CloudQuant’s AI push adds RavenPack for alt-data

CloudQuant announced the addition of RavenPack analytics to their trading strategy incubator. Crowd researchers can now use RavenPack historical data to discover tradable alpha signals on CloudQuant’s online Python and JupyterLab-based tools.

In the next couple of months, the incubator will be releasing CloudQuant AI, which gives anybody with an internet connection the ability to do data science or alpha signal studies against alternative datasets as a free service, said CloudQuant CEO Morgan Slade on the sidelines of the Battle of the Quants conference.

“The outcome of that is an alpha signal which we can then integrate directly into our investment process, so we are really allowing people who aren’t involved in anything to do with Wall Street the ability to play a hand in running a hedge fund,” said Slade. “What we’re doing is bringing together large numbers of alternative datasets with crowd researchers, with the machine learning platform.”

One of the major tech breakthroughs of CloudQuant’s platform is something called bitemporal tables, which keeps track of multiple versions of history. For example, if you query earnings per share for financial services companies over a decade, depending on which day you ask, it will be a different answer because of revised earnings.

“There’s lots of different versions of history, and having technology that contemplates that is a new thing,” explained Slade. “It gives a correct answer instead of an answer that is not really achievable because they’ve got forward-looking bias…and that may look like alpha when in fact it’s not.”

Also presenting at Battle of the Quants was Peter Hafez, chief data scientist at RavenPack, who detailed how tier one banks are using alternative data sets.

J.P. Morgan, for example, proposed a framework for style timing in cross-asset risk, using various machine learning techniques to generate views on expected returns. And Citi’s quant research team used RavenPack’s event data, which is automatically extracted using natural language processing, to investigate how investors can profit from CapEx announcements.

In a company statement, Slade said the CloudQuant community is already finding promising signals using the datasets. Crowd-based research tools are increasing in popularity along with the rapidly growing data science field, motivating participants to access Wall Street professional-quality tools and datasets.