Author Topic: Max short holding period will make shorts paying interest easy to implement  (Read 5030 times)

0 Members and 1 Guest are viewing this topic.

Offline Ander

  • Hero Member
  • *****
  • Posts: 3506
    • View Profile
  • BitShares: Ander
This solution seems excellent.  Shorts compete in a free market system by setting interest rates they are willing to pay.

Competing over collateral lead to issues as others pointed out - in order to compete, one would need to borrow a bunch more collateral and post it.  This made the short look like they were well collateralized when they actually were not.
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Offline tonyk

  • Hero Member
  • *****
  • Posts: 3308
    • View Profile
I am partially surprised to say it, but I kind of like this solution (for more than one reason)...

Are you gonna allow closing using the collateral or you have not decided on that yet?
Lack of arbitrage is the problem, isn't it. And this 'should' solves it.

Offline bytemaster

I think we need to consider the difference between "interest" and "fee"...

In the original design their was a fee which was basically bidding up the spread and lengthening the time until a short can cover at a profit.

Under an interest system the fee is a percent per unit time rather than an upfront cost.   An interest system encourages covering sooner rather than later as the fee grows with time and thus creates an opportunity cost. 

Increasing the collateral requirement and fixing it at 3x  (2x from short, 1x from long) would provide greater protection and the interest rate would then rise to compensate the longs for any risk.   The increase in collateral also provides enough room such that for all reasonable interest rates collateral will always be over 2x even with interest due.

The default interest rate would be 0...

Bidding up the collateral rate has diminishing returns... while it is a pricing system and will work, the results it will yield for the network as a whole are less.

 +5% +5% +5%

I think I understand this - collateral is fixed but at 2X from the short and instead they compete on an annual interest rate that they are willing to pay.

Do you think market makers will be fine with this? Or would it work better if the interest rate only started after you hadn't covered for a week.

I had considered a 24 hour grace period, but concluded that the interest accrued over such short periods of time would be far less than the spread and thus not worth considering.
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 Empirical1.1

  • Hero Member
  • *****
  • Posts: 886
    • View Profile
I think we need to consider the difference between "interest" and "fee"...

In the original design their was a fee which was basically bidding up the spread and lengthening the time until a short can cover at a profit.

Under an interest system the fee is a percent per unit time rather than an upfront cost.   An interest system encourages covering sooner rather than later as the fee grows with time and thus creates an opportunity cost. 

Increasing the collateral requirement and fixing it at 3x  (2x from short, 1x from long) would provide greater protection and the interest rate would then rise to compensate the longs for any risk.   The increase in collateral also provides enough room such that for all reasonable interest rates collateral will always be over 2x even with interest due.

The default interest rate would be 0...

Bidding up the collateral rate has diminishing returns... while it is a pricing system and will work, the results it will yield for the network as a whole are less.

 +5% +5% +5%

I think I understand this - collateral is fixed but at 2X from the short and instead they compete on an annual interest rate that they are willing to pay.

Do you think market makers will be fine with this? Or would it work better if the interest rate only started after you hadn't covered for a week.

Offline bytemaster

I think we need to consider the difference between "interest" and "fee"...

In the original design their was a fee which was basically bidding up the spread and lengthening the time until a short can cover at a profit.

Under an interest system the fee is a percent per unit time rather than an upfront cost.   An interest system encourages covering sooner rather than later as the fee grows with time and thus creates an opportunity cost. 

Increasing the collateral requirement and fixing it at 3x  (2x from short, 1x from long) would provide greater protection and the interest rate would then rise to compensate the longs for any risk.   The increase in collateral also provides enough room such that for all reasonable interest rates collateral will always be over 2x even with interest due.

The default interest rate would be 0...

Bidding up the collateral rate has diminishing returns... while it is a pricing system and will work, the results it will yield for the network as a whole are less.

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 GaltReport

This may not be valuable input but as a user who has bought BitUSD and Shorted BitUSD and thought about it (granted not at a very sophisticated level :) ) I was way more excited about the interest than worried about the collateral level.

Offline Empirical1.1

  • Hero Member
  • *****
  • Posts: 886
    • View Profile
I think I basically agree with OP, that the previous solution price fixed collateral and the current implementation price fixes interest both are not efficient.

https://bitsharestalk.org/index.php?topic=9029.msg117925#msg117925

While not understanding this stuff, it seems the last solution price fixed collateral (Sub-optimal)  but this one price fixes fee/interest. (Sub-optimal)

This solution makes it better for market makers but decreases genuine BitAsset demand as it offers interest rates that are 300%+ lower probably??

Isn't the best system one that encourages market makers and genuine BitAsset holders, Optimal interest + optimal collateral?

...

50% of new BitUSD is awarded to highest collateral and 50% to highest fee/interest when demand is at or above feed.

Then we have a system with more collateral and attractive interest that encourages market makers and genuine BitAsset demand (who will also appreciate more collateral & tighter range)

I guess below is probably more optimal...


