Given market tick data for a single tenor for a futures and a set of options on the futures, how can I say if the IVs in the market at this point is sticky strike or sticky delta?
I was trying to solve the problem looking at change in the futures price and the comparing the change in IVs of the options but have run into issues when the options quote change is not in sync with the futures prices. I was wondering if anyone here has worked on similar problem and can share their ideas?
I work in a relatively illiquid and old-fashioned market (options on power), where trades are arranged via phone & broker, so the issue of low underlying liquidity is definitely there. To remedy this, all options are dealt with delta hedge, where the price level of the delta hedge is pre-agreed, so market moves during arrange a trade do not matter as much (unless of course they are very substantial).
In your case, I would refer to end-of-day quotes, where in the case of exchange-traded options, you have closing prices for options and futures. In this case, the exchange will probably poll several dealers in order to give a realistic market picture. In OTC markets, brokers will show end of day option rates, and explicitly reference them to a closing price of the underlying.
As for judging the mode of behaviour (sticky strike vs. sticky delta) intraday, I would be cautious. Imho, if you base your hedging decisions on this, you may overengineer, potentially not doing yourself a favour.
I have mostly been working on the assumption that a rangebound market with very modest moves will be sticky strike, whereas during more volatile periods it will behave in a sticky-delta way. Not having tested this explicitly, I would say you could try to look for a criterion along the lines of:
$S\sigma/\sqrt{252}\gg\mathit{daily move}$ (sticky strike) resp. $S\sigma/\sqrt{252}\ll\mathit{daily move}$ (sticky delta)
What you could do to make this into a more sound methodology is to run volatility analysis on end-of-day data and relate to daily moves.
I believe you won't be able to infer much, as sticky-delta versus sticky-strike is defined by the model and market-maker. And this pre-defined sticky-delta/strike is then complicated by the actual market-moves.
Consider, I've calibrated my model to existing market-prices in terms of IV (implied vols).
If I've defined my vol-surface, as ATM-vols and OTM risk-reversals/butterflies; then I make an explicit assumption (usually for stress-tests and delta/vega/gamma hedging) on whether to use sticky-delta, or strike * if * the markets were to move; in the generation of new implied-vol-surface. This is at time, T =0.
Now, at the next time-slice, and markets * have * moved, the new implied vol-surface will reflect the market-consensus of prob.dist at T = 1. If it was purely sticky-delta, the new ATM vol will be at the same level; if it was purely sticky-strike, the new ATM vols will be the level implied at strike K at time T= 0.
But the above never happens. Hence, market vols move in a behaviour where it is a mixture between the sticky-strike and sticky-delta. But it is the market-maker who will know (and have set) their models was from a sticky-strike/delta behaviour.
That's my two cents. It's possible to calibrate out the general ratio (0 to 1) of sticky-strike/delta the market is behaving as; but it is more involved.
Hope that helps!
