Author Topic: BitAssets and Black Swan Events [BLOG POST]  (Read 4851 times)

0 Members and 1 Guest are viewing this topic.

Offline Markus

  • Sr. Member
  • ****
  • Posts: 366
    • View Profile
"In order to break BitUSD the value of BitShares would have to fall by 67% in a market where no one was willing to sell enough BitUSD to allow all existing shorts to cover. For all practical purposes this fall would have to occur over just a few days, in thin markets, with no expectation for a rebound in value."

OK, let's see what thin market means in this case. Currently there are about 600 000 BitUSD shorts expired and a few BitGold and BitBTC.

Offline liondani

  • Hero Member
  • *****
  • Posts: 3737
  • Inch by inch, play by play
    • View Profile
    • My detailed info
  • BitShares: liondani
  • GitHub: liondani
can we make please such experiments on DevShares before we made them on BitShares?

Sent from my ALCATEL ONE TOUCH 997D


Offline Chronos

Would the black swan event described in this post result in the liquidation of all BitUSD in existence? Is it possible for any BitUSD to survive the margin call? What about other BitAssets -- can a black swan event liquidate BitUSD but leave BitGold untouched?

A black swan event could negatively effect bitUSD and at the same time positively effect bitGOLD and/or other bitAssets (ie. such as a huge unexpected devaluation of the USD).
That doesn't quite answer my questions, but I don't mean any arbitrary black swan event. I mean a BTS black swan as described in the blog post. This one:



"In order to break BitUSD the value of BitShares would have to fall by 67% in a market where no one was willing to sell enough BitUSD to allow all existing shorts to cover. For all practical purposes this fall would have to occur over just a few days, in thin markets, with no expectation for a rebound in value."

Offline onceuponatime

Would the black swan event described in this post result in the liquidation of all BitUSD in existence? Is it possible for any BitUSD to survive the margin call? What about other BitAssets -- can a black swan event liquidate BitUSD but leave BitGold untouched?

A black swan event could negatively effect bitUSD and at the same time positively effect bitGOLD and/or other bitAssets (ie. such as a huge unexpected devaluation of the USD).

Offline Chronos

Would the black swan event described in this post result in the liquidation of all BitUSD in existence? Is it possible for any BitUSD to survive the margin call? What about other BitAssets -- can a black swan event liquidate BitUSD but leave BitGold untouched?

Offline mdj

  • Full Member
  • ***
  • Posts: 192
    • View Profile
  • BitShares: mdj
I think the main concern with immediate liquidation would be the loss of all positions. I think after a substantial crash already triggering the event, the sentiment would encourage the immediately liquidated BTS to be sold. It wouldn't be a nice situation and I can't think of a good solution, it would hurt regardless. The problem is if compared to the current banking system, banks would get bailed out by the government (not saying this is a good thing) but this approach is more like what we saw in Cyprus where everyone's savings take a hit.

Offline arhag

  • Hero Member
  • *****
  • Posts: 1214
    • View Profile
    • My posts on Steem
  • BitShares: arhag
  • GitHub: arhag
Your approach is complicated to implement, complicated to explain, and converting to BTS is just as good as BitUSD IOU only it is likely to rebound faster than your uncollateralized USD-IOU. 

*SIMPLE* matters more than anything else when it comes to BlackSwans that are already scaring people.

I still think the approach I outlined is superior to immediate liquidation and should be implemented in the long term. But, I certainly don't want development effort going towards implementing it now (I admit it is a more complicated system to implement) which would further delay 1.0 release. In the mean time, a solution is necessary that is better than freezing the markets, so I am happy that you guys will be implementing the immediate liquidation in case of undercollateralization feature (for now).

However, I think that eventually something like what I described is important to build, especially before the feature to allow any user to create their own BitAssets with custom price feeds, margin call limits, and backing collateral BitAsset choice.

