Perpetual Volatility Swaps

[schwofi]

Hi all. I have been (mostly passively) following Vega for a while and was invited to begin the discussion here on Perpetual Volatility Swaps. Here are some initial considerations;

The idea with this class of products is to bring transparent & efficient Greeks exposure on an arbitrary underlying asset to Vega. Some advantages: -Effectively pure (realized) volatility exposure. -No roll costs (self-rebalancing) -Can recreate any (vertical) option/strategy defined in terms of Delta, Moneyness, etc. -Efficient Margining -Greatly simplifies task of making markets on assets where demand is untested (ie low barrier to entry)
-Unique in the space (Are there any comparable centralized product offerings across TradFi or Crypto)?

While there are papers and blog posts that dance around this idea (can google XPOs (Perpetual Options) or see this paper that Paradigm put out: https://www.paradigm.xyz/papers/everlasting_options.pdf) for the most part these contemplate contracts with a fixed strike (Paradigm paper does mention the floating-strike variant almost as an after-thought). Contracts designed to have adaptive strikes seem generally more desirable as they solve position management inefficiencies in the vertical (price) dimension, as well as horizontal (time).

Possible mechanisms: -Trade price is the funding rate -Payout curves can be customizable, incorporating potentially multiple strike levels (eg Strangles, Call Spreads, Ratio Spreads, etc) -Contracts can be defined to settle at a desired regular interval (eg 1min, 30min, Daily). Additionally its conceivable a price threshold could trigger settlement as a measure to further increase short-side margin efficiency -Underlying price pulled from oracle at settlement, which serves as the strike price for the subsequent interval as well as determining the realized volatility payout; Settlement or TWAP of contract for the previous interval may determine funding payout (if ratio spreads are modeled funding and volatility payouts could be positive or negative) -A Latency of 1 settlement interval to participate in payouts on either side is a possibility; it would remove any incentive to 'trade' the settlement, keeping potentially undesirable noise out of the price action

The biggest issues seem to be around risk modeling and fair settlement. Though non-linear payout contracts may present challenges, any such product should be considerably cheaper to margin than its fixed strike, fixed-expiry equivalent.

[davidsiska-vega]

There are two issues I see that would need a bit of work: 1) Oracle. With the oracle framework we have it may be possible to ingest just the price feed for the underlying and have the realised volatility calculated as a "filter" on the oracle framework. That's definitely solvable. 2) Risk model. I think we'll be fine with a Heston model but I have to check the details.

By the way, from your point of view @schwofi, is there much difference between the usefulness of a volatility swap and a variance swap? The reason why ask is that under the Heston model I have an explicit formula for the price of the variance swap but not the volatility swap ;-).

[Lucas-Kohorst]

Hey!

I am also new to vega, and thought this was an interesting discussion. There are actually a few more papers on this topic specifically constant greek exposure, or automated / self-rebalancing options.

Constant Gamma Exposure with a Power Perpetual, this is further detailed in Raw Moment Derivatives, Unbundling Gamma, and Power Claims.

[schwofi]

Var/Vol My thinking on the variance/volatility question is pragmatic. The payout of a variance swap is non-linear (see: http://sp-finance.e-monsite.com/pages/variance-swaps/mechanics/convexity.html) which may be desirable but in many cases is not, for example when replicating individual options & simple option strategies. Linear payouts should also much simpler from a risk modeling standpoint. Margin requirement for selling a "Perp ATM Straddle" (terminology tbd!), for example, would be similar to the underlying asset's requirement (ie roughly equivalent to a Perpetual Future based on that underlying).

Risk One possibility (though I acknowledge it may seem problematically circular) may be to define the margin as a function of the contract price itself (ie the funding rate/implied volatility), since, assuming some reasonable liquidity threshold, this may well give a more accurate/safe measure of near-to-long term risk vs attempting to define a static vol parameter to insert into a model. Governance could decide/vote on a scalar or linear offset/buffer to the market view of volatility which is in turn fed into Heston or other. I am definitely no expert in this area, but it seems desirable that margin parameters float with market conditions to some extent, and this seems like an interesting mechanism for 'staying current' (and may be conservative enough, since risk premiums tend to fall less and rise more than realized market volatility in the medium-to-long term).

Dev What is the strategy for zeroing in on a spec? It seems to me the cornerstone contract is the Straddle, as it provides maximum capital efficiency for liquidity providers and can be used in combination with a corresponding Futures contract to take a Call or Put-side position without need for the additional instruments. Might this be a good place to start?

Liquidation/Settlement A technical question: with Perp Futures, is funding typically paid to trader's wallet/balance at each settlement, or do funding debits flow to a pool that is drawn from when trader closes position? I am curious since it seems possibly simpler for volatility payouts to flow to a pool that is drawn from when long positions close out, rather than to all open positions @each settlement, since, in the event of a liquidation, short positions may, on a given trade, not be able to meet their obligation to buyer.

Naming Conventions "Perpetual Floating-Strike Volatility Swaps" is something of a mouthful. Maybe Vega/Vol Perps, or just V-Perps?

[schwofi]

(sorry didn't mean to close)

Just saw Lucas' post which serendipitously applies to the discussion around exponential vs linear payouts! The power perp appears to be a directionally biased variance swap, mimicking holding a strip of options out into infinity, which may be the exposure one is looking for, and may not. In any case, due to the tendency of such assets to blow up in extreme conditions, implementing good risk controls may be a challenge.

Down the road, expanding the asset class to mimic arbitrary higher order Greek responses curves to price movement is definitely a powerful idea. Constant Gamma is just one such construct; another may be Constant Speed (Gamma of Gamma), which would have an even crazier payout curve as the Gamma would increase linearly with price movement (eg ETH^2^2). So I think the Paradigm paper's suggestion that the power perp is the generalized options contract and can absorb much of existing markets' liquidity is not accurate. Though, to be fair, it is quite attractive in the sense that the Straddle variant (ie variance swap) consolidates the entire options skew in a single instrument.

[davidsiska-vega]

Hi @schwofi the risk for any product on Vega is the expected shortfall of the derivative position at a future time \tau > 0 in whatever risk model you choose. See Section 6.1 in https://vega.xyz/papers/vega-protocol-whitepaper.pdf. The challenge here is to come up with a nice model (Heston will work). The main risk parameters would in that case be mean reversion speed, mean reversion level and vol of vol.

[davidsiska-vega]

By the way, when discussing derivatives it would be really good to actually write the payoff function so e.g. for a variance swap this would be

[davidsiska-vega]

The reason for this is that I am not up-to-date with all the jargon and I get confused easily. When I have a formula I at least have a chance to understand correctly.

[davidsiska-vega]

By the way, if you want to see a bit more about the design principle of margins on Vega you can find it in: https://vega.xyz/papers/margins-and-credit-risk.pdf