In other words, just specify some hard-coded function f(collateral_ratio, interest_rate) with df/dx and df/dy both positive, and sort shorts according to their f() value.

Basically f() represents to what extent the network is willing to reward higher collateralization with an interest rate discount.

But otherwise, what is so bad about awarding 50% of new shorts based on the collateral system and 50% based on the interest system?

In my non complicated maths mind, it seems then both systems will achieve an optimal collateral and interest rate and buyers of BitUSD will get the benefit of both?

Offline arhag

  • Hero Member
  • *****
  • Posts: 1214
    • View Profile
    • My posts on Steem
  • BitShares: arhag
  • GitHub: arhag
Where to draw the line between meaningful protection against volatility and waisted collateral is a judgement call that is difficult to make.   
I think it's fine that the market decides how much is too much.  Let the shorts make the judgment call as they compete to short.

The current system of prioritizing by collateral doesn't let the market make that judgement call. The judgement call being whether the marginal reduction in black swan risk is worth the opportunity cost of not having extra DAC revenue to use for various purposes (increasing BitAsset demand through higher yields, funding development and marketing, etc.). I don't think the market can make that judgement call. I think that balance is something the shareholders need to decide.


Offline Agent86

  • Sr. Member
  • ****
  • Posts: 471
  • BTSX: agent86
    • View Profile
Where to draw the line between meaningful protection against volatility and waisted collateral is a judgement call that is difficult to make.   
I think it's fine that the market decides how much is too much.  Let the shorts make the judgment call as they compete to short.

I think if people want long-term shorts and high yield USD then the solution is a Certificate of Deposit of a fixed term.   Someone with USD would actually LEND it to someone backed by sufficient collateral for a specified period of time and an interest rate.

Now you are moving the long-term shorts out of the market engine and into private CDs.  You are also separating the USD holders demanding liquidity from those demanding a return.  As a result you also remove all restrictions on short selling for someone who legitimately borrows the USD from a long rather than creating USD to sell.   

So now "shorts" are given an option:  pay interest to a long-term USD holder and have freedom to sell at any price and any time  *or* be restricted by the price feed and post even higher collateral.

I think providing this alternative lending process is better than attempting to integrate it all into a single market engine with competing goals.
+5%
I agree that we need a separate market to address interest rates.  I think the whole focus on yield through transaction fees is a distraction.

I think of CDs as non-transferrable whereas bonds are transferable/tradeable.  I think it is much more useful if there is a bond-market so that people holding the bonds (pays x bitUSD at x time) can trade them rather than wait for the maturity date.


Offline arhag

  • Hero Member
  • *****
  • Posts: 1214
    • View Profile
    • My posts on Steem
  • BitShares: arhag
  • GitHub: arhag
Thinking as a BitUSD holder (which I am) I care more about liquidity than I do interest. 
...
At 2x collateral I can still see potential for black swan events.

What about the idea of not complicating the shorts and just sticking with a fixed 2x collateral with some kind of fee-based prioritization that doesn't hurt liquidity (either the interest proposal here, or the capital gains tax method), but letting the DAC further reduce black swan risk by holding a higher reserve of BitAssets at the expense of smaller yields?


Offline bytemaster

Are we getting less "interest" with the new short rules? 

If so, I don't think this is a good trade off for excess collateral.

Thinking as a BitUSD holder (which I am) I care more about liquidity than I do interest.   The higher the collateral level, the more "leverage" I (as BitUSD holder) have over the short to make sure I am given a fair price. 

The law of marginal utility does come to play here... at some point the collateral level is so high that adding additional collateral is meaningless. 

At 2x collateral I can still see potential for black swan events... this potential is reduced significantly by having a 30 day max holding period which forces losing shorts to boost their collateral back to 2x.   At 2x collateral you are protected against a 50% fall.  At 3x collateral you are still getting meaningful reduction in potential for black swan events, being protected by a 66% fall.   At 4x collateral you are talking about a 75% fall in BTSX price in a very short time... at 5x collateral you are talking about an 80% fall in BTSX price in a very short time...  increasing collateral by  25% (from 4x to 5x) only gains you an extra 5% protection.  Doubling it from 5x to 10x collateral you are only protected by another 5%... a fall of 90% (almost instantly, not gradually)

So you can see that after a certain point the value of additional collateral is no longer protecting you against realistic volatility and such a rapid fall would only happen if a catastrophic failure of the network happened meaning all BitAssets will have trouble finding liquidity even if on paper they are still trading at parity.

Where to draw the line between meaningful protection against volatility and waisted collateral is a judgement call that is difficult to make.   

What is likely to happen with the collateral only approach is that the quantity of USD available to short will end up decreasing because there is only so much BTSX that is interested in shorting at all.  So the higher the collateral the less USD each short can back and thus the short wall decreases.  The less USD for sale means that it is more likely that a surge in demand for BitUSD will push through the high-collateral shorts and hit the lower collateral shorts.   In effect, collateralization cannot go to infinity because quantity of USD shorted would go to 0.

