Any pointers on construction. I'm not having a lot of luck on this. For futures where option expiries do not overlap (e.g. ES) I'm getting kinked surfaces (not smoothly increasing or decreasing over distant time). For futures where option expiries overlap (e.g. ED) I am getting different adjusted-to-implied-futures-underlyer curves significantly different for the same expiry (different underlying futures). Averaging among these still leaves an undulating surface along the time dimension. Just to be clear I am trying to construct an implied vol surface (delta-or-moneyness by time) in terms of implied vol of the futures underlying not options underlying (when options and futures expiries are coincident, should be the same).
Kevin: I think there are multiple factors to account for (Assuming I understand you correctly). This is a non-trivial exercise. It may be helpful to simplify your approach to determine where the issue lies. Many people can guess, but you should be able to find for sure. I am still unclear precisely what you mean by IV of the Futures, but not of the Futures options. For a simpler task, consider doing the same for SPX (instead of ES), and first insure the issue with time does not exist there! -- If it does exist there, correct it with SPX first before tackling ES. -- I have done some work with this on SPX, but not ES.
Futures underlying / options underlying are the same re IV. Doesn't matter... If the options have the futures as official underlying, it's still the actual index that's the basis... Where are the 'kinks' at? is it between maturities? Show us your surface please?
In which case, vol curves for two coincident expiries (e.g. for options on EDH17, option base ED, and EDH21 option base E4, both expiring 20170210) should be the same or at least have the same shape and be roughly parallel. This is not the case. Or if it is. then either my vol calcs are off or my marks are off. Mark I have on 20170113 for EDH17 atm call 98.875 is 0.0625 and for EDH21 atm call 97.50 is 0.13. Futures marks are 98.925 and 97.53. Between maturities with different underlying futures expiries but the same ultimate underlying index where expiries don't overlap. For example between ESH17 20170317 expiry and ESM17 20170331expiry. Not a really hard kink, just a reset in curves monotonically increasing in levels. I don't visually inspect the surfacs, just use them for constant maturity calcs such as level, slope, and curvature (anologous to atm straddle, 25d RR, and 25d fly), so I don't have any plots. Matplotlib doesn't work on my machines due to numpy dependency and I have never learned plotting in matlab or R. I will post a plot once I figure out how to get plotting working.
EDH17 exp 13 March 2017 EDH21 exp 15 March 2021 I don't have experience with eurodollar futures or options... but that seems like 2 different expiry dates to me... Also, do you fit your own curves? How is liquidity at later expiry dates?
I don't really get what you're aiming at... what do you mean with overlapping expiries? So where is the big kink exactly? Or do you mean the gradual rise in IV when expiry is later and later dated?
Those are the expiry dates of the underlying futures, the options in question both expire Feb 10th of this year. They settle in different futures, one in the March 2017 contract and one in the March 2021 contract. So the options expire on the same day. They show different implied vols. I am sure there is a good explanation, I just don't know what it is. I fit my own curves which I use to derive indicators used to model the and trade the outrights. I have been doing this on equities and index options for years. I just bought in backhistory of futures options since 2005 and want to test if the same metrics calculated on futures options would be useful to me. BTW, I used to buy these calced numbers (vol surfaces and metrics) from ORATS, but I found I could do it myself more cheaply after they moved to a new pricing model.
Okay, got it... So Eurodollars being in interest rate product... longer dated futures should probably move a lot more correct? At least, so I would assume... similar with bund/bobl/schatz ratios... IV for bund is about 2.5x bobl, and 10x schatz. So, I would think something similar would be for the eurodollar futures... so if the 03-2021 future is more volatile than 03-2017... the IV's would be higher as well. '21 is about 3 to 4 x the '17?
When, for example, an option on an "M" expiry future expires before one or more option expiries of an "H" future. Sort the list below by the third column, option expiry, and you'll see what I mean: EDH17,20170313,20170210 EDH17,20170313,20170313 EDM17,20170619,20170413 EDM17,20170619,20170512 EDM17,20170619,20170619 EDU17,20170918,20170918 EDZ17,20171218,20171218 EDH18,20180319,20170113 EDH18,20180319,20170210 EDH18,20180319,20170310 EDH18,20180319,20180319 EDM18,20180618,20170413 EDM18,20180618,20170512 EDM18,20180618,20170616 EDM18,20180618,20180618 EDU18,20180917,20170915 EDU18,20180917,20180917 EDZ18,20181217,20171215 EDZ18,20181217,20181217 EDH19,20190318,20170113 EDH19,20190318,20170120 EDH19,20190318,20170127 EDH19,20190318,20170210 EDH19,20190318,20170310 EDH19,20190318,20190318 EDM19,20190617,20170413 EDM19,20190617,20170512 EDM19,20190617,20170616 EDM19,20190617,20190617 EDU19,20190916,20170915 EDU19,20190916,20190916 EDZ19,20191216,20171215 EDZ19,20191216,20191216 EDH20,20200316,20170113 EDH20,20200316,20170120 EDH20,20200316,20170127 EDH20,20200316,20170210 EDH20,20200316,20170310 EDH20,20200316,20200316 EDM20,20200615,20170413 EDM20,20200615,20170512 EDM20,20200615,20170616 EDM20,20200615,20200615 EDU20,20200914,20170915 EDU20,20200914,20200914 EDZ20,20201214,20171215 EDZ20,20201214,20201214 EDH21,20210315,20170113 EDH21,20210315,20170210 EDH21,20210315,20170310 EDM21,20210614,20170413 EDM21,20210614,20170512 EDM21,20210614,20170616 EDU21,20210913,20170915 EDZ21,20211213,20171215 EDH22,20220314,20170113 EDH22,20220314,20170210 EDH22,20220314,20170310 EDM22,20220613,20170413 EDM22,20220613,20170512 EDM22,20220613,20170616 EDU22,20220919,20170915 EDZ22,20221219,20171215 I mean when the gradual rise is interrupted by one lower curve (then rises again) when the underlying future (but not the underlying index) changes.
You're right, that is probably the answer. I need to figure out a conversion formula for options on later futures expiries so that I can create a consistent curve going out to 2021. My experience with options on rates derivatives or even rates futures themselves is limited so I'll have to hunt around for the correct formula.