Feedback on System Analysis: TrendFaker

[hrokling 10]

Although I've been trading for more than 10 years, I've just recently started some serious automation efforts. My first attempt is a system that's only been tested on five months of 1m data, but is so frequent that it's still pretty interesting.

 

My objectives is high-frequency, smooth equity curve. Not too big drawdown (less than 10%).

 

Attached is the Equity Curve as well as most of the analysis from MultiCharts (some has been removed, as I don't want discussion on market selection). Any feedback is welcome.

 

A few additional comments: As you can see, the strategy had a major drawdown in October. I have a catastrophical stoploss built in, and it got triggered multiple times. Stopping the system for the remainder of the day would improve the results, but I don't know how to code that yet

 

Also, the position sizing has not been properly done. Right now it just reinvests profits - and trades with a leverage of 1.2 to 1. Is there a "simple" way of working with position sizing in MC?

Trendfaker.xls

[Dacamic 10]

Did your tests include commissions and slippage?

[hrokling 10]

Yes, real commissions (IB) and realistic slippage. It's listed in the Excel-sheet, Strategy Analysis tab, row 25 and 26.

 

The historical data I've been using is from eSignal, and I've now in the meantime compared this with 1m-data from IB which makes me question the realism of the backtest. The data are not quite the same - the price movements at IB are tighter and more jumpy, whereas it seems more rangebound with eSignal. This affects the system performance, although I'm still investigating this. See attached chart for the same time period, IB on the left side.