Offline bytemaster

Your approach is complicated to implement, complicated to explain, and converting to BTS is just as good as BitUSD IOU only it is likely to rebound faster than your uncollateralized USD-IOU. 

*SIMPLE* matters more than anything else when it comes to BlackSwans that are already scaring people.   
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.

Offline arhag

  • Hero Member
  • *****
  • Posts: 1214
    • View Profile
    • My posts on Steem
  • BitShares: arhag
  • GitHub: arhag
Your approach is similar to what is currently implemented:  there is no yield until all BitUSD is reclaimed as fees.

I don't think it is the same. If a sudden price feed correction happens that causes all of the shorts to be margin called and placed at the new price feed. However, this new price feed would be so high (in BTS/BitUSD) that some of these shorts would not be able to collect enough BitUSD if their order was matched to pay off their BitUSD debt to the network. Let's assume that after many of these margin calls were matched, the BitUSD debt they owed minus the BitUSD they collected from the matching is greater than the amount of BitUSD in the yield pool. In that case, the current system should be left with 0 BitUSD in the yield pool, some BitUSD liabilities by the network that is not backed by any of the remaining short positions, and the remaining short positions margin called at the price feed (but not yet undercollateralized at that price).

Now this undercollateralization (let's say the deficit is 1000 BitUSD) causes everyone to panic and get out into BTS while they still can. I will assume the price does not adjust any higher (in BTS/BitUSD) to cause more of the remaining margin call orders to become undercollateralized. Even with that assumption, in the current system (actually not the current system, but a system very similar to the current one but where the markets didn't freeze as a result) some people could be left with worthless BitUSD. If I have 1000 BitUSD in my wallet and am slow to move (everyone else moves into BTS before me), all the short positions will be covered and I will be left with 1000 BitUSD with 0% backing. Furthermore, everyone knows that if a new short was to be created, the system would start in a state of undercollateralization.

So how is that different from what I am proposing? Well, in my proposed system, the moment in which undercollateralization occurs (assuming the yield pool has been depleted already), everyone's BitUSD is split into BitUSD and BitUSD-IOU. Let us say there is an outstanding supply of 1,000,000 BitUSD. Before the undercollateralization event, the short positions collectively owe the network 1,000,000 BitUSD. The black swan happens, the price feed adjusts, the shorts are margin called. If placing any short's margin call order at the price would cause it to be undercollateralized if matched, then the system has to do something different. It first considers that short position the property of the DAC (meaning the owner of that short won't be getting back any BTS at all). It takes the collateral from all of the shorts matching that condition and determines how much BitUSD it could get with it at the price feed. It then calculates the amount of BitUSD it is missing to pay off the collective debt owed by those shorts. Let's pretend that the deficit is 1000 BitUSD. In that case what it conceptually does is convert 1000/1,000,000 = 0.1% of all outstanding BitUSD in the system into BitUSD-IOU (owned by the same owner as the BitUSD converted) and it prints up the 1000 BitUSD and holds on to it to cover the deficit when those undercollateralized short margin call orders are matched. The outstanding BitUSD supply is still 1,000,000 BitUSD, which is the same as the debt owed by the network. However, the remaining collateralized shorts + the BTS collateral at the price feed which was collected from the undercollateralized shorts give the network to claim to 999,000 BitUSD worth of BTS assets (at the particularly BTS/BitUSD price feed). Also, there is an additional 1000 BitUSD-IOU liability by the DAC that is held by the original BitUSD holders. As the price feed adjusts, it will also make the necessary adjustments in the BitUSD vs BitUSD-IOU distribution in order to maintain the amount of BitUSD necessary to cover whatever deficit occurs. That means if the price goes back up, the DAC could return the BitUSD that it held to cover the deficit back to the BitUSD-IOU holders (in exchange for destroying those BitUSD-IOUs). This all happens automatically in the blockchain without user intervention necessary.

