4Ds for Traders & Investors - Deep Pipes

This post marks the first in a series of 4 posts, that cover next generation building blocks for AI-driven applications in finance: Deep Pipes, Deep Learning, Deep Graph and Deep Threat.

It is a widely accepted fact that any digitization effort goes hand in hand with the increase in the cyber-attack surface. There is almost no day when the media does not report on attacks and blackmail attempts. The effects can be dramatic, and that’s why we always add ‘Deep Threat’ as the fourth building block of any AI-driven application.

Make Alternative Data Smart

The amount of data generated every single day is enormous and the ratio between non-traditional or alternative data and traditional finance data is increasing fast.

Financial institutions and individual traders and investment managers know about the huge potential of data beyond SEC filings, financial reports, or market data in search of alpha-generating ideas, evaluating investments and managing portfolios.

Data sources like (IoT) sensors and satellite imagery can help measuring economic activities including the time of day when activities peak. And use cases where alternative data provide a solid foundation for valuable insights for the finance domain are continuously increasing.

The reason is crystal clear: A quarterly time grid is no longer sufficient to monitor volatile business & economic environments to differentiate between opportunities and risks.

Knowing about the benefits of alternative (big) data and being able to generate insights, however, is a completely different story, hindered by many technological barriers. Even worse, it is no longer sufficient to implement point to point connections from a certain data source to a specific (e.g.) data warehouse.

Beneficial data sources are continuously changing, and their management must keep pace with ever-changing data requirements.

Fast and flexible access to a wide variety of data sources, however, is not the end of the game. Sorry for that but connecting data silos with some intermediate data transformations just shifts the problem to the last mile (or application layer).

Gaining value and ROI requires more than flexible data fusion platforms:

Artificial intelligence must be applied to the first mile to contextualize data and make them smart before reaching their destinations.

Need for Industrialized AI

AI applications right now are artisanal handcrafted products. Accompanied by a plethora of different technologies and environments. Today’s AI production process is a bottleneck in volatile environments where the right data-centric answer must be given on demand and just in time.

AI products must be turned into industrialized assets that can be produced fast and with a unified technology and production line for the entire AI triad: data, algorithms, and applications.

Deep Pipes

Deep Pipes is a smart data fusion component for traders & investors and defines the first ‘D’ of our next generation 4D architecture.

‘Deep Pipes’ seamlessly integrates with PredictiveWorks. and brings the power of this declarative code-free AI application factory to data fusion.

In this way, alternative data from whatever direction and sources becomes smart data right from the start. Stay tuned and next, we briefly sketch, industrialized deep learning.