical side.