Xcelerit Runs the World’s Fastest Monte-Carlo Option Pricing Computation
Xcelerit software demonstrated a record speed in a Monte-Carlo simulation on a 1U server for European-style options using the industry’s fundamental pricing model, known as the Black-Scholes model.
Xcelerit announced the world’s fastest execution of a Monte-Carlo option pricing algorithm (Black-Scholes model) on a single unit rack-mounted system.
1st of all I don’t know that there is some industrial standard for option pricing benchmark(s).
Moreover, I didn’t find any comparison w/other GPU-based option pricing calculations.
And AFAIK, Monte Carlo simulation is quite different model than Black-Scholes. Last is based on partial differential equations which have simple analytical solving.
BTW, how actual are calculations for millions of option pricing – i.e. it’s interesting only for trading using software robots or for “human trading” also ?
euler-maruyama discretization or other discretization with faster convergence (like runge-kutta or modified duffy)? and SDE only geometrical brownian motion or generalized SDE in multi-dimensional space? 😀
The Black-scholes pricing model can be solved either with a partial differential equation or with a Monte Carlo simulation, which turn to be more efficient when the number of underlying is high.
Also Monte Carlo is an excellent example of simulation with high parallelization ratio because there’s no communication between the different instances. The only drawbacks (i not sure here 😀 ) are the number of repetitions which is directly related to the precision of the final result and the generation of random numbers.sorry, all stochastic models can be solved with that tools, not only bs! and all can be parallelized fine