Appreciate your very thoughtful response. I am not strapped for cash but don't like to pay if I can get them for free. Also my broker provides historical data, just don't know how accurate it is. If I use historical stock data for back test, should I use actual or adjusted? Regards,
That's up for debate. The answer is "whatever you'll be trading on". The prices you see on your charts are probably back-adjusted for splits, etc. I'd use your broker's data over any free source. Remember, the free data sources have no incentive to give you good data. You might think you are getting a great deal until your system fires off a bunch of terribly timed signals because your training set was bad. You'll pay for the data one way or the other.
What is Excel Solver? I Googled, it is an Add-in and my Excel does not have that feature. For my VBA macro, VBA does not recognize NORMDIST, I have to calculate NORMDIST in Excel and then access it in my VBA subroutines. To write those functions myself, it is beyond my capability.
Hmmmm. Have you tried calling it as Application.NormDist? as in: d1 = (Log(underly / strike) + rfrate * t) / dtmp + 0.5 * dtmp d2 = d1 - dtmp dtmp = strike * Exp(-rfrate * t) vcall = underly * Application.NormDist(d1, 0, 1, True) - dtmp * Application.NormDist(d2, 0, 1, True)
If you are accessing a "worksheet function" you usually have to put the word "Application." or "Application.WorksheetFunction." in front of it. Like this: Application.NormDist(a,b,c,d) Yes, very possible. As long as your strat is not latency-sensitive or requires truly massive or complex calculations, you should not have any problems. Yes, and Yes. Full version license for Matlab runs into the $thousands though (and free clone Octave just doesn't cut it), so you might consider switching over to R.
See here: https://www.elitetrader.com/et/thre...her-traders-coders.313213/page-3#post-4518050 and, See emphasized text within below quote. ("...A while back I introduced you to the Excel WORKSHEETFUNCTION. What happened with that?...")
If you need a good IDE use the community version of Pycharm works with Java, python and others just makes it easy for debugging . Python add-ons for writing out Excel , has HTML parsers for scraping pages, Url libraries for downloading data. Matlab data structures, plotlib for plotting charts and on and on
What I like about MATLAB that you can use Java and Perl directly in with your MATLAB code. I write a witch's brew using all three as each has its own strengths.
The future is with Python I guess. Easy to learn and implement your ideas. Basically it's good because it has in built compiler so you have to write less code. The second is how variables are defined - they can be defined anywhere in the code. But well there is drawback with Python and it's speed. If you build sophisticated strategies where ms matters you have to dig into python compiler to make some tweaks and increase speed of execution
I don't quite understand the Python bashing. There's more ways than one to use Python. I do everything from research to automation in Python and thus simplify things quite a bit. The difference is that the code is converted and compiled to C using Cython, not for speed but for obfuscation. I've yet to encounter any issues.