So market makers are in the business of volume, high collateral limits the volume of USD they can process at one time and thus greatly hurts their profits and bottom line. 

I think if people want long-term shorts and high yield USD then the solution is a Certificate of Deposit of a fixed term.   Someone with USD would actually LEND it to someone backed by sufficient collateral for a specified period of time and an interest rate.

Now you are moving the long-term shorts out of the market engine and into private CDs.  You are also separating the USD holders demanding liquidity from those demanding a return.  As a result you also remove all restrictions on short selling for someone who legitimately borrows the USD from a long rather than creating USD to sell.   

So now "shorts" are given an option:  pay interest to a long-term USD holder and have freedom to sell at any price and any time  *or* be restricted by the price feed and post even higher collateral.

I think providing this alternative lending process is better than attempting to integrate it all into a single market engine with competing goals. 




 
 
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 GaltReport

Are we getting less "interest" with the new short rules? 

If so, I don't think this is a good trade off for excess collateral.

Offline bytemaster


Why not instead prioritize first by collateral ratio, starting from the minimum (say 200%) up to a maximum (say 400%)? Then beyond that point only the interest rates are used to prioritize matching. Do we really need more than 400% reserve? How conservative do we want to get here?

 +5%


I like this logic

More complex, but worth considering if collateral ratios prove to get that high. 
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 liondani

  • Hero Member
  • *****
  • Posts: 3737
  • Inch by inch, play by play
    • View Profile
    • My detailed info
  • BitShares: liondani
  • GitHub: liondani
Why not instead prioritize first by collateral ratio, starting from the minimum (say 200%) up to a maximum (say 400%)? Then beyond that point only the interest rates are used to prioritize matching. Do we really need more than 400% reserve? How conservative do we want to get here?

 +5%


I like this logic

Offline theoretical

Do we really need more than 400% reserve? How conservative do we want to get here? Like drltc pointed out, there is a balance here: the more conservatively the DAC operates the less value it can capture for useful purposes such as higher yields on BitAssets, or even payments to delegates/workers to grow the DAC in other ways.

I think this argument conclusively shows that having shorts pay interest will be better for the BitShares ecosystem than having shorts develop excessive collateralization.  bytemaster's comments in this thread and elsewhere lead me to believe that he disagrees; he thinks collateral ratio is more important.  (At https://bitsharestalk.org/index.php?topic=9532.0 I develop at great length another objection to excessive collateralization.)

I was trying to argue that there should be one variable and that variable should be interest.  Given bytemaster's strong support for collateralization ratio as the one variable, I thought interest on shorts would be likelier to be implemented if I offered the compromise of including both.  You rightly point out that this compromise will add complexity and make it harder for market participants to formulate strategy.

However, I worry that your function f is going to be way too complicated for market participants to reason about. How exactly is someone considering shorting supposed to decide on their strategy? They now have three variables to tweak: price, collateral ratio, and interest rates (I am assuming the market decides interest rates rather than having it be universally imposed by the network at any given time).

Dealing with the three-dimensional nature of the problem makes it difficult.

We could get rid of the "price" dimension by having two different types of shorts:  A fixed-price short with 2x collateral and no interest, and a "wall short" or "feed short" that moves with the feed but has variable collateral and interest.  Fixed-price shorts would only execute when there are no wall shorts, and they wouldn't really be useful when there's a lot of people wanting to short.

I was thinking of showing the f() score of each offer.  With one pretty graph of the total BitUSD of orders that would be ahead of you as a function of the interest rate assuming you kept the current leverage ratio, and another pretty graph of the total BitUSD of orders ahead of you as a function of the leverage ratio assuming you kept the current interest rate.

Then I realized that f() of each order changes with the price level.  Whether there's some f() that will result in this effect helping or hurting all orders equally and preserving their relative positions in the sequence is an interesting and non-trivial theoretical problem.  If you don't have an f() with this property, then the two pretty graphs are useless and misleading.  I think most f() will not have the necessary property unless you design it in.

Also, keeping some of the BitAssets from the yield fund in a reserve fund in case of a black swan event can be another mechanism, beyond collateral ratios, for the DAC to further reduce black swan risk (at the expense of lower yield rates) without needing to add complication to the shorts.

This is an idea I support, but I think bytemaster's plan for now is, if it becomes a problem, make a hardfork that can spend the "untouchable BitUSD" in the yield fund.  I agree that it should be considered low-priority given the market behavior and the other major changes they're working on, discussing, or have already rolled out.
« Last Edit: October 01, 2014, 07:36:14 am by drltc »
BTS- theoretical / PTS- PZxpdC8RqWsdU3pVJeobZY7JFKVPfNpy5z / BTC- 1NfGejohzoVGffAD1CnCRgo9vApjCU2viY / the delegate formerly known as drltc / Nothing said on these forums is intended to be legally binding / All opinions are my own unless otherwise noted / Take action due to my posts at your own risk