~ FT-Test Day 2 ~
One exit for a loss (XLB -%0.91), several new entries. It is already looking very hard to share the amount of data I am collecting at one time, so I will give you an idea. I have several different views of this test in Excel. I am using TDA’s data feed to supply the current information and locking it in manually as needed. I will explain the categories at the end of the test and in the final analysis.
The overview (not meant for reading here):
Here is the jist when it comes to what I am doing:
What is the version of Fisher Transform used all about?
The indicator I built is the traditional FT where you have the length setting at 10 and the first line (FT_Line) is created. Normally the signal line is FT or if FT is higher than the bar before then long else short. Instead, I take a super fast hull moving average of the FT and use it as a singal line, and if FT is higher than the signal line long else short.
What prices are taken in these trades?
All prices are END OF DAY readings (unlike real-life where you need to be filled). REMEMBER! This is a simulation test, but moving forward live instead of backtesting. I want to categorize the signals and see which signals traded better than others. Therefore, I plan to take EVERY single signal from the 142 names I watch. This will load me up with data. All trades are binary, so if a short signal comes I go short, and close the long if I was long of course. Commissions are not being used as the profit run makes no difference. I am much more worried about measuring the signal categories against each other.
To what end does the test aim?
I want 80-120 trades minimum in each category, making for a lot of data to collect. It will end when it ends, and I will crunch the numbers. 🙂
How is the data being collected?
I am using a leger that automates the statistics, however, I manually enter and exit the trades in the log. In the past I shared this each and every day on the QA page. Now, I think I will just have everyone trust that I am thorough with my data, and that when the test concludes, ALL of the evidence will be available and support my findings. Remember, I am looking for an amazing setup, and I DO want the test to show me how this indicator fails, and I plan on sharing all of it. Keep in mind that every time I do a new test, I build a new statistics modeling machine specific to the test that I am doing. Never has a template NOT caused problems, so it is just easier this way believe it or not! At this point there is more model to build.