A recent press release reports, “Alpha Vertex (www.alphavertex.ai), a financial technology company that offers analytical solutions powered by proprietary AI algorithms to financial institutions and investors, today announced the launch of Alta, its alternative data product. Created with enriched alternative datasets using big data techniques such as NLP, anomaly detection, sentiment and relation extraction, Alpha Vertex’s proprietary AI algorithms produce informational advantages and detection of new investment risk profiles for investors. Alpha Vertex takes care to control model complexity to avoid over-extrapolation and rigorously tests for the probability of false discovery. Its alternative data product and a few of its machine learning models are currently in use by a multi-billion dollar hedge fund.”
The release continues, “The use of alternative data products to glean insights is nothing new to hedge funds who have been early adopters of the solutions. In 2018, Deloitte predicted that alternative data will likely transform active investment management over the next five years. Yet, studies and leading research reports have demonstrated that most ML models and investment strategies — which use alternative data and other datasets — uncovered by companies are false.”
Mutisya Ndunda, Co-founder and CEO of Alpha Vertex, noted, “There is a growing number of companies offering Al algorithms in a box and use of alternative datasets that claim to provide an advantage or ‘alpha signals’ to investors… I would say that 90% of these are false signals for investors. We are in the 10% category because we developed an approach to root out false positives as much as possible. We don’t offer a massive array of signals and ML models and see what would work because, statistically speaking, a few random options out of 100 models will work every time. We offer very few and intentionally selected machine learning models to financial institutions. The random approach is not a long-term investment strategy.”
Read more at PR Newswire.
Image used under license from Shutterstock.com