Speculative attacks on markets have been thwarted repeatedly by the various interventions of governm...
5 Steps to Developing a Trading System
07/01/2013 6:00 am EST
It takes time and experience to build a successful trading system that can generate consistently profitable trades, but Alexander Nekritin of TradingMarkets.com lists the first steps below to get you started.
Each trading system has to have an edge. The edge is what should make your system have positive expectancy. In other words it must be profitable over the long run. We will go into the calculations and common misconceptions about expectancy later. But for now it is just important to understand that an edge should ideally make the system profitable in the long run and make you more likely to make money than lose money over a large enough sample size of trades.
2. Indicator Co-Linearity
When selecting an edge, it is very crucial not to over-optimize or fit the data. One common mistake that people make when developing a system is using two similar or confirming indicators and optimizing them. This causes the system to look great historically; however, the system will not do as well in the future.
Indicators are Broken Down into Five Types:
Examples of Each Indicator Type:
Trend = Moving Averages, ADX
Volume = On-Balance Volume, Accumulation/Distribution
Overbought/Oversold = CCI, RSI
Momentum = Stochastics, Rate of Change
Volatility = Bollinger Bands, Keltner Bands
3. System Robustness
It is also important that the edge is robust. A system is robust when it maintains positive expectancy. The system should be tested on an up, down, and sideways move. Many trend-following systems perform well when the instrument trends but don’t do as well when the instrument is in a sideways whipsaw period. It is crucial that the period is taken into account during back testing.
4. Amount to Back-Test
I recommend back-testing on at least 2000 bars. If you are back-testing a system on the daily charts; I recommend using 10 years. On the intra-day charts, I recommend back testing the systems as far back as your data vendor will allow. This is usually six months to a year.
5. Back Testing Programs
It is important to use professional level software with back-testing capabilities when developing your system. To name a few: MetaStock, BigTrends.com MetaStock Toolkit, TradeStation.
By Alexander Nekritin of TradingMarkets.com