Systems development with the use of computers is very important. It helps us prove to ourselves that certain methods work and others do not. It is better to find out if your strategies have worked consistently in the past rather than just randomly hope that they will work in the present. Below are some tips to keep your system development strong and effective.

1. Use logic first. The first thing to do is to observe the markets and the movements of individual stocks. Do you notice recurring patterns? When you think of stock market logic, is there something you think will work yet you have not tested it? Test it. Also, periodicals and industry websites such as Yahoo finance, thestreet.com and many other trading sites can offer unique ideas to test. If someone brings up an interesting point about trading, test the validity of their claims.

2. Test Objectively. It's okay to hope that you will find a good strategy, but it's not okay to hope too much. If you do, you will jump to conclusions too quickly. Also, forget about your favorite indicators. Don't have favorite indicators. Just have indicators that test well and ones that don't test well. That way if your "favorite" no longer tests well, you won't be married to that indicator.

3. Form test lists. First of all, you want to test long periods in the stock market (from 2-10 years). It would be good to have 2 bull markets, 2 bears, and a sideways market. In the 20th century, there were 33 down years and 66 up years for the SPX. With that in mind, you could test multi-year periods when there were two years for every down year. For example 2002, 2003, and 2004 where there were two years up and one down. Although we don't expect there to be 66 up years in this century, we base our testing on some sort of logic. For it would not make since to assume that there would be 66 down years this century, unless you have some strong logic to back it up. Also, you'll need to put together random lists of stocks to test your strategies on. I recommend lists of 30 stocks. That is enough to have statistically significant results. Test those lists of 30 stocks over those same time periods you plan on using. Then look at net profit, percentage profitable and other factors you deem important and compare the results using different systems and indicators. You can answer the age old question of "which indicators are most effective?" You can also throw away the indicators that don't work.

4. Decide on pertinent data. We've already mentioned net profit and percentage profitable as relevant measurements of the strength of various systems. There are other important factors such as win to loss ratio, standard deviation and max drawdown. I use these and so should you. Also, be on the lookout for other factors, such as a smooth equity curves in tests and stay away from strategies that never trade or trade too much. Also remember this: Don't use biased data. You see, many traders in the late 90's and early in the 21st century used the data from just the 90's to back test. Do you think those same strategies worked will in the last few years? No, they haven't. The market has been different since 2000 and we must test all sorts of markets so that your system does not depend on particular market conditions to survive.

5. Be sure that the data is good. When you form lists of stocks and start testing, be sure to properly label each list. I use Excel and then I am sure to label each stock list properly. Also, I encounter bad data all of time. I see unusually low numbers mostly or zeroes where they don't belong and can see quickly that the data is bad.

6. Test simple indicators first. Before you start testing the strategies that require the stars to line up for entries, test the simplest indicators first. For example, avoid testing a stochastics, RSI, ADX, volume combo at first. You should just test ADX first and then decide if it is effective and then later you can test ADX in combination with other indicators to form a strategy. Don't try to be unique. Just try to make money. That's the purpose of trading.

Price Headley is the founder and chief analyst of BigTrends.com.