Machine Learning in Capital Markets : Part III

This part is the continuation of part II. With the help of fundamental analysis, the value of a company is assessed taking various factors into account. The fundamental analysis approach follows the assumption that the company in question has an economic value to which the associated share price adjusts in the long term. The value of the company in question is defined as the intrinsic share value. The return on shares is assessed using the internal share value, which enables conclusions to be drawn about the profitability of the sector. A common method is the present value theory. The Present Value Theory states that the intrinsic value of the share is due to the sum of dividends and other monetary benefits per share and is discounted to the current day. The present value or present value concept enables the calculation of the intrinsic value. The simplified concept of the Price Earning Ratio (PER) method allows conclusions to be drawn about the future prospects of course profits from a company that has been studied. This means that the most immediate insight into the price gains of company shares is gained. It is mandatory to take into account that the chances of price gains are dependent on the intrinsic value of the share.

Technical analysis

It is worth noting that conclusions on the global market, based on the sketch of a single company, are of insufficient quality. It should also be avoided to draw conclusions for a specific company based on the global market. In order to draw conclusions of acceptable quality, it is mandatory to assess company-specific, industry-specific and global aspects in a centralized manner. Methods for studying company-specific, industry-specific and global aspects are top-down analysis and bottom-up analysis. The top-down analysis follows the approach of specifying global data for the industry, the conclusions established thereby are in turn specified for the companies. In the bottom-up analysis, vice versa attempts are made to generalize the company-specific conclusions to the industry, which in turn are generalized to the global market. Technical analysis – sometimes known as chart analysis – tries to predict price trends by observing market movements. It is assumed that all necessary market information is already included in the price trends and that an analysis of the company’s business data is not necessary.

Everything that can influence the prices is reflected in the market price. It is believed that prices rise as demand increases and supply falls as supply exceeds demand. This assumption makes it possible for machine learning to create a behavioral pattern from the course and the market price in order to be able to derive a prediction for the further course of the course.

An analysis of stock trends requires that these trends be divided into trends. Trends illustrate stock trends and thus enable probability forecasts to be made about the further course. As a premise for the division into trends, the number of trend types to be specified must be defined. Furthermore, it is part of the premise which distinctions exist between the specified trends and how these trends differ. In order to define trends, the course directions of the short-term course peaks (zenith) and the short-term course depths (nadir) are studied. This defines the trends as follows. A sideways movement of peaks and valleys defines a sideways trend, a successive increase in peaks and valleys defines an upward trend vice versa, a gradual decline in peaks and valleys defines a downward trend.

Each trend is categorized as long-term, short-term or medium-term. It has to be taken into account that every trend type exists infinitely often within the course of a share. In order to minimize the number of trends to be considered, it is mandatory to define which period is valid as long-term, short-term or medium-term. A period definition for long-term trend is a trend duration of more than six months. Medium-term trends are defined as constant trends from three weeks to several months. All trends that last less than three weeks are defined as short-term trends.

In summary, it was defined that trends consist of peaks and valleys. When a valley is low enough, the selling behavior changes into buying behavior. This makes the prevailing downward trend an upward trend. This exact point in time is defined as support. Vice versa, the exact point in time of a change from upward trend to downward trend is defined as resistance.

A resistance line is unable to interrupt a rising trend, as a result the resistance line is defined as a short-term interruption. The same principle applies to a support line. A continued upward trend can be determined by the fact that the successive supports describe a higher price point than the respective previous support. A continuing downward trend can be determined as soon as the successive resistances describe lower price points than the respective previous resistance. It is essential to define how course changes are determined. A price drop can be determined as soon as a price drop falls below the previous support line in an upward trend. According to this law, it can also be determined that there is a price increase as soon as the previous resistance line is exceeded in a downward trend. The determination of a price rise or fall enables a trend reversal or the transition to a sideways trend. If, after a trend change from a downward trend, the highest previous resistance line is clearly exceeded, this will be rated as a new support line when the trend changes. This is called changing roles of support and resistance.

The key to understanding the future lies in studying the past, or the future is just a repetition of the past. This assumes that stock trends have already played out in the past. Past price information or market influences that have changed after having to repeat can be used for a prediction.

To be continued in next part.