Skip to main content

Stock trading system

Stock trading system - how to test?

If you are interested in investing in the stock market and if your goal is to make money, sooner or later you will be interested in the subject of system testing. Where "investment system" means a set of rules when you buy, when you sell, and how much you buy.

Most indicators and oscillators can not earn on the stock market. Those that will give you need to use it differently than it is commonly used. The same general book knowledge does not work. Do not kid yourself. This is one of the reasons why only every 10 investor earns. Which group do you want to belong to?

It is imperative that you review your investment methods before investing. You need to know what profits they can generate, and how often these two are occurring. Relying on information you just heard will always lead your investment account to bankruptcy.

Testing, testing, analysis. Name it as you wish. You are always doing this on historical data, but you have several options for doing so.

Stock trading: manual work.

Watch charts in the analysis program. Draw lines, mark places and levels. Mark all the necessary elements of your system. Then you have to count the situations when you make money, the situations when you lose. Count earned and lost amounts. Calculate whether your method is profitable or just a myth.

This method is always applicable. This is its primary advantage. It is possible to test all assumptions, whether or not they are programmed or not. However, it has serious disadvantages. First, it is very laborious and secondly, it is a serious mistake. Evaluation of man

Human eye and brain are unreliable and unfortunately tend to be distorted. First of all, it is easy to miss something when you die in large amounts of data. Secondly - the human brain has in nature the search for patterns. That's why it fits so easily with what he sees with his convictions. Very serious flaw

Eliminate these drawbacks by simply commissioning a computer. There are more opportunities here.

Stock trading: let the computer be your eyes.

Since people easily skip or twist what they see, have your computer find important components of your system. Let the computer find these places and levels. He will do it perfectly perfectly and do not have to wonder "do these candles overlap or is there a tiny gap?"

He is counted and decisive within a fraction of a second. You will also eliminate subjectivity in this way, but you will still have a labor-intensive problem. If your computer only points and levels, and you still have to count on profit and loss alone - you still have a lot of work to do.

However, staying true at this stage does not make sense. Since you already have an almost programmed system, you only add to it the transaction calculation and you can go on to the next four methods.

Stock trading: backtest.

The simplest version of the automated test. It does all the work of the man from the previous points. The computer takes historical data and simulates that with your system made a transaction. This allows you to see the effects of using your method. The amount of statistical data will always outstrip what you prepare yourself. And this is just the beginning of the possibilities.

Stock trading: optimization.

The second step is to automatically find the best parameters. If you are investing in a simple intersection system, for example, you need to decide what lengths of medium you use. By making a plain backtest you have to decide what average to use and see the result of the investment.

You can change the mean and compare to see if the result will improve. However, you are not able to see thousands of possibilities in this way. And here comes the help comes with optimization. This is simply a backtest carried out repeatedly in order to find the parameters you seek for. This type of test brings much more knowledge, though, and he has the disadvantages.

Stock trading: walk-forward.

It largely eliminates defects in optimization. Real life with the system is to find the best parameters and invest with them. However, from time to time these values should be searched from time to time. The nature of the markets is changing, investors' emotions change, so the system must also match.

Walk-forward simulates just such real life on historical data. Takes a piece of data, optimizes on them, and invests in the next section. Once again, it stops, optimizes and returns to investing again. It goes through all available data as if you personally made the decision month after month or week after week.

However, despite the fact that walk-forward is called the test of all tests, there is another step above it. In addition, this stage must be performed personally.

Stock trading: optimizing walk-forwardu.

Although it sounds paradoxical, walk-forward also needs to be optimized. I do not know the program that would do this for you. The walk-forward alone has parameters and must decide on its operating principle. Simply put, you have to decide how long it takes to optimize and how long to invest.

Do you optimize on monthly or yearly data? Or maybe always on data from the beginning of quotations? How often do this? Weekly or monthly? Or maybe once a year? Walk-forward must have it given. He must know how to go through the years of historical data.

This is where the knowledge of the market matters, the type of instrument you are investing in. Its variability, leverage, whether it is trendy or even what turns. Of course, you can also test this by performing several walk-forward with different settings.

In all these tests, much remains to be learned. Although system stability, bankruptcy probability, system drawdown, etc., however, without the programmed system it will be impossible to perform any of the steps described above except for the first.


Popular posts from this blog