The point is that if Alice is moving faster than Bob to liquidate into BTS, she won't really gain an advantage based on the information she has (she may have an advantage or disadvantage as the price changes in the future just like anyone does currently, but no one can predict the future). If Alice has 999,000 BitUSD and Bob has 1000 BitUSD in their wallets, then the undercollateralization event described in the above paragraph would cause 0.1% of each of their stakes to be converted to BitUSD-IOU. Alice will have 998,001 BitUSD and 999 BitUSD-IOU, while Bob will have 999 BitUSD and 1 BitUSD-IOU. If Alice buys BTS at the price feed with all of her BitUSD, the DAC will have some collateralized margin called shorts remaining that collectively provide the network a claim on 999 BitUSD worth of BTS assets (999,000 - 998,001 = 999), and the network will now owe a debt of 1,999 BitUSD (but it holds 1,000 BitUSD collected through the 0.1% tax, so in reality it is a 999 BitUSD debt owed). Now, assuming the price stays the same, Bob can come along anytime later and trade his 999 BitUSD for 999 USD worth of BTS (which the network still has assuming the price had remained the same). This would close all the remaining margin called shorts (which means the network would not have any more BTS assets to back the BitUSD market), and reduce the debt owed by the network to 1,000 BitUSD. But that is exactly the amount of BitUSD the network was holding to keep things balanced, so it can just cancel the liability and assets out and be left with zero BitUSD debt and zero BitUSD assets. Finally, Alice is left with 999 BitUSD-IOU and Bob is left with 1 BitUSD-IOU. These assets will likely be worthless.


Your graceful approach is similar to Mt. Gox experiencing a black swan event and then attempting to gracefully recover by selling bonds that it would take "for ever" to pay off.

So, what if it does? If it never pays off its bonds it is equivalent to charging all of its holders a percentage tax on their balances to make itself whole. That is better than a bank run that could leave certain slow movers with nothing.

Offline graffenwalder

When BitUSD is force-selled, are you going to prioritize BitUSD that hasnt moved the longest first?

i.e. If a BitUSD holder lost his private keys years ago, then his BitUSD should be the first to sell.

Everyone gets the same price so order doesn't matter.

IIUC it matters in that not all 100% of existing BitUSD needs to be sold in a black-swan to cover shorts - just some of it. Thats where you can prioritize so most active BitUSD holders wouldnt even notice.

I don't think it is wise to treat some BitUSD differently than other BitUSD.   

You are right that in theory I only have to force *SOME* BitUSD to cover, but prioritizing the oldest requires keeping balances sorted by date (big performance hit for normal operation).   

I think calling it a clean slate is best for everyone.  Chances are that if one guy has hit the limit then almost EVERYONE is below the 200% minimal margin but above the 100% technical solvency level.   Bottom line, *IF* we hit the Black Swan event the whole system needs reset anyway because all hell is breaking lose out there.
The whole system, or just that particular bitasset?

Offline bytemaster

When BitUSD is force-selled, are you going to prioritize BitUSD that hasnt moved the longest first?

i.e. If a BitUSD holder lost his private keys years ago, then his BitUSD should be the first to sell.

Everyone gets the same price so order doesn't matter.

IIUC it matters in that not all 100% of existing BitUSD needs to be sold in a black-swan to cover shorts - just some of it. Thats where you can prioritize so most active BitUSD holders wouldnt even notice.

I don't think it is wise to treat some BitUSD differently than other BitUSD.   

You are right that in theory I only have to force *SOME* BitUSD to cover, but prioritizing the oldest requires keeping balances sorted by date (big performance hit for normal operation).   

I think calling it a clean slate is best for everyone.  Chances are that if one guy has hit the limit then almost EVERYONE is below the 200% minimal margin but above the 100% technical solvency level.   Bottom line, *IF* we hit the Black Swan event the whole system needs reset anyway because all hell is breaking lose out there.
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.

