Siva

Algo trading bitcoin, ethereum and other liquid crypto. All things #DeFi. And maybe other stuff.

[Siva @8bitbytebit]

@SGJohnsson @danrobinson You're right about the offsetting positions issue and the lack of transparency.

I think you are under weighting both how much an exchange should avoid having to use ADL at all and also the value of impacting the least amount of people as possible.

[Siva @8bitbytebit]

@SGJohnsson @danrobinson Think of it in terms of moral hazard. No exchange wants to employ ADL unless absolutely necessary so a well designed ADL queue should act as to minimise the chance of it being needed.

[Siva @8bitbytebit]

@SGJohnsson @danrobinson I'm saying that pro rata ADL is unfair.

Having an ADL queue and sorting by leverage is fair because it focuses the "pain" on the source of exchange insolvency risk -- high leverage positions.

[Siva @8bitbytebit]

@SGJohnsson @danrobinson It's unfair. It punishes every account/position instead of focusing on the ones carrying the highest leverage.

It's being overly leveraged, whether long or short, that creates the insolvency risk for the exchange in the first place.

[Siva @8bitbytebit]

@0xnagu @danrobinson @tarunchitra For the same reason -- punish insolvency risk generators -- I also think positions with large unrealised pnl should not be higher in the queue.

[Siva @8bitbytebit]

@0xnagu @danrobinson @tarunchitra Understood, thanks. I thought you meant that Tarun was right in the normative sense.

IMO the redistribution of equity should be uneven. Specifically punishing opposing positions that are also highly leveraged because that is what generates exchange level insolvency risk.

[Siva @8bitbytebit]

@danrobinson Why is unrealised pnl used for ADL ranking anyway? It incentivises the wrong thing as it punishes "good" trading on average. Same for pro rata schemes.

Just prioritise highly leveraged positions only in an ADL queue. That's what drives solvency risk in the first place.

[Siva @8bitbytebit]

@ConcretumR Unbelievable. Thank you so much for this.

[Siva @8bitbytebit]

@danrobinson The trilemma is the most treasured part of Tarun's argument. Sad really.

You stated somewhere in the storm of tweets that you are in favour of pro-rata equity haircuts. Why?

Pro-rata seems dangerous to me on moral hazard grounds.

[Siva @8bitbytebit]

@fiddybps1 @nfhen any thoughts on this 😂

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ Check your DMs please. This conversation will go much better there.

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ Ok, great. In the case of it being 8am UTC is there at that precise moment (and no other) an option on deribit that matches exactly with each paradex option. Same price according to BSM and same Greeks, but leaving theta out of it for the moment. (I'll come back to it shortly.)

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ For the sake of being super careful. Two options have the same underlying, same strike, same time to expiry. If they have the same Greeks will they have the same price and vice-versa?

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ I find it interesting that you are unwilling to say "yes" to my question now.

Breaking it down further and please be direct for the sake of clear communication:

If two options have the exact same spot/Greeks (all of them), should they have the same price and vice-versa?

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ I think you are confusing replication with pricing. Let's break it down step by step please:

Do we agree that at precisely 8am UTC a deribit option with expiry in 24h should have the same price as a paradex perp option. All else same (spot, IV, strike, etc.).

Yes or no?

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ It must not be what you have now. The gradient of the linear interpolation of the time value of the option going to zero over 24 hours.

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ But every moment in time is a moment when there is a theoretical dated option expiring in 24 hours time that has a theta value. So the instantaneous funding cost for the perp option has to be that theta value.

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ I think you're caught up in replication. Pretend that deribit has options that expire every second of the day. Each passing second there is one deribit option that is identical. It expires in 24h. Shouldn't the paradex option have a funding cost equal to that option's theta?

[Siva @8bitbytebit]

@nfhen @fiddybps1 @wufasa_ From your docs:

"If Alice buys 5 BTC-USD-101000-C. After 1 hour and if market prices are unchanged, Alice will have a Funding PnL of -5 * (500 * 1 / 24) = -$104.17"