This is the continuation of part VI of Machine Learning in Capital Markets.
In the course of this work, a critical examination of the use of various learning methods in the analysis of financial market-specific data showed that machine learning methods are only conditionally suitable for successfully predicting share prices. A look at the results shows that there are sections of the stock market that can be analyzed and forecast using machine learning. Nevertheless, it was and is not possible to use the existing combinations of different classifiers and characteristics to predict a complete forecast of a share price.
Besides, it was found that a large number of factors must be taken into account when analyzing capital markets, particularly the stock market, to be able to successfully predict stock prices. Because of the exponentially increasing volume of data, it is an elementary factor to check the quality of the data used, since only prepared and cleaned data sets for a classification procedure provide positive results and thus have a strong influence on the success factor concerning the forecast. Concerning the conciseness of the aforementioned data premise, known and foreseeable exogenous factors must be taken into account and integrated into the analysis of the database. This work also demonstrated that direct and indirect characteristics play an elementary role in the informative value of the object under consideration. The aim was to achieve a high level of generalization of the characteristics so that they could be used afterwards for further calculations and comparisons. In a subsequent classification, the data only had to be divided so that it could be used due to its predictability, controllability and adaptability.
Two different approaches were examined in the analysis techniques. A distinction was made between fundamental analysis and technical analysis. In contrast to technical analysis, fundamental analysis proved to be beneficial if the “intrinsic value” of a share should be used. By comparing the intrinsic and actual value of the stock, there is another way to evaluate the buy or sell decision. In principle, fundamental analysis is used to evaluate a company. In addition to fundamental analysis, the background and functions of technical analysis were also examined. The technical analysis tries, among other things, to classify the courses in different trends to be able to filter downwards or sideways trends. For the implementation of this analysis technique, various technical indicators were examined for their advantages, disadvantages and possibilities.
In addition to the analysis technology and the cleaning of the data, it is also an important factor to prioritize the sources and evaluation methodology. As an example, the use of neural networks, their weighting properties and methods were discussed. For example, it was discovered that even a minimal change in the weighting of a decision-maker has a significant impact on the decision made. For the implementation of this analysis technique, various technical indicators were examined for their advantages, disadvantages and possibilities.
To enable the various methods of AI to be able to evaluate data mechanically, in addition to preparation, it must also be transferred into a machine-readable format. A candlestick pattern can be used to mathematically represent price trends, i.e. maximum, low, opening and closing prices for the machine to be learned. Other indicators that turned out to be successful in a combination were the momentum and RSI indicators. These made it possible to smooth out individual outliers caused by a sudden increase or decrease in the share value. After successful scaling and selection of the existing characteristics, the objects were classified. The aim of this was to look for recurring feature patterns, to be able to react better to new situations during the training phases. These feature patterns can be then assigned to different subsets and included in the analysis.
In conclusion, it can be said that scientific discourse is continuously expanding with the application possibilities of machine learning. On the one hand, this is due to the progressive development in technology and science, which makes various processes increasingly complex and abstract. Globally renowned companies invest enormous capacities in the area of machine learning to be able to automate processes with increasing complexity by machines.
This combination of increasing demands and capacities results in the following scenario concerning future analyzes: Will machine learning enable complete transparency of the financial market in the future?