Monday, July 6, 2009

How to find trading system

How to a find trading system is a question that I’m asked very often.

There is an ever going quest to find a good trading system and once you have one that is making money, you will probably try to get an even better one - by trying to improve existing one and by squeezing even more points out of it.
Web is full of trading systems, sold for $99.95 or so, with promises of making you million or three over next month or so. I would like to ask - If it is so profitable, why the owner would sell it for that amount, when he/she could trade it and retire after few months.

Finding a trading system is a time consuming process. One should have reliable data for market that one plans to trade and software with back testing capabilities. I use Metastock for back testing (same software is used for charts) while for ALSI data feed I am subscribed to local data provider - Hisat.

Metastock has very nice back testing capabilities and it comes with approximately 40 trading systems, ranging from simple to very complex. All systems are explained and focus is on what they are using as signal and what is being done with optimization parametars, with possibility to test across array of numerous time-frames, stop-losses, etc. So, it is possible to input any parameter that you can think off and see what the outcome would be.

There is somewhere on the web a trading idea to optimize some trading system by going only long after winning trade and short after a loss. So, absolutely anything is testable and tradable.

When it comes to data that is to be used for back test, one should choose market and time frame first. Also, one should have idea what one wants to trade - ALSI futures or white maize maybe; maybe gold; pork bellies or government bonds. There are hundreds of markets out there, so one should be wise when choosing one. There is no need and urgency to go for main-stream markets, but for one that is liquid and relatively easy to trade. Time frame could be changed, as back test will allow you to look into all time frames you like - from 1min to weekly chart.

I would recommend starting with some trend following system, i.e. moving average penetration or crossover. After one finds system that has positive expectancy, next step is to paper trade it in live market conditions. One should record all trades - every winner and every loser and I would recommend for that trial phase to last for at least six months. System must prove itself over as many trades as possible, with some bare minimum of 400 trades. Only after all these steps are completed, I would recommend that one should go in with real money.

This is just for starters. From there on, one should develop as a trader. System is only 10-20% of success, as there are many other factors one should work on after system with positive expectancy is in place.

Trade with trend.

NOTE: Positive Expectancy = (Average Winning trade x Winning %) / (Average Losing trade x Losing %) > 1