Author Topic: please ask delegate to stop using ALT's delegates auto feed script!!!!  (Read 4965 times)

0 Members and 1 Guest are viewing this topic.

Offline emski

  • Hero Member
  • *****
  • Posts: 1282
    • View Profile
    • http://lnkd.in/nPbhxG
The configuration in my version of the script allows each delegate to put different set of exchange trust levels.
Essentially this produces unique feed for any different configuration allowing every delegate to use slightly different rules.
When you add the random time between feeds you get harder to predict median.

Offline biophil

  • Hero Member
  • *****
  • Posts: 880
  • Professor of Computer Science
    • View Profile
    • My Academic Website
  • BitShares: biophil
I don't know if what I will ask makes any sense...
should the system take the media price feed also ... time weighted ?
Hug ... nice thinking ..

not sure about the economical implications ... but sure worth discussing!

I'd hesitate to do anything that increases the lag of the price feed. Lag is just as bad as determinism. Rather, price feeds should have a small degree of randomness in them.

Sent from my SCH-S720C using Tapatalk 2

Support our research efforts to improve BitAsset price-pegging! Vote for worker 1.14.204 "201907-uccs-research-project."

Offline biophil

  • Hero Member
  • *****
  • Posts: 880
  • Professor of Computer Science
    • View Profile
    • My Academic Website
  • BitShares: biophil
Can I also point out that any price feed script should add noise to the price feed, both in when it updates (so that nobody can predict exactly when it updates) and in what price it reports (so that nobody can predict exactly what price it will report). If the bid/ask spread is 1%, a price feed script should report a random variable whose mean is the mean of the spread, but that could take any value within there.

If people don't get it I'll post a stronger justification later.

How can you validate if the delegate is trying to manipulate that?
I don't think random numbers will help. Care to elaborate ?

Yeah, good question about delegate manipulation. That's a tricky one, but it seems to me that manipulation by delegates is a separate issue. Since there are so many delegates, there should be many feeds close to the median, so it should be difficult in general for one evil delegate to simultaneously be the median and manipulate the price. If he changes his price too much, he'll cease to be the median.

My thought with randomness is about minimizing manipulation by whales on external exchanges, which is what these Chinese guys are saying is happening.

If 90% of delegates use Alt's feed script and it publishes prices deterministically, a bad actor can tell precisely what the median price feed will be at all times. If they can tell what the feed will be, they can manipulate external prices and predict their precise effects on the median feed and profit at the expense of the credibility of the internal market.

Here's where randomness helps: if delegates don't report prices deterministically, but rather allow their reports to be randomly distributed around the true price, the bad actors will never be able to precisely predict the changes in the feed price, and their profit opportunities will be decreased.

I doubt this would completely solve the problem, but it would certainly help. I encourage all price feed script publishers to add randomness.

Sent from my SCH-S720C using Tapatalk 2

Support our research efforts to improve BitAsset price-pegging! Vote for worker 1.14.204 "201907-uccs-research-project."

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
I don't know if what I will ask makes any sense...
should the system take the media price feed also ... time weighted ?
Hug ... nice thinking ..

not sure about the economical implications ... but sure worth discussing!

Offline liondani

  • Hero Member
  • *****
  • Posts: 3737
  • Inch by inch, play by play
    • View Profile
    • My detailed info
  • BitShares: liondani
  • GitHub: liondani
Isn't that why we
- average over multiple exchanges
- run the script in different time patterns
- use the volume weighted average
- build the median over all feeds
???

I don't know if what I will ask makes any sense...
should the system take the media price feed also ... time weighted ?

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
Isn't that why we
- average over multiple exchanges
- run the script in different time patterns
- use the volume weighted average
- build the median over all feeds
???

Offline alt

  • Hero Member
  • *****
  • Posts: 2821
    • View Profile
  • BitShares: baozi
I have no idea to avoid  manipulation,
I thought volume can avoid this  yesterday, but somebody remind me, volume is even more easy to manipulation.

