Can someone suggest a topic or some reasonably narrow area in financial time series analysis (e.g. statistical, machine learning, etc.) which can make a good topic for a master thesis? By 'good' I mean that it's perspective i.e. can be extended to PhD-level studies, but still manageable in 3-4 month time for an applied math master student. Thank you.
Edit: Thank you for the feedback. The field is indeed very broad and well developed that's why I would like to have some input from the community. Let's narrow down to high frequency financial time series or market microstructure (but if you come up with ideas in different areas I would be most grateful to hear it).