BSE & NSE Indian Shares - Daily Stock Market Report
Indian Shares - Stock Tips - Top Gaining Shares - Daily Update Free

Author: Doug Newberry

Article: If you don't have a trading plan, you're setting yourself up for failure in the market. It's likely that you've heard this advice everywhere from the corner stock trading office to the Wall Street high rises. Your system provides the blueprints for how your trades will progress. If you're just starting to trade stocks, you'll probably want to test your system to make sure it makes money. One way to test is to use system development tools that make it possible to quickly test, in a short time period, thousands of combinations of system parameters. Of course, once your new trading system has been thoroughly tested, you can begin to make money with it. As you're developing a trading system, you must be aware of the phenomenon of over-optimization. There are many parameters available in looking at your system to see how well it performs. You should evaluate the smoothness and drawback of the equity curve increase and the profit and the loss. The danger is to focus on one extremely high parameter. A better strategy is, instead of just looking for one outstanding set of parameters, look for a cluster of terrific results. If you choose parameters in the middle of a good cluster of results, this is likely to provide you with much more consistent ongoing results. Make sure there are enough trades in the resulting test set so you'll know they're significant statistically. It's very unwise to decide on any system if you don't have at least 30 trades in the test period. You'll finally be able to have some confidence in the system after more than 300 trades in the test period. Of course, it's important that you didn't over-optimize those results. You still won't be sure you have a reliable system after 300 trades. Many over-optimized systems perform well for 300 trades and lose money after that. Holding back some historical data will allow you to check your parameters as a test set. However, if you use the test set more than once at the end of your system development, beware. Your optimization may be compromised because your "test set" has become part of the optimization procedure. Some people tend to fine tune their system too much to fit past performance. Here is a hypothetical example of one such system: In your "winning" system, you only buy stocks that: *open between $6.11 and $11.23 *rise between .37% and .78% between 9:23 and 9:51 a.m. *have a trade volume between 97,836 and 143,687 shares per day with an exit strategy as follows: *close the position by 11:54 am *wait until the stock price rises .63% And that's not all. Positions are closed while holding your breath on the day before a full moon or the nearest Thursday, whichever comes first. While his exaggeration is obviously absurd, it does make a point. Any particular system may make you money in the short term, even if this money-making is purely a coincidence and has nothing to do with how well you have devised your system. This is why you must be careful to make sure you have a statistically significant number of trades in your test before drawing conclusions about how well your system works. Keep your eyes open for a medium range of success in between two and five parameters. If you're able to find this sort of system, you're very likely to use it successfully in the marketplace. Since you're dealing in probabilities, you can't ever be sure you've got a truly reliable system. You could just be hitting a possible though unlikely series of bad or good results right at the beginning. But if you're careful not to over-optimize, you'll probably have a consistently profitable system.

About the author: A Director of Investing Systems Network, Mr. Newberry consults on the conceptual development of portfolio management tools and software to trade stocks. Investing Systems Network makes these available to its customers, who are traders and investors in more than 70 countries.