Author Topic: First BitShares-sponsored research paper submitted for review  (Read 1329 times)

0 Members and 1 Guest are viewing this topic.

Offline biophil

  • Hero Member
  • *****
  • Posts: 880
  • Professor of Computer Science
    • View Profile
    • My Academic Website
  • BitShares: biophil
First BitShares-sponsored research paper submitted for review
« on: August 02, 2019, 05:46:43 pm »
I've submitted a draft of a research paper to the Decentralized 2019 conference in Athens, Greece (the same one that is sponsored by clockwork's worker proposal). If the paper is accepted by the conference's peer review committee, it will be published as the first BitShares-sponsored academic research, and I will travel to Greece to present the research at the conference. I'm providing a link to a draft of the research paper, but before I do, let me clarify a few things:
  • The paper considers an extremely oversimplified model of the BitShares system. The reason for this is threefold: 1) I wrote the paper in only a few weeks, so time constraints limited the amount of detail I could include. 2) At the start of a project, you really don't want to draw the wrong conclusions just because you rushed an overly-complex model to press. 3) A simple model often illuminates interesting points and helps guide the way forward.
  • One issue that the paper illuminates is that the BitShares incentive mechanism is probably very sensitive to our choices of MCR and MSSR, in the sense that sometimes if you change one of these parameters a little, it could have a huge effect on the behavior of BitAsset shorters.
  • The specific numbers used in the paper (MSSR= 1.005, 1.01, 1.02 and MCR= 1.4, 1.5, 1.6) are to be taken with a grain of salt. I used them to illustrate the general shape of the problem, not to draw any specific conclusions about those particular parameter choices.
Having given those caveats, I welcome your comments on the paper draft itself, available here.

Finally, the worker proposal which will fund ongoing research in this area needs a few more votes! Please vote for worker 1.14.204 or ask your proxy to vote for it.
Support our research efforts to improve BitAsset price-pegging! Vote for worker 1.14.204 "201907-uccs-research-project."

Offline pc

  • Hero Member
  • *****
  • Posts: 1492
    • View Profile
    • Bitcoin - Perspektive oder Risiko?
  • BitShares: cyrano
Re: First BitShares-sponsored research paper submitted for review
« Reply #1 on: August 03, 2019, 10:54:43 am »
Wow, excellent!

I encourage everybody to support this worker proposal. I think proper research on this matter could produce very interesting and helpful results.
Bitcoin - Perspektive oder Risiko? ISBN 978-3-8442-6568-2 http://bitcoin.quisquis.de

Offline abit

  • Committee member
  • Hero Member
  • *
  • Posts: 3703
    • View Profile
    • Steemit Blog
  • BitShares: abit
  • GitHub: abitmore
Re: First BitShares-sponsored research paper submitted for review
« Reply #2 on: August 03, 2019, 12:17:20 pm »
I convinced cn-vote to vote for the worker.
BTS account: abit
BTS committee member: abit
BTS witness: in.abit

Offline bitProfessor

Re: First BitShares-sponsored research paper submitted for review
« Reply #3 on: August 03, 2019, 02:57:34 pm »
I convinced cn-vote to vote for the worker.
Good

Offline Digital Lucifer

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 310
  • 13 years of being Slackware abUser
    • View Profile
    • BitShares 3.0
  • BitShares: dls.cipher
  • GitHub: dls-cipher
Re: First BitShares-sponsored research paper submitted for review
« Reply #4 on: August 03, 2019, 08:10:11 pm »
I've submitted a draft of a research paper to the Decentralized 2019 conference in Athens, Greece (the same one that is sponsored by clockwork's worker proposal). If the paper is accepted by the conference's peer review committee, it will be published as the first BitShares-sponsored academic research, and I will travel to Greece to present the research at the conference. I'm providing a link to a draft of the research paper, but before I do, let me clarify a few things:
  • The paper considers an extremely oversimplified model of the BitShares system. The reason for this is threefold: 1) I wrote the paper in only a few weeks, so time constraints limited the amount of detail I could include. 2) At the start of a project, you really don't want to draw the wrong conclusions just because you rushed an overly-complex model to press. 3) A simple model often illuminates interesting points and helps guide the way forward.
  • One issue that the paper illuminates is that the BitShares incentive mechanism is probably very sensitive to our choices of MCR and MSSR, in the sense that sometimes if you change one of these parameters a little, it could have a huge effect on the behavior of BitAsset shorters.
  • The specific numbers used in the paper (MSSR= 1.005, 1.01, 1.02 and MCR= 1.4, 1.5, 1.6) are to be taken with a grain of salt. I used them to illustrate the general shape of the problem, not to draw any specific conclusions about those particular parameter choices.
Having given those caveats, I welcome your comments on the paper draft itself, available here.

Finally, the worker proposal which will fund ongoing research in this area needs a few more votes! Please vote for worker 1.14.204 or ask your proxy to vote for it.

As I said before even worker was around - PROPER ACADEMIC RESEARCH can only bring BENEFITS! Please let us know once draft becomes final and when we can(if we can) upload it to bitshares.org under Documentation (with full source)

Please continue

And please VOTE FOR THIS WORKER IF YOU HAVEN'T.

Worker ID: 1.14.204   
Name: 201907-uccs-research-project
URL to escrow/details

Chee®s
Milos (DL) Preocanin
Owner and manager of bitshares.org through Consensus
Move Institute - RN: 2098555000
Murska Sobota, Slovenia, SI.

Offline sschiessl

Re: First BitShares-sponsored research paper submitted for review
« Reply #5 on: August 03, 2019, 10:33:22 pm »
In what way/form could I give feedback on the paper?

