2 Comments
User's avatar
Rainbow Roxy's avatar

Thanks for writing this, it clarifies a lot; thinking about volatility as a lifecycle makes so much sense, it's way more intutive than just a plain number. It kinda makes you wonder how much more predictive power you'd unlock once you layer in some proper ML models to detect those transitions dynamically, especially with all the unstructured data floating around.

K. Iyer's avatar

Thank you. And I agree. In fact my two recent posts looked at replicating the results from a recent paper on “predicting” crashes. Got some decent results.

https://open.substack.com/pub/kniyer/p/testing-a-hybrid-ml-ensemble-arxiv