In arriving at a personal trading system that suits you and that you feel comfortable trading, there are various choices that you must make, as follows:
· Fundamental vs. Technical
· Long-term vs. Short-Term
· Discretionary vs. Mechanical
The choices shouldn't be considered strictly black or white; shades of gray are permissible - so for instance you may be a technical trader who uses fundamental analysis in arriving at the universe of stocks to be considered for purchase. However, one thing you do have to be sure of in order to have the confidence to trade your system is to know that it has a trading 'edge'.
Your edge
A Trading Edge is an advantage that allows your system to be traded profitably. As an example of a trading edge, consider the roulette wheel at a casino. The wheel has 36 numbers on which the house pays out, but it also has two slots numbered 0 on which all bets are lost. Therefore the house has a mathematical edge of 2/38, i.e. 5.26%. Of course in the short-term the number zero may come up less than expected, but according to the Law of Large Numbers over the long run it will perform as expected. If we spin the wheel 38,000 times, we can expect the number zero to come up very close to 2,000 times - the house has an inbuilt 'edge'. This is the reason that casinos have so many tables - over a large number of plays, despite the occasional punter who goes home happy there will be many more that won't, as the casino's edge will come into play.
Back-testing
How do you ensure that your system has a Trading Edge? You will not know for sure until you actually trade it, perhaps not necessarily with real money but rather paper trading it for a month or two, i.e. recording all the purchases and sales according to your system's rules but not actually placing the trades. However, you can get an idea of the best candidates for further research by back-testing. In a back-test, you apply your prospective system using historical data, and track the results. Many technical analysis software systems, such as Metastock or Telechart, allow you to back-test systems provided that you can clearly specify the rules, i.e. buy on next open when the 26/12 MACD (Moving Average Convergence Divergence) line crosses the signal line. However, if your proposed system allows some discretion in whether you would take the trade or not, it is not possible to test it in this manner. The only way then would be to go to a stock chart, set your starting date as the last point visible on the right hand side of the chart, and then click forward one day at a time, noting whether you would buy, sell, or do nothing, according to your discretionary 'rules'.
Curve Fitting
One of the potential pitfalls to be aware of when backtesting is the practice of curve fitting. If you start off with one or two rules but end up with nine or ten exceptions, additional rules which if followed improve the results, then you are curve fitting. You will have arrived at a set of rules which may work very well with this set of data, but may not work when you try to apply it in practice. A good practice is to retain a set of 'out of sample' data on which you can test your optimized trading rules. So, for instance, you might optimize a system on the previous five years' data, and then test it on the five years before that. The basic rule of thumb when designing a trading system is 'the fewer rules the better' - always try to keep it simple.
Mick Brooks is an educator, a public speaker, and an avid stock market investor. As a UK-qualified CPA, he thought making money in the stock market would be easy, but his 'education' cost him around $30,000, so now he makes it his goal to help others to avoid replicating the more obvious mistakes. Visit his website, http://www.beginning-investing.net/ for more advice and information, and don't forget to pick up your FREE copy of the 'Investing Secrets - Day Trading' report!
No comments:
Post a Comment