Old versus new tools in Quantitative Finance

I am not just a researcher. I actively trade stocks and options. As a theorist and trader, I am not satisfied with all those time series tools currently taught in Quantitative Finance. I think they are outdated, lack predictive accuracy and are too rigid to be useful in practice. My suspicion is that large financial institutions, and high frequency traders in the first place, have developed new revolutionary methods. These methods allow them to extract high profits and therefore are kept in secret. However, researchers outside the financial services industry little by little are inventing methods that prove to be better than ARIMA, ARCH-GARCH and the like. Dmitry Rastorguev has provided a nice selection of 2017 papers (available for free) on new methods in finance.

Once upon a time I looked up the 1940's issues of Doklady Akademii Nauk, a highly respected Soviet journal. They were full of papers on convergence of numerical series. These days that topic is not considered serious. I am sure the same will happen to all papers on time series models that flood contemporary journals. The sooner the better!

The only reason the traditional time series models are alive is the quality standards set by research journals: if the paper is complex and in the mainstream of contemporary research, it gets published.