Offline speedy

  • Hero Member
  • *****
  • Posts: 1160
    • View Profile
  • BitShares: speedy
When BitUSD is force-selled, are you going to prioritize BitUSD that hasnt moved the longest first?

i.e. If a BitUSD holder lost his private keys years ago, then his BitUSD should be the first to sell.

Everyone gets the same price so order doesn't matter.

IIUC it matters in that not all 100% of existing BitUSD needs to be sold in a black-swan to cover shorts - just some of it. Thats where you can prioritize so most active BitUSD holders wouldnt even notice.

Anyway this is just a suggestion.
« Last Edit: January 27, 2015, 10:38:19 pm by speedy »

Offline bytemaster

When BitUSD is force-selled, are you going to prioritize BitUSD that hasnt moved the longest first?

i.e. If a BitUSD holder lost his private keys years ago, then his BitUSD should be the first to sell.

Everyone gets the same price so order doesn't matter.
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.

Offline speedy

  • Hero Member
  • *****
  • Posts: 1160
    • View Profile
  • BitShares: speedy
When BitUSD is force-selled, are you going to prioritize BitUSD that hasnt moved the longest first?

i.e. If a BitUSD holder lost his private keys years ago, then his BitUSD should be the first to sell.

Offline bytemaster

Who makes good on this IOU?

The blockchain takes any new BitAsset fees (mostly new short interest) and instead of putting it in the yield uses it to convert as much of the BitAsset-IOU in circulation back into the corresponding BitAsset (in proportion of course so it is all fair). In reality, I expect the implementation to have some global fraction parameter for each BitAsset that distinguishes how much of each user's balance is BitAsset vs BitAsset-IOU, and the DAC simply adjusts the parameter as it is able to in order to ensure the amount of outstanding real BitAssets is always at least 100% backed by the BitAsset debt owed by the open short positions.

What if income from USD fees is not able to make it whole?

Anyone could take their real BitUSD and sell it for BTS on the exchange as usual. But the BitUSD-IOU would be in a separate market and since BitUSD-IOUs cannot be used to cover short positions, the margin called / expiring shorts cannot match against the BitUSD-IOU. This means that (assuming further undercollateralization didn't happen beyond what had already happened) all of the BitUSD holders could exit out into BTS unwinding all of the short positions. What would be left over is some amount of BitUSD-IOU that would likely be worthless (although maybe not because they are claims on the future interest paid by any new shorts that reappear in the fresh BitUSD/BTS market). The DAC would not put any new BitUSD into the yield pool until it paid off all of the BitUSD-IOUs first.

Also, it is true that if undercollateralization occurs, then some people exit out into BTS, and then the price drops again and further undercollateralization occurs, the people leaving after the second undercollateralization event would be worse off than the ones who left after the first but before the second undercollateralization event. But one can make the same argument comparing to sellers before the first undercollateralization occurs. The ones who sold their BitUSD for BTS and then for real USD before the first undercollateralization event would be better off than the people who sold (or were forcefully liquidated) after undercollateralization into BTS and then converted their smaller amount of BTS into real USD. I think this is an acceptable risk because I imagine undercollateralization events like this will only happen after some sudden news breaks out to cause a sudden flash crash, and I'm not as concerned about a black swan happening on top of a black swan before people are able to place their relative BitUSD sell orders just under the price feed after the first black swan event.

How do we draw the line and declare the system insolvent and in need of liquidation?

When undercollateralization occurs the system is already insolvent. I am just arguing for a more graceful way of handling the resulting outcome than forceful liquidation. It could allow the DAC to earn enough BitUSD income to repay the old BitUSD holders that weren't able to get their full promised value back because of the undercollateralization.

Your approach is similar to what is currently implemented:  there is no yield until all BitUSD is reclaimed as fees.   Your graceful approach is similar to Mt. Gox experiencing a black swan event and then attempting to gracefully recover by selling bonds that it would take "for ever" to pay off.   

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.