BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:www.tcs.tifr.res.in/event/269
DTSTAMP:20230914T125917Z
SUMMARY:Nearest Neighbor Method in Simulation based Approximate Dynamic Pro
 gramming for Option Pricing
DESCRIPTION:Speaker: Ankush  Agarwal\n\nAbstract: \nAs is well known\, opti
 ons and other financial derivatives acquire their value from the underlyin
 g assets such as stocks and bonds and are traded extensively in the financ
 ial markets. We consider the problem of pricing an American option\, that 
 is an option with an early exercise feature. This problem  can be formula
 ted as an optimal stopping time problem for stochastic processes. In high 
 dimensions\, typically no closed form solution is available for  American
  option prices and numerical methods relying on solving associated partial
  differential equations are also not viable. Recently there has been consi
 derable research in developing simulation methods to solve this optimal st
 opping time problem via approximate dynamic programming. We overview commo
 nly used pricing methods that use nested simulation and regression based t
 echniques. We propose a new pricing algorithm wherein nearest neighbor est
 imator is used to estimate the continuation value functions required for t
 he American option pricing. We derive asymptotic mean square error (MSE) o
 f the option price estimator and find the optimal parameter for the neares
 t neighbor estimator that minimizes this MSE. This asymptotic MSE decays t
 o zero as the allocated computational effort increases to infinity. We als
 o discuss the impact of dimensionality on this rate of convergence.\n
URL:https://www.tcs.tifr.res.in/web/events/269
DTSTART;TZID=Asia/Kolkata:20120424T140000
DTEND;TZID=Asia/Kolkata:20120424T153000
LOCATION:AG-66 (Lecture Theatre)
END:VEVENT
END:VCALENDAR