Offline binggo

  • Hero Member
  • *****
  • Posts: 1366
    • View Profile
Re: First BitShares-sponsored research paper submitted for review
« Reply #6 on: August 03, 2019, 10:35:03 pm »
hehe

Offline biophil

  • Hero Member
  • *****
  • Posts: 880
  • Professor of Computer Science
    • View Profile
    • My Academic Website
  • BitShares: biophil
Re: First BitShares-sponsored research paper submitted for review
« Reply #7 on: August 05, 2019, 09:25:51 pm »

As I said before even worker was around - PROPER ACADEMIC RESEARCH can only bring BENEFITS! Please let us know once draft becomes final and when we can(if we can) upload it to bitshares.org under Documentation (with full source)


We're excited to keep it going!

I will let you know about hosting it on bitshares.org; each publisher has different rules about precisely how various papers can be hosted where. However, anywhere I publish should allow full version of the paper to be hosted on https://arxiv.org/, which I will do if this paper is accepted by the Decentralized conference.
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
Re: First BitShares-sponsored research paper submitted for review
« Reply #8 on: August 05, 2019, 09:27:59 pm »
In what way/form could I give feedback on the paper?

Feel free to email me at philip.brown (at) uccs.edu.
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
Re: First BitShares-sponsored research paper submitted for review
« Reply #9 on: August 05, 2019, 09:29:54 pm »
I convinced cn-vote to vote for the worker.

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

Offline bitcrab

  • Committee member
  • Hero Member
  • *
  • Posts: 1567
    • View Profile
  • BitShares: bitcrab
  • GitHub: bitcrab
Re: First BitShares-sponsored research paper submitted for review
« Reply #10 on: August 06, 2019, 10:13:06 am »
have gone through the paper without checking exactly the derivation of the formulas, just try to find what sense does the conclusion make.

it seems:

it make sense to set MCR<1.53, to make the optimal collateral ratio "snaps" to the right.

while MCR<1.53, then either MSSR=1.01 or 1.005 make little difference.

anyway, as the conclusion based on a simple model, so we need to consider more factors to evaluate the effect while changing these parameters.


Offline biophil

  • Hero Member
  • *****
  • Posts: 880
  • Professor of Computer Science
    • View Profile
    • My Academic Website
  • BitShares: biophil
Re: First BitShares-sponsored research paper submitted for review
« Reply #11 on: August 07, 2019, 03:47:00 pm »
have gone through the paper without checking exactly the derivation of the formulas, just try to find what sense does the conclusion make.

it seems:

it make sense to set MCR<1.53, to make the optimal collateral ratio "snaps" to the right.

while MCR<1.53, then either MSSR=1.01 or 1.005 make little difference.

anyway, as the conclusion based on a simple model, so we need to consider more factors to evaluate the effect while changing these parameters.

It depends on what you're trying to do, and you're right - more study is needed.

If your goal is to incentivize low collateral ratios (more borrowing), then you need to set MCR low enough (in the paper, less than 1.53) to get that to happen.

However, low collateral ratios intrinsically increase the risk of undercollateralization, which is something that should probably be avoided.

On the one hand, low MCR should lead to tighter pegging; on the other hand, low MCR should lead to more risk of undercollateralization. This suggests that there is a tradeoff between tight pegging and solvency, which needs to be studied further.

For the research, we're starting to work up more-expressive models that should help capture these tradeoffs more clearly.
Support our research efforts to improve BitAsset price-pegging! Vote for worker 1.14.204 "201907-uccs-research-project."

Offline bitcrab

  • Committee member
  • Hero Member
  • *
  • Posts: 1567
    • View Profile
  • BitShares: bitcrab
  • GitHub: bitcrab
Re: First BitShares-sponsored research paper submitted for review
« Reply #12 on: August 07, 2019, 05:10:18 pm »
have gone through the paper without checking exactly the derivation of the formulas, just try to find what sense does the conclusion make.

it seems:

it make sense to set MCR<1.53, to make the optimal collateral ratio "snaps" to the right.

while MCR<1.53, then either MSSR=1.01 or 1.005 make little difference.

anyway, as the conclusion based on a simple model, so we need to consider more factors to evaluate the effect while changing these parameters.

It depends on what you're trying to do, and you're right - more study is needed.

If your goal is to incentivize low collateral ratios (more borrowing), then you need to set MCR low enough (in the paper, less than 1.53) to get that to happen.

However, low collateral ratios intrinsically increase the risk of undercollateralization, which is something that should probably be avoided.

On the one hand, low MCR should lead to tighter pegging; on the other hand, low MCR should lead to more risk of undercollateralization. This suggests that there is a tradeoff between tight pegging and solvency, which needs to be studied further.

For the research, we're starting to work up more-expressive models that should help capture these tradeoffs more clearly.

Yes, I believe we need to incentivize more USD borrowing at this moment, that's why I suggest to change USD MCR to 1.5, unfortunately it hasn't gotten enough support.

Another factor is that with GS protection, we have minimized the pain of undercollateralization, in recent past we have watched that both bitCNY and bitUSD have experienced GS protection,  with no obvious impact to the market.
« Last Edit: August 07, 2019, 05:11:50 pm by bitcrab »

Offline Customminer

  • Hero Member
  • *****
  • Posts: 703
  • Bitshares FTW!
    • View Profile
  • GitHub: grctest
Re: First BitShares-sponsored research paper submitted for review
« Reply #13 on: August 17, 2019, 03:26:09 pm »
Great work biophil 👍

Supporting your worker proposal should be a high priority for all 👌