Kwantx publishes research on future equity return and risk profiles, including state of the art machine learning architectures, frameworks for timing factor returns, and orthogonal alpha factors. We welcome collaborators from academia and industry for future research publications.
FinBert To Detect Twitter Breaking News Events
We design a framework of a bidirectional Bert model to understand text, aimed to detect which Twitter stories are market moving on a one-minute time scale. The framework can be used for equities and may be scalable to other asset classes and time frequencies.
Factor Crowding Detects Abnormal Returns
We summarize existing literature and propose a framework from Dou, 2014 that uses pairwise correlation for measure crowding of equities and factors. The framework identifies a statistically significant negative abnormal return in the momentum and value factor future returns.
Subscribe To Future Research
Stay up to date on future research publications.