Suppose we actually didn't need forced settlement (for bitUSD etc)...

[merivercap]

The argument you'll get against this is that we don't want to be so dependent on the price feed and I understand the sentiment, but the system will be far more robust to depend on the price feed when there is no liquidity. 

Isn't that what we are doing:

our DEX at low liquidity = the feed price
our DEX at high liquidity becomes the market (that most people use) and the tail (us) is now wagging the dog (the markets), so only then is when we can take the training wheels off

But are we not doing this currently? and if not, why?  I thought's that's why we had to have the "peg" in the first place (only for low liquid times)

Not really.  The price feed is not really used at all at the moment other than for forced settlement afaik.  Since we reduced the SQP lower it got closer to the price feed in the bitUSD market, but still is 10% off.  There is not much buying pressure and it seems people just put orders just below the SQP.  That's why if you put SQP at the price feed you'll effectively be basing settlement on the price feed.

[void]

The price feed would be used for collateral requirements and for working out real time the equivalent  voting weight and possibly in the future, yield, which would be due.   For these things, we just need a price feed that is roughly accurate.  It wouldn't need to be hyper accurate on an intra-day basis because we wouldn't have the request settlement option.

The problem with current system which this system might fix is the *uncertainty* for the short position thus leading to low supply and a high premium for the asset.

Currently, a shorter may get their position closed at any time, and it might not be for a good price.   (This is why I am personally wary of using the current system to short anything.)

In the proposed system, all the shorter needs to worry about is that their collateral doesn't drop below a certain threshold on the average price on the external exchanges.  There margin rule would be very simple and it would not depend on any other players in the internal system.

For example, I want to short some bitBTC and short some bitUSD, so that I can trade these two against each other.   I don't really want to care about the price in BTS.  In the current system there is uncertainty because these short positions can be closed at any time due to unknown variables.  This makes me avoid doing this.  In the alternative system, provided I maintain a fixed and known colleteral level, I am assured that my bitBTC / BitUSD positions won't be closed.   This is the major difference

This, along with possibility that the assets might trade exactly around their price feed over time

[merivercap]

I'm not sure about the details you're proposing, but I agree with the direction.  We can set the SQP to 1000 which is effectively the same thing.  (Alternatively we can design margin calls to settle at the price feed before walking the book)

The argument you'll get against this is that we don't want to be so dependent on the price feed and I understand the sentiment, but the system will be far more robust to depend on the price feed when there is no liquidity.  Furthermore you can design a price feed to include pricing from the internal exchange so you can slowly move away from external pricing influence gradually anyways.     

[void]

UPDATE: I have just thought of a possible idea of how this new version of bitUSD without the forced settlement option (suppose we call it bitUSD2), could have intrinsic value. 


Suppose we could make it have *precisely* the the same intrinsic value of the equivalent BTS which is held as part of the collateral, at the current price feed?


I.e., All the rights/benefits of the equivalent BTS at the current price feed would be transferred to the bitUSD2 holder, and deducted from shorter's collateral which is held against it.  This calculation would update every time the price feed changed.
This would transfer the *voting* rights to the bitUSD2 holder.
And it would transfer any *yield/dividends* paid, if we were to introduce that in the future for BTS, to the bitUSD2 holder.
(This would not be like the previous idea of yield.  Just the equivalent of anything that BTS would get.)
But I think this would be a nice sell (both for BTS and bitUSD2)?! 


So, would this not mean the the bitUSD2 had the equivalent value in actually fact to the BTS at the price feed?

And therefore, there would be no reason that bitUSD2 shouldn't trade at exactly around the price feed?!

(likewise for any other version of this - CNY, etc.)


Bytemaster / anyone got views on this?
I'd be interested to hear if this might in any way be technically possible :-)

[void]

I posted a reply earlier today where I said I thought it was absolutely required(!)...but then started wondering a bit!!

Suppose the price feed combined with collateral rules,
(along with the interests of the parties wishing to short and hold bitUSD),  mean that we could in practice still track the real price.

The argument against is that buyers of bitUSD wouldn't do so unless they had a guaranteed way out at around the real price.
But, while they hold bitUSD someone else has to have their BTS tied up as collateral. 
So at any point in time there should always be holders of bitUSD and shorters of bitUSD wishing to exit.
An agreed price would be settled upon.

If the price of BTS goes up, the bitUSD shorter can remove collateral.
If the price of BTS goes down, the bitUSD shorter has to add collateral.
If the shorter's collateral to back the asset falls below e.g. 1.5 x the current value, the system takes ownership of it and a buy order is placed at the price feed.
Could this be enough to mean we track the real price?

I guess we maybe tried something like this with BTS1.0, but the thing I'm wondering is if what was tried was exactly like this - I.e., price feed for collateral, but no other rules?
Also BTS1.0 didn't have a simple way where anyone could just create bitUSD (although this may be a minor thing).

(Another argument is that nuBits still seems to be working...)
I probably wouldn't use nuBits because it's not backed in a trust-less way (so not much better than an exchange).
But this smart-coin would still be collateralized at x times a real-value asset (BTS, where the value comes from voting stake + maybe even at some stage, profit  ). 
With nuBits, all of the asset created is done by the market maker.
In this system it wouldn't be.  But, nevertheless,  shorters might not typically wish to short much below the real price;
buyers might not typically wish to buy much above the real price.
And perhaps we'd see interested parties creating 1% spreads exactly on the real price...

The present system is very cool (and an amazing achievement IMO), but the fact that holders can settle at any time means the shorter must build in a high premium to guard against volatility because they can't exit at any time.  But pricing the required premium is difficult.  How do the market participants figure this out (10%, 20%, 50% above..).  With the system where neither side can automatically exit, the pricing can naturally be simple and the market might conceivably centre on somewhere near the actual price.

But I still find it hard to imagine this would work, and that common-sense dictates the current system is required(!)  But maybe this is wrong for the above reasons, and that the thing that shouldn't possibly work might actually work!
So perhaps it so could be worth trying, with an experimental new alt. asset - e.g 'bitUSD-exp'?