We develop a MATLAB code for pricing plain vanilla European options with underlying dynamics driven by Black-Scholes-Merton model (BSM) endowed with jumps, stochastic volatility and multi-factor Gaussian interest short rates. We take advantage of a Monte-Carlo method (MC) known in literature as dimension-reduction Monte-Carlo (drMC). We measure the efficiency of the drMC with respect to the basic MC through numerical experiments conducted on three- and six-factor models. Starting from the drMC pricing machinery, we implement a code for calibrating the three-factor model to market data. The peculiarity of the drMC is the reduction of the number of simulated processes from all those driving the model to only one. The reduction of the number of simulated processes reduces the variance of the final prices and the computational cost. The technique for reducing the number of randomness sources requires to condition iteratively the canonical expectation of the discounted payoff on all those sources of randomness (Brownian motions, BM) associated to the interest rates and to the stochastic volatility. This procedure gives back the BSM dynamics endowed with jumps, which has a known Partial-Integro Differential Equation (PIDE) associated to. The PIDE is solved via Fourier Transform in the frequency domain. The Antitransform of the solution in the frequency domain is either computed numerically through the Fast Fourier Transform algorithm, or analytically, whether the jump-size distribution of the model provides a closed-form solution. The Gaussian nature of the interest rates also allows a closed-form solution for part of the conditional expectations, thus relaxing the conditioning up to the BM associated to the volatility process only. As a consequence, the simulations finally reduce to one single source of randomness: the BM associated to volatility.
Sviluppiamo un codice MATLAB per il prezzaggio di opzioni europee plain vanilla con dinamica del sottostante guidata dal modello di Black-Scholes-Merton (BSM) arricchito con salti, volatilità stocastica e tassi di interesse multi-fattore gaussiani. Ricorriamo ad un particolare metodo Monte-Carlo (MC) noto in letteratura sotto il nome di dimension-reduction Monte-Carlo (drMC). Diamo una misura dell’efficienza del drMC rispetto al MC tramite esperimenti numerici condotti su modelli a tre e sei fattori. Basandoci sul meccanismo drMC per il prezzaggio, implementiamo un codice per la calibrazione del modello a tre fattori su dati di mercato. La particolarità del drMC è la riduzione del numero di processi simulati da tutti quelli che costituiscono il modello a uno soltanto. La riduzione del numero di processi simulati ha una doppia valenza: riduce la variabilità dei prezzi finali e riduce il costo computazionale. La tecnica di riduzione delle sorgenti di aleatorietà su cui si basa il drMC prevede di condizionare iterativamente la canonica attesa del payoff scontato rispetto a tutte quelle sorgenti di aleatorietà (moti Browniani, BM) associate ai tassi di interesse e alla volatilità stocastica, riottenendo così la dinamica BSM arricchita con salti per la quale è nota l’equazione integro-differenziale alle derivate parziali (PIDE) associata. La PIDE è risolta tramite trasformata di Fourier nel dominio delle frequenze. L’antitrasformata della soluzione nel dominio delle frequenze viene calcolata con l’algoritmo Fast Fourier Transform, o analiticamente qualora la distribuzione dei salti lo permetta. La natura gaussiana dei tassi di interesse permette di calcolare analiticamente parte delle attese condizionate e dunque di rilassare ulteriormente il condizionamento fino al solo BM associato al processo della volatilità. Le simulazioni si limitano pertanto ad una sola sorgente di aleatorietà: il BM associato alla volatilità.
Dimension reduction Monte Carlo : a code implementation and calibration to market data
CIPOLLONI, PIETRO
2017/2018
Abstract
We develop a MATLAB code for pricing plain vanilla European options with underlying dynamics driven by Black-Scholes-Merton model (BSM) endowed with jumps, stochastic volatility and multi-factor Gaussian interest short rates. We take advantage of a Monte-Carlo method (MC) known in literature as dimension-reduction Monte-Carlo (drMC). We measure the efficiency of the drMC with respect to the basic MC through numerical experiments conducted on three- and six-factor models. Starting from the drMC pricing machinery, we implement a code for calibrating the three-factor model to market data. The peculiarity of the drMC is the reduction of the number of simulated processes from all those driving the model to only one. The reduction of the number of simulated processes reduces the variance of the final prices and the computational cost. The technique for reducing the number of randomness sources requires to condition iteratively the canonical expectation of the discounted payoff on all those sources of randomness (Brownian motions, BM) associated to the interest rates and to the stochastic volatility. This procedure gives back the BSM dynamics endowed with jumps, which has a known Partial-Integro Differential Equation (PIDE) associated to. The PIDE is solved via Fourier Transform in the frequency domain. The Antitransform of the solution in the frequency domain is either computed numerically through the Fast Fourier Transform algorithm, or analytically, whether the jump-size distribution of the model provides a closed-form solution. The Gaussian nature of the interest rates also allows a closed-form solution for part of the conditional expectations, thus relaxing the conditioning up to the BM associated to the volatility process only. As a consequence, the simulations finally reduce to one single source of randomness: the BM associated to volatility.File | Dimensione | Formato | |
---|---|---|---|
2018_07_Cipolloni.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
1.15 MB
Formato
Adobe PDF
|
1.15 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/141752