Offline emski

  • Hero Member
  • *****
  • Posts: 1282
    • View Profile
    • http://lnkd.in/nPbhxG
Can I also point out that any price feed script should add noise to the price feed, both in when it updates (so that nobody can predict exactly when it updates) and in what price it reports (so that nobody can predict exactly what price it will report). If the bid/ask spread is 1%, a price feed script should report a random variable whose mean is the mean of the spread, but that could take any value within there.

If people don't get it I'll post a stronger justification later.

How can you validate if the delegate is trying to manipulate that?
I don't think random numbers will help. Care to elaborate ?

Offline biophil

  • Hero Member
  • *****
  • Posts: 880
  • Professor of Computer Science
    • View Profile
    • My Academic Website
  • BitShares: biophil
Can I also point out that any price feed script should add noise to the price feed, both in when it updates (so that nobody can predict exactly when it updates) and in what price it reports (so that nobody can predict exactly what price it will report). If the bid/ask spread is 1%, a price feed script should report a random variable whose mean is the mean of the spread, but that could take any value within there.

If people don't get it I'll post a stronger justification later.
Support our research efforts to improve BitAsset price-pegging! Vote for worker 1.14.204 "201907-uccs-research-project."

Offline fluxer555

  • Hero Member
  • *****
  • Posts: 749
    • View Profile
Perhaps the best script would look at each individual trade from an exchange (or all exchanges at once?), take a rolling window of X number of BTS traded, and average the price from those X BTS. X could be determined by a percentage of the 1-hour window of total volume across all exchanges, so that when there is more volume, it takes more BTS into account when making the average.

If there aren't public APIs for this data, you could probably pull it from the front-end JSON API for listing recent trades...
« Last Edit: November 20, 2014, 08:51:45 pm by fluxer555 »

Offline emski

  • Hero Member
  • *****
  • Posts: 1282
    • View Profile
    • http://lnkd.in/nPbhxG
Isn't this free market?
Isn't this supposed to work this way?
My version of the script is extra sensitive to negative price movements 0.1% (at default settings) trigger feed update.

The problem is that the current price feed system does not factor in volume into the spot price, and spot price has little to do with actual valuation in an illiquid market. Look at Mastercoin's market cap, for example, vs its volume. If you were a MSC whale, you'd have a very hard time divesting at the 'spot price'.

My version of the script uses the volume reported by the exchange. If the volume is low its (that exchange's) price weight is reduced.

The only way to make more accurate script is if we collect all deals from all exchanges in the last few minutes and average everything accounting for volume and price of each trade.

Offline fluxer555

  • Hero Member
  • *****
  • Posts: 749
    • View Profile
Isn't this free market?
Isn't this supposed to work this way?
My version of the script is extra sensitive to negative price movements 0.1% (at default settings) trigger feed update.

The problem is that the current price feed system does not factor in volume into the spot price, and spot price has little to do with actual valuation in an illiquid market. Look at Mastercoin's market cap, for example, vs its volume. If you were a MSC whale, you'd have a very hard time divesting at the 'spot price'.

Offline emski

  • Hero Member
  • *****
  • Posts: 1282
    • View Profile
    • http://lnkd.in/nPbhxG
Isn't this free market?
Isn't this supposed to work this way?
My version of the script is extra sensitive to negative price movements 0.1% (at default settings) trigger feed update.

Offline taa

  • Full Member
  • ***
  • Posts: 54
    • View Profile
PMed.  ALT price feed trigger  is default to 0.5% change on bter and btc38

Explain to me what is going on?

If you manipulate the feed you can in theory cause some shorts to execute at a slightly better price (say the price was manipulated 1% on an exchange)... but the spread on the internal market is in excess of 1% so the accuracy of the feed below those tolerances is not an issue.

Can you show a time where the feed was "manipulated" by more than 1%?

Offline bytemaster

Explain to me what is going on?

If you manipulate the feed you can in theory cause some shorts to execute at a slightly better price (say the price was manipulated 1% on an exchange)... but the spread on the internal market is in excess of 1% so the accuracy of the feed below those tolerances is not an issue.

Can you show a time where the feed was "manipulated" by more than 1%?

For the latest updates checkout my blog: http://bytemaster.bitshares.org
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.