This will be an experiment, like Bitcoin.  The closest thing to BitUSD is EuroDollars.   

Eurodollar deposits are redeemable for USD at the maturity of their deposit time.  This redeemability is backed by faith in the solvency of the bank that holds your deposit.  What we have been debating is why the bitshares market in BitUSD would actually track the BTS/USD rate, which would be necessary to redeem your bitUSD from the market for the value of 1 USD denominated in BTS.  So this is a bit tangent to our discussion, and again is a financial product where lump-sum cash flow changes hands at some point in time.  That contractual obligation (modulo trust) is sufficient to maintain a market value for Eurodollars that tracks dollars. 

I asked for an example of a financial derivative traded on a market that tracks an underlying merely via a mechanism of yield-adjustment, without lump-cash settlements that compensate directly and linearly for price changes of the underlying.  I've never heard of such a product, and it's premise seems to me to be ill-founded.

If I had to think of the closest product to bitUSD, I think it would be an equity.  Equities just pay dividends, as bitUSD does, and there's no notion of any terminal value of an equity.  However, equities give ownership in the net assets of a company, hence have backing in this sense.  They also don't track anything other than the value of the company (by definition), which in old-school finance was proposed to be the discounted expected value of the entire future stream of dividends from that company.  This makes sense as a way to value bitUSD also.  However, the future dividends of bitUSD depend directly and merely on its price (in the case where margin depends on bitUSD's price). There's no underlying net profits as there would be for a company to distribute as dividends.  But the price is valued from the dividends.  So we have a bit of a self-referential situation going on here, and it should be clear that in this valuation model there's no reason for the market price to converge to any particular value.

In the case where bitUSDs dividends are a function of an embedded BTS/USD price, then that price can be seen as analogous to something proportional to net earnings of a company.  In that case, we may expect some rational valuation model to work, but I don't see any reason why the value of the dividend stream from bitUSD would converge to a certain specific proportionality factor (the one needed to make the ratio of bitUSD/(BTS/USD) prices be 1) times its net income (BTS/USD price).  The price of bitUSD would be sensitive to all kinds of exogeneous things, first and foremost interest rates in the overall economy.

I appreciate your time in replying, again.

I could debate the tracking issue to great extent, but I think the ultimate proof will be when we launch the test network.

Indeed, it will be an interesting experiment.  However, BitShares, and by association Invictus, risk losing some credibility if it fails, so it's worthwhile to think it through carefully first.  I can tell that you are quite confident, although I'm still unable to fathom why from this conversation. 

If my theory is wrong, then all that is needed to 'make it work', is an external price feed.   

I don't agree.  I think this would help substantially, since it would give the yield incentive that you mentioned earlier.  However, as I pointed out, yield incentives are never sufficient to overcome fear of near-term losses in turbulent financial times.  That issue is further exacerbated by the fact that there is no guaranteed convergence of BitUSD at any terminal horizon (as there would be for bonds or futures), because there is no terminal horizon.  With typical financial derivatives, a buyer knows an upper bound on how much time he will have to bear price deviations before convergence.  With BitUSD, there is no such upper bound.

I base my 'price tracking' on the behavior of prediction markets that also have no ties to the real world and yet can track arbitrary ideas or concepts. 

Can you provide an example of what you mean?  The prediction markets that I'm familiar with have a terminal payout.  Do you know of ones that successfully track without a terminal payout?

Here is my challenge to you, role play longs and shorts against one another and see if you can manipulate the price without one side digging in.  Keep in mid that longs are competing against other longs and shorts against other shorts.   Then define the direction the market will move and why.

I'm not quite sure what you mean "longs are competing against other longs."  If I'm long, I'd very much like to see an increase in demand for new longs, as this pushes up the price, making my position more profitable. 

Furthermore, my argument isn't based on game theory, and game theory is superfluous in analyzing my argument.  If BitShares is successful, the market will be big, and typical players will be small.  This means they have negligible ability to manipulate the price.  (Of course in practice there are often big players that can manipulate prices at certain points.  That poses further vulnerabilities for BitShares, but my argument doesn't rely on price manipulation).  I'm only assuming agents will act out of their self-interest and/or emotions.  Crazy price fluctuations do not necessitate price manipulators.  They sometimes occur endogenously (black Monday, 1987), and sometimes are triggered by exogenous events such as the Japanese Tsunami and subsequent Fukushima fallout.  When abnormally large deviations in bitUSD vs BTS/USD occur, this creates some level of fear that tracking is failing.  When participants know that the only reason to believe tracking will succeed is if others believe that tracking will succeed, then this fear alone is sufficient to cause tracking to fail.  It has been so many many times in financial history, with assets that gave market participants more incentive to cause them to track than bitUSD vs BTS/USD gives. 

The main mitigating factor that I see with respect to my argument is that your margin requirements are much much higher than is typical in financial markets, so that should result in reduced feedback effects from selling inducing more selling due to solvency pressure.  As I said above though, reducing this feedback is not equivalent to incentivizing convergence.  When (not if) large deviations occur, it's essential that the market have high incentive to reverse the deviation if tracking is to occur.  I just haven't seen what that incentive might be.

Thanks again for your reply, bytemaster, during this busy time. In response to your previous points:

BitUSD Short margin requirements are based upon the highest unaccepted bid to buy BitUSD for BTS.

Shorts MUST buy back at some point if they ever want to use their collateral for anything else.   

