Is it possible to model time series data that fluctuates. The main solution is to take first differences and make it easier to fit conventional models. What if non-linear models are built? Can they solve a noisy time series (e.g stock returns) and make good predictions? Can adding a square term or a trigonometric term or something else non-linear work? Has some researched the topic?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.