Last week's silence

After last week's dismal performance (-64 pips) I decided to pull back on trading and focus a little more on understanding how I make a trade and what makes it work (or not). I capped off a bunch of work that I had been doing on a testing harness for my system backtesting and I now have a great little collection of modules (in Python) that allow me to do proper backtesting of any system I can code up. Creating a new system is easy: I just create a new module and in it I implement a TradingSystem class with a tick method. The tick method is called by the test harness every time a tick occurs in the "virtual exchange" that is modelled. This allows me to register a number of systems and test them simultaneously.

I also finished off a very difficult battle with the performance of these modules by doing a little profiling of the code and ironing out a few choice kinks.

OK, so after all that bla-bla what have I learned?

1. I am developing the feeling that pivot points (and the resistance and support levels they define) are much better used as support and resistance than as areas in which to enter/exit trades
2. I found that in backtesting several variations on my TS01 system, the runs with looser stops (10-20 pips) and large T/Ps (up to 150 pips) actually perform best over time
3. A key factor in the success of my runs has been the margin I use to confirm a trend: too big a margin and I don't enter early enough to offset losses, too small a margin and I get caught by whipsawing prices