A novel data-driven neuro arch (DDNA) model for option pricing on cloud

Unpredictable nature of financial derivatives market has always been a challenge for investors to profit from these markets. It is known that computing volatility of underlying assets is a source of this challenge. In this work, we present an application of our data-driven neuro arch (DDNA) volatility model that is driven by the market data of the stock prices to price options. That is, we use the volatility forecast from our DDNA model along with the Monte Carlo (MC) simulations to compute option prices. Since the MC method requires a large number of simulations for better precision, we implement the proposed model on two cloud resources (Amazon’s EMR and Google’s Cloud DataProc) using the Hadoop MapReduce paradigm. Also, since the MC strategy is prone to errors due to uncertainties and random numbers, we propose to generate a fuzzified range of option prices instead of a single crisp option value to minimize these errors. Our experimental configuration consists of c3.xlarge instances and n1-standard-4 instances, both having 4 vCores, for EMR and GDP respectively. For a largest number of 10 million simulations on 40 VMs, the option pricing results (computed in 39 and 33 s respectively on EMR and GDR) are close to the last traded option value found on option chain tables with an error of 0.00101 (0.1%). The proposed DDNA model for forecasting volatility together with MC option pricing model implemented on MapReduce outperforms the existing option pricing models in terms of efficiency and accuracy. This proposed strategy could be used by investors for computing option prices precisely with relative ease, allowing them to value the numerous available option contracts for their investment decisions.