I'm going to resurect some work I did on mechanical system trading with Sierra Chart and I thought I'd share it in the hopes that more minds might make a better mousetrap. This thread isn't going to be about specific systems although we might need some basics to generate discussion; its going to be about the underlying test harnesses and analysis software.

This is independent of SC's own backtesting facilities. My initial goal is to create a system that:
- runs a set of bar based backtests with two configurable variables
- marks the charts with position opens & closes plus limit and stop orders for visible inspection
- produces a single contract OTE/CTE line (variable contracts would be an extension)
- generates a timestamped text file with order, position, equity and other changes for analysis
- generates a variable based summary text file of each trading run with key outputs like pf, mfe, mae
- the order generating code should be in a form that could be shimmed into live trading modules.

Thats all in C++ as dll modules. I've got a lot of this stuff now but need to revisit it as its a couple of years old.

Then you take the text files and analyse and visualize the data using Python (3.4).

Is there any interest in participating? Does anyone think this is the wrong approach and have a better suggestion for all or part.

kiwi 

I'm going to resurect some work I did on mechanical system trading with Sierra Chart and I thought I'd share it in the hopes that more minds might make a better mousetrap. This thread isn't going to be about specific systems although we might need some basics to generate discussion; its going to be about the underlying test harnesses and analysis software. This is independent of SC's own backtesting facilities. My initial goal is to create a system that: - runs a set of bar based backtests with two configurable variables - marks the charts with position opens & closes plus limit and stop orders for visible inspection - produces a single contract OTE/CTE line (variable contracts would be an extension) - generates a timestamped text file with order, position, equity and other changes for analysis - generates a variable based summary text file of each trading run with key outputs like pf, mfe, mae - the order generating code should be in a form that could be shimmed into live trading modules. Thats all in C++ as dll modules. I've got a lot of this stuff now but need to revisit it as its a couple of years old. Then you take the text files and analyse and visualize the data using Python (3.4). Is there any interest in participating? Does anyone think this is the wrong approach and have a better suggestion for all or part.

Im an avid fan of Sierra, and i've just started getting into Python along with matplotlib and pandas. The main reason for getting into Python has been to analyze my own performance because i've felt too limited by excel. So ill certainly be following along but im probably not experienced enough on the development side to contribute much to your thread. Will never the less be great to see how you incorporate Python into your analysis.

Will you be using matplotlib and Pandas?

Not sure DarkPool. In fact I just got distracted for a month (oops).

I like matplotlib and pandas but I'll probably try to do most of the work in C with Sierra at the base level. I guess I'll think a bit more about that as I get deeper into it. I've just setup Zim wiki to keep track of what I'm doing so it seems likely that this sputtering restart might be just about underway.