This new thread will be a discussion and tutorial on system development. The example will be done mostly in TradersStudio with a few of them in TradeStation. You can get a free demo of TradersStudio by just registering on my web site. This thread on ET explains the demo. http://www.elitetrader.com/vb/showthread.php?s=&threadid=232738&perpage=6&pagenumber=1 I have a plan on how I am going to present the material , but let me know if you have any ideas. I will post in this thread at least several times a week.
Let's begin our journey into the world of systems. Let's first discuss peoples views of mechanical trading systems. Trading systems bring on strong feelings both ways with different groups of people. Some people will only trade using mechanical systems. Others do not believe in systems at all. They believe they do not work long term. There are also people who believe discretionary traders can beat systems. The problem with that thought is that discretionary traders are like a baseball pitchers who can throw 95 , born with god given gifts. Great system trades just need to follow the system. Next week we will start with our system development course.
Let's begin this conversation with my view on technical analysis and system development. First most classic rules for technical indicators do not work. You can create systems which will backtest well but will not work well out of sample. Momentum indicators might work in trend following, but most classic indicator rules do not. Patterns , setups and fundamental relationships are what is required to build reliable systems. Indicators need to be understood at a much deeper level than the classic rules handle to be of any use.
Let me add one as a professional developer. TEST YOUR EDGE CASES. I was in a discussion some weeks about with someone who had a system that in the core was a stochastics based reversal system. He had some bad trades in a row and wondered that "Stochastics do not work". Well, lets get this straight - if the market moves long, storahcstics "draws out" (close to upper limit) and the SMALLEST reverse movement triggers it to drop. This is a known mathematical issue with that approach - and one that people should be aware of. Looking at the chart in this example, would could hardly see the price going down - more like "going sideways / slightly up". I am not advocating Stochastics here - I use it as example. You MUST check the behavior of whatever you use in various extreme situations. And your code MUST handle that. Some of the indicator formulas have quite extreme edge cases, mathematically.
Sorry had some personal issues and fell behind will start up again next week. I had some family medical issues.