FinEquityVolSurface about financepy HOT 14 CLOSED

Sure. Do you have a view on the strike interpolation scheme ? SABR or polynomial or other ?

from financepy.

I would use a polynomial interpolation - ideally a SVI implementation

from financepy.

from financepy.

OK. I am looking at SVI surfaces.

from financepy.

Section 5.1 of the below article describes a fast and robust implementation of the SVI algorithm:
https://zeliade.com/wp-content/uploads/whitepapers/zwp-008-RobustNoArbSSVI.pdf

First the "Slices" are calibrated producing a set of parameters for each expiry/curve provided as input.

Then section 7.2 and 7.3 describes how to interpolate (or extrapolate) for any expiry using the previously calibrated parameters.

This jupyter notebook describes well the parametrisation: (The first section can be dismissed as implied vols are given as input in our case and not derived from option prices).
http://www.cmap.polytechnique.fr/~demarco/files/calibrationParis6/TP_M2_probFin_implied_vol_parametrisation_solutions.ipynb

from financepy.

OK. I will look at this today.

from financepy.

SSVI would be polynomial in delta.
SSVI is often used with Listed Option prices as input in the literature which requires a first step to imply the Black-Scholes vols from the prices before then moving to the SSVI polynomial fit.
In this case the vols are given as input so first step is not necessary here.

Further useful resources may be found here:
https://wwwf.imperial.ac.uk/~ajacquie/IC_AMDP/IC_AMDP_Docs/Code/SSVI.pdf

Also happy to take a look at SABR. Is it already available in a VolSurface Class method?

from financepy.

from financepy.

SABR is in the SABR class (shifted and basic) under models. I have just used it in FinFXVolSurfacePlus under market->volatility and I am about to use it again for swaptions. On 12/01/2021 08:42, mattdoub wrote: Cool. SSVI would be polynomial in delta. SSVI is often used with Listed Option prices as input in the literature which requires a first step to imply the Black-Scholes vols from the prices before then moving to the SSVI polynomial fit. In this case the vols are given as input so first step is not necessary here. Further useful resources may be found here: https://wwwf.imperial.ac.uk/~ajacquie/IC_AMDP/IC_AMDP_Docs/Code/SSVI.pdf Also happy to take a look at SABR. Is it already available in a VolSurface Class method? — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#63 (comment)>, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABJ73PK5JXOZZMUNTJEMVWTSZP4PJANCNFSM4V5MI4BA.

Apols meant SSVI polynomia in strike not delta...

from financepy.

Just checked in FinEquityVolatilitySurface. It uses Gatheral's SVI to fit the data you provided. See the test case TestFinEquityVolatilitySurface. I will look at SSVI over the next few days. It's already coded up but not yet wired in.

from financepy.

I am closing this for the moment.

from financepy.

I have added an access function to get the vol called

volatilityFromStrikeDate(K, expiryDate):

from financepy.

Looks v good indeed, and it fits well! ;)

Just noticed a tiny bug:
once the curve is calibrated, if you want to see the object in the console it returns an error:
'FinEquityVolSurface' object has no attribute '_atmMethod'

from financepy.

Fixed.

BTW I don't know if I got lucky with the fit as it is unconstrained. I need to build in constraints and no-arb checks. I seeded the longer expiry date fits with the SVI parameters from the previous expiry date which helps.

from financepy.