Conversation

3) “traditional model hasn’t had any problems” —CME LOL nickel

300

4) DCOs can have similar bankruptcy protections as FCMs

5) Traditional models only need credit checks because they extend credit We require margin up front

126

6) as others have said, comment about finality is bizarre and seems to forget that we are a centralized exchange and don’t use blockchain for every trade

121

2:35 PM · May 25, 2022Twitter for iPhone

7) glad we’re introducing WMD language to this discussion, really appropriate and well-scoped to DCO margin applications, certainly nothing else recently that would fit it better

Replying to

8) price bands prevent flash crashes. We have price bands. Not sure why some are assuming we wouldn’t.

106

9) no recourse, people keep forgetting this

10) with delayed margin calls you have to choose between liquidating too early and often, needlessly closing down customer positions, or waiting too long, like LME nickel With real time margin you can be precise, preserving customer positions longer while avoiding systemic risk.

110

11) real time margin is risk reducing and protects customers!

109

12) yes the large amount of margin is important in traditional models Which is why we also require margin! We have tens of billions of collateral just for digital assets on FTX.

13) to clarify, I think the risk model generally works well for digitally settled assets; it would require more work to be as appropriate for goods whose settlement is primarily in physical warehouses

14) moderator bringing up improvements in payments settlements instead of improvements in margin settlements It’s a really good point! …why not both?

15) Christine Parker /

@coinbase

making the same point as me, but more eloquently

16) agree also with Parker that intermediation in risk models is different from intermediation in order processing or customer facing

17) re: RJO: there’s a balance on margin call timing; if the current system actually _is_ real then the worries about real time settlement are misplaced

18) agree with ICE that the system should allow people to extend credit to those in danger of margin calls But that shouldn’t be the DCO on behalf of its other customers and members—that is one party posting another’s assets without asking!

19) in our model, others are still able to extend emergency credit to a firm—using their own capital, not other customer’s!

20) frankly the thought of the risk manager for a DCO on the phone with a large customer, choosing to not margin call them and *risking the DCO’s other assets on that call* is scary! It should be forms/FCMs/etc risking their own assets on that call.

21) and crucially, other FCMs/members/customers should not be exposed to that decision

22) one party said, as — I think — a criticism, “if customers can choose whether or not to use an FCM, they won’t” First—that seems to imply FCMs don’t provide any value. I think many do, and so customers would want to keep using them! But also, if customers choose something…

Risk Management, Process, and the Future – An Update

23) more vague allusions to “but what would happen in large moves!” Check out ftxpolicy.com/posts/risk-man; we have empirical data!

ftxpolicy.com

FTX US Derivatives (FUSD) has an application before the U.S. Commodity Futures Trading Commission (the “FTX Application”) that would amend its existing license to operate a clearing house.

24) note, again, that FTX doesn’t rely on credit, and so doesn’t require additional capital backing credit—though it still does absolutely require both initial margin, and a guarantee fund etc.

25) great point by Virtu—FCMs can still provide credit/capital to customers to prevent margin calls! But the capital ends up at the DCO one way or another, so other members aren’t impacted

26) recent comment seems to forget—as is clearly stated in the “stylized facts”, our application, etc.—that we have a guarantee fund in addition to margin

27) also once again ignores all the existing data about risk model performance And in the end, CFTC/DCR are experts in analyzing this and wouldn’t ultimately approve a model they found inadequate

28) really agree with the points Cisewski is making! We need more competition in futures markets. And in the end, the actual performance of the models is what matters.

29) there are already DCOs live without FCMs—this doesn’t require new rule making And we would like to allow for FCMs as well

30) on self-certification and new asset classes—DCR can approve margin models for particular products/asset classes, which limits spread

31) and also CFTC can object to particular self-certs!

32) glad to hear talk from a competitor about needing engagement before self-certs, guessing they would support having a round table discussion before each of their self certs?

33) a lot going on here—simultaneously debating crypto, algorithms, computers, retail, 24/7, etc. most of these already exist on other DCOs, and really unclear what her point is. And weird that she’s skeptical that markets increase liquidity!

34) remember price bands—people are protected from 1 minute market moves! Also—users know _way_ more than any of these people about these markets, ironic to hear the opposite condescendingly said.

35) “retail” is not a monolithic category!

36) good point by Dave Olson—this is the model that many are already used to!