Why is this necessary?
Every day new information appears on pages of business papers, news agencies and Internet sites devoted to financial markets analysis.
Every day this information is read by traders. They read it and make decisions. Some make decision instantly, some - within the day, some will decide to wait until tomorrow. Some already made their decision yesterday.
How will decisions made by traders reflect on your stock prices, on technical indicator and stock index values?
Is it possible to evaluate force of trend in time series matching stock prices, technical indicator and stock index values?
SerieSelect has been created to answer this question.
SerieSelect breaks the entire multitude of time series into three groups:
- The first group embraces series with self-sustaining behavior: probability that the trend will remain intact is very high.
- The second group consists of non-trending series, and there are all grounds to believe that such tendency will persist.
- The last group is made up of series for which the trend, most likely, will be opposite to the previous one.
|
 |
In the basis of SerieSelect there lies calculation of Hearst parameter. It is by its values that one can judge direction of series trend. Three figures, which illustrate changes in this parameter depending on series analyzed, are presented below.
Microsoft daily close prices for the period from April 14 to March 30, 2001 have been chosen as a sample.
In fig. 1 on the left, you see a price series, for which the analysis is carried out. The main operating schedule used to calculate Hearst parameter is located on the right.
Figure 1. Daily price (on the left) and Hearst parameter calculation (on the right).
| Price |
Hearst parameter H=1 |
|
|
In this case, H value parameter is equal to 1, which means high probability of the current trend preservation. As a comparison, figure 2 presents parameter H calculation for daily yield series over the same period.
Figure 2. Daily yields (on the left) and Hearst parameter calculation (on the right).
| Yield |
Hearst parameter H=0.58 |
|
|
For this series H value parameter is about 0.5, which means absence of trend component and preservation of this tendency in the future.
As for time series with H values much lower than 0.5, such series occur in financial sphere extremely seldom. However, time series composed of yield increments displays exactly such a type of behavior, though on the surface it differs from yield series very little. You can see it on fig. 3. H value is equal to 0.32, which speaks about high probability of change in the current trend.
Figure 3. Yield change (on the left) and Hearst parameter calculation (on the right).
| Yield change |
Hearst parameter H=0.32 |
|
|
|
Main
