Sorry to say, this is meaningless unless you are using actual market data. Why not set up a collective2 account something similar then at least you will have an idea if your method makes sense--- https://www.collective2.com/options
Man, it is impossible to make a profitable trading from truly random data. Unless it's something goofy like I could beat the roulette if I could double my bet infinitely until I make my initial $1 bet back. You must have a flaw in your model, my feel is you misprice the options somehow - for example if you don't calculate the volatility correctly, you may end up with premium too high or too low for the underlying asset. If your options pricing get a little off then it's easy to curve fit the whole system to generate profits that quickly balloon into 1000% as you claim. Another mistake might be that you're off by 1 bar somewhere in your calculations. To reveal the flaw you should really test your system against real market data - not only stock data but you must also use the real options price data to see if your system really works.
I think perhaps you should produce some conventional statistics. Using actual data could be highly recommended. http://www.elitetrader.com/et/index.php?search/3265671/&q=options+data&o=date&c[title_only]=1 http://www.elitetrader.com/et/index.php?threads/tick-by-tick-historical-option-data.295583/
I'm not going to argue any further about the system. Some people think no system can ever work. So let them their own belief. Regarding random data: the changes are random (ie. normal distribution/gauss/bell shape), but the result is log-normal --> wiki Regarding Black-Scholes: here is some sample output for verification.
Fact remains that there are pockets of irrationality in markets, and there's a chance that your simulation data was off in some way (doesn't matter which). Until you give your system some paper trading experience with actual market data, it won't be taken seriously. A month should suffice to at least start demonstrating what works as expected and discovering what might not.
Here is a procedure to backtest such a system: http://www.elitetrader.com/et/index.php?threads/a-proposal-for-testing-black-box-systems.296945/ The interested party would feed the system remotely with data, and will get instant response about the decision the system takes (ie. open/close/adjust pos). The data can be historic or simulated, and the whole test (1 year) could be finished in less than 2 hours. This is my offer to all interessted parties: just send the tick or bar data one by one over a TCP socket connection to the system and receive on the same line the activity data of the system again immediately one by one. By this, the customer side can be sure that the system works as announced.
Here's GBM formula and Excel code for generating simulated stock prices: http://investexcel.net/geometric-brownian-motion-excel http://investexcel.net/wp-content/uploads/2011/10/Geometric-Brownian-Motion.xlsx
You might have some discrepancy between the data you generate and the parameters in the option calculation. For example the Excel spreadsheet you refer to shows volatility 0.07 and drift 0.1 while the black-scholes formula you refer to shows volatility 0.3 and no drift. Are you sure the data you generated for your test didn't have any drift in their price?
Come on please! The Excel stuff is just an example I found yesterday on the net, you can change its input params. And: it is not from me, I just thought that it maybe could be of some use for others to play around with GBM. I myself am doing this stuff now for several years, albeit not in Excel but in C++. And, yes, the drift param in my setup is 0.