So you apply game theory between the longs and the shorts.   The longs will hold out for a fair price, and it costs the shorts a lot of money to 'let it ride'.  Also, for someone who is short to maximize the return on their investment they need to cover, take a profit, and re-short at the new price.   As the value of BitUSD falls the collateral backing it goes from 2x to 3x to 10x to 100x... and thus the interest rate earned by those who hold BitUSD rises and the opportunity cost for the short also rises. 

As a result there is constant pressure for shorts to cover and and re-short at the new price.  The only way for them to cover is to convince the longs to give up their higher yield and thus agree to a price near BTS/USD.

Now the longs cannot hold out for ever either because at any time new shorts can enter the market if the longs try to push up the price by holding BitUSD off the market.

1.) So you're saying that you won't allow shorts to reduce their margin positions even if/when the bitUSD price has moved very much into the money for the shorts? 

This seems silly, as the shorts can still effectively reduce their margin by covering their previous position and initiating fresh positions at their desired level of exposure with new 2x margin, which would be lower than the previous 2x margin. It forces the shorts to needlessly make extra transactions to achieve the same position they could have had by just reducing their margin. Forcing shorts to cover in this way won't cause net long demand, hence won't cause the price to rise.


2.) Fungibility: If you don't allow shorts to reduce their margin when their positions are in the money, then the bitUSD contracts lose fungibility.  This is because the margin requirement of each would be contractually coded in at the time of creation, and would depend on the bitUSD price at that time.  I doubt that this is really the way you intend for things to operate.

Ok, those who own BitUSD are earning dividends at 2x the rate of those who own BTS. 
Those who are Short BitUSD are paying interest at 2x the value of BTS.

The only price at which someone holding BitUSD would be willing to sell is the BTS/USD exchange rate. Otherwise, they sell the BitUSD to someone who wants a high yield investment backed by BTS.

These dividends will scale down as margin scales down as shorts reduce their margin, which they will do even if you force them to cover their initial position and reenter a new one to do so. Since the margin requirements are based on the bitUSD price and not the BTS/USD exchange rate, then the goodness of the deal for the longs doesn't depend on the deviation of bitUSD from BTS/USD, so this is not a mechanism that keeps the prices tracking.

Furthermore, in panics yield-incentives typically are surpassed by fear of near/intermediate term price losses. Interest only accumulates linearly in time, whereas prices can change arbitrarily quickly. LTCMs positions had crazy yields at the market prices before they had to liquidate.  Still, buyers didn't enter and didn't cause their illiquid bonds to converge to the values of similar liquid ones.  Whereas in the setup you describe. . .

1.) Yields (dividends) won't effectively increase as bitUSD price drops, because shorts will reduce their margins - by exiting and reentering if you force this upon them.

2.) There's no terminal payout, so no implied yield (price appreciation) as there is for e.g. bonds.
 
I still see no incentive whatsoever for any market agent to buy and sell in such a way that e.g. the price of bitUSD will track the BTS/USD rate, other than the belief that it is common knowledge that they "should."  I still see this as a far fragiler incentive structure that has existed in many instances in financial history where instruments that were expected to track/converge had huge deviations and didn't.  On the plus side, I'll point out that your system requires far higher margin than is typical in financial markets, which should reduce the pro-cyclical feedback that occurs in financial panics.  Reducing this feedback is not, however, the same thing as incentivizing convergence.

I'm open to hearing new arguments or learning where I've misunderstood how the proposed system would work.

Anyone else losing connectivity quite a bit all of a sudden?  We have a few different IP options right?  I'm using 162.243.67.4.  What are the others?

Status?

Quote from: Winslow Strong on November 10, 2013, 09:19:50 pm

I don't understand the mechanism by which bitUSD would track the exchange rate of paperUSD/BTS.

Let's compare it to an ETF.  There, the value of the ETF tracks the value of the basket of equity/commodities or whatever else it is composed of.  Speculators rationally buy the ETF when its price is below the price of the basket, because the shares of the ETF itself could be redeemed for the actual underlying basket (modulo large size and fees).  Hence, the price of the ETF tracks the price of the basket due to arbitrage opportunities being available when there's a substantial disagreement between the two.   

But it doesn't seem that bitUSD are redeemable for paper USD.  So why would the market rationally buy bitUSD when its value (in BTS) is less than that of BTS/PaperUSD?

The idea is, people will bid whatever they think it's worth, and that price will track the 'true' price of USD. There is no physical USD being exchanged or stored, but a number of people trading on the platform will maintain that price. It's just like a prediction marketplace where equilibrium will bring it to the 'true' price. Of course, if for some reason, everyone believed that bitUSD should instead track bitCENT, then it will, but it would be unlikely.

Sorry, maybe I wasn't being clear enough. For the price to track the true USD price, its necessary that buyers would have incentive to buy when the price dips sufficiently below the true USD price.  What is that incentive? 

In a prediction market, the incentive is that eventually the share matures and is paid out, so there's a terminal value.  Similarly in futures markets.  With ETFs, there's no terminal date, but if the price of the ETF dips too low, theres an arbitrage opportunity for people to buy shares of the ETF and exchange it for a basket of the underlying.

It sounds to me like you're saying there is no such mechanism for bitUSD.  Then you are just relying on putting this idea out there that the prices should track, and hoping that others believe that others believe that others believe. . . that it should.  Without some type of redeemability, I don't see how anyone could believe this will actually work.

I've added 100+ nodes with same address, is that OK?

I wouldn't.  Myself and at least one other had balances wiped out when doing that.  For me there was a crash that occurred which may have been a cause.

All of my unpaid (and paid) balances were wiped clean upon transition, even though I'm not sticking multiple miners on the same address anymore.  Is this the same for everyone?