Yet 10-15 years ago I have read a lot of articles about SVM using on finance (it was a special site about different SVM applications), but...results were not good enough. It seems me, that for time-series we should use other algorithms.
I think SVR would fit better than SVM in financial area, as they are regression case studies. Because most of the financial data are numeric (i.e., the targets).
For example, predicting the stock market prices using SVR..
Moreover, SVM has been applied for the problem of detecting the fraud in the banks transitions.
Supporting vector machines (SVMs) and neural networks are among the most current and innovative models at the moment. Implementing them in the area of public finances would be a new scope of application of this meaningful but sophisticated algorithm and would provide useful answers to your questions.