0
$\begingroup$

There are many stochastic volatility option models not only require significant more computation/simulation comparing to the standard BSM model but also introdue large source of possible problems at model-calibration.

A quick computation on some real stock option prices data give me the impression that stochastic volatility models fail to demonstrate superiority over BSM models in a statisically meaningful manner.

Am I get the wrong impression here?

$\endgroup$
  • 1
    $\begingroup$ Models are usually calibrated to match prices of liquid instruments, typically vanilla options, so I don't see your point. A real test would be to measure if, using a model, you are able to lock the premium when you sell an option and delta hedge it under your working modelling assumptions. Also, model complexity plays a role when pricing exotic options. $\endgroup$ – Quantuple Feb 2 '17 at 12:55
  • $\begingroup$ @Quantuple, well the reason is we do simulation of assert price paths and want a reliable pricing model to translate the assert prices to its option prices at that path, and we try to calibrate models and use their parameters to compute option prices of the assert in the next few time steps, but we found this method work actually poorly or at best, half-half, comparing to the standard BSM. $\endgroup$ – user2188453 Feb 2 '17 at 13:03
  • $\begingroup$ I'm against for closing. The question is clear. $\endgroup$ – SmallChess Feb 3 '17 at 11:36
  • $\begingroup$ Are you aiming to calibrate a model to current option prices, such that you can accurately predict option prices conditional of the future spot price? Ie you want a model that has the property of being accurate after the spot changes? $\endgroup$ – will Feb 6 '17 at 23:59
  • $\begingroup$ @will exactly, we just need a translator that can translate spot price into option price $\endgroup$ – user2188453 Feb 8 '17 at 14:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.