10.6 Conclusions

As we have shown, SVMs are capable of extracting information from real life economic data. Moreover, they give an opportunity to obtain the results not very obvious at first glance. They are easily adjusted with only few parameters. This makes them particularly well suited as an underlying technique for company rating and investment risk assessment methods applied by financial institutions.

SVMs are also based on very few restrictive assumptions and can reveal effects overlooked by many other methods. They have been able to produce accurate classification results in other areas and can become an option of choice for company rating. However, in order to create a practically valuable methodology one needs to combine an SVM with an extensive data set of companies and turn to alternative formulations of SVMs better suited for processing large data sets. Overall, we have a valuable tool for company rating that can answer the requirements of the new capital regulations.