KarelZe / MC-VAR-Sim

Monte Carlo Sim of a stock portfolio using c++ AMP

MC Sim

This Visual Studio project is my coursework for CMP 202 at Abertay University, Scotland.

underlying concept

Using a Value at Risk approach we try to make a statement similar to this: I am X percent certain there will not be a loss of more than V dollars in the next N days. (Hull, 2012) So X is the confidence level, where V is the loss and n is the holding period.

dS = μ S dt + σ S ds

S = stock price
μ = exp. return
σ = volatility

approach

repeat n times
  repeat t times
    generate normal distributed number
    update end price
  save end price to path array  
extract the 0th quantile
print results

boundary

• Assuming a stock without dividend payments
• single stock portfolio
• using C++ AMP
• extents that don't require padding
• optimized for a NVIDIA GeForce 940MX

dependencies

• Commandline parsing is handled through TCLAP
• Concurrency Visualizer Marker through Visual Studio Extension
• Random number generation through amprng

usage

Using the program is simple:

cd path
REM access help for explanation on cl arguments
C:\path>ConsoleApplication3 --help
REM sample usage of calculation with a tile size of 16 and 512k paths
C:\path>ConsoleApplication3 -x 16 -p 524288 -i 10 -r 0.05 -v 0.04 -d 300 -t 300

sample output:

Todos

• Prepare proposal
• Implement MC kernel
• Implement path array to value at risk functionality
• Optimize memory access, loop unrolling etc.
• measure performance
• Prepare presentation
• Enhance to a multi stock portfolio (optional)
• Download historical data from quandl (optional)
• Calculate volatility using EWMA method (optional)

Contact

Feel free to send me a mail