Thank you all! After reading the forum again and again, I especially agree with 2rosy's comments. I have a bit research of R and Python, and I have found R - quantstrat useful and quite popular in the community. Does Python has something similar to quantstrat?
Python: These guys have an open source backtester: https://www.quantopian.com/ https://github.com/quantopian/zipline If you are not a trader, my experience is that Amibroker will get you trading much faster than other languages. You can backtest many ideas very fast against many types of stocks and assets. It's also mature and so you can trust and debug the backtesting results. It will also naturally lead you into a diversified style of trading. Keep in mind AFL is array based. Easier for non-programmers. Walk forward optimizing is very fast and easy to use. If you need the simplicity of Amibroker coupled with C# you can try QuantShare You can code in array like AFL or in C# with access to external c# libraries. I am also learning Python and like it. But I couldn't trade on it, just yet. R i have not tried (I have access to Matlab...)
Use assembler. It's as fast as it gets, and simple -- only a small number of instructions to memorize. No need to bother with all this fancy architecture .
Maybe they do now, in this decadent era of Lite beer, hand calculators and "user-friendly" software but back in the Good Old Days, when the term "software" sounded funny and Real Computers were made out of drums and vacuum tubes, Real Programmers wrote in machine code. Not Fortran. Not RATFOR. Not, even, assembly language. Machine Code. Raw, unadorned, inscrutable hexadecimal numbers. Directly. sorry, just could not resist, even I spent the whole day writing R code today
This post is 4 years old. TradeStation 9.5's optimizer and walk-forward analyzer are no longer an issue. TradeStation 9.5's optimizer now runs on more than 1 CPU core. TradeStation 9.5 is now 3.5 times faster than TradeStation 9.1 on a 4 core CPU. I assume 9.5 runs even faster, possibly 9 times faster, on a 10 core CPU. It may no longer be necessary to write in Python or R code. Writing in Python or R takes longer than in TradeStation's EasyLanguage. Not only must you develop new programming skills but you must also write a walk-forward analyzer from scratch. It is more productive to use TradeStation EasyLanguage.
Still, Easylanguage is slow as hell. I tested a small strategy that needed 4 seconds for a backtest in C. The same strategy needs 150 seconds on Multicharts in EasyLanguage. In favor of EasyLanguage, it's indeed much faster than R or Python.
Before you can test a new idea, you have to find an edge; that means you have to analyze data to formulate a hypothesis. This needs to be done in some statistic software. For me, the most productive tool is Python with Pandas in Jupyter Notebook, it is really amazing, how fast you can test ideas as a non programmer. At the end of the day you will end copying snippets from different existing scripts, so don't think to much about the "best" tool, just start with one and see what happens. The difficult part is not to master the tool, but to master the process of developing a system, and that can be done with all of them. I recommend to have two independent systems, where you code the same strategy twice to see, if it generates similar results; don't forget you are willing to risk real money on your strategy. Sure Python is much slower than other options, but as you said productivity of your time is the measure, not execution time, if it is fast enough, its ok. At the end of the day trading is not rocket science, so you don't need to faster algorithm on earth as a retail trader.