How to influence time to alpha using API data sources?
Everyone is on the hunt for new and useful data sources to assist them in making investment decisions. Ideally, we have access to robust data sets, complete with 3-5 years of history, which can be used in training our machine learning models, which help us understand our investment performance against specific market indexes. However, there are many valuable signals emerging from across social networks, digital sensors, traffic cameras, and other emerging API-driven sources, which can be used to make our models more reactive, and used to understand market movements in real time.
Even with these rich API resources emerging, going from discovery to alpha takes a lot of work. You can’t just tap into the Twitter API or the neighborhood traffic cameras, and immediately begin harvesting signals that can be used to better understand how markets are moving. The time to alpha involves discovering new APIs, mapping out their surface area, defining and configuring authentication, then dialing in exactly the type of integration you need based upon the paths and parameters that are available. Then once you have a successful integration with a new API data source, you still have to understand how it all fits in with your existing models, how it can be used to better understand the landscape, based upon your current investments.
The time to alpha is significantly reduced when you are working with APIs that are predefined, using machine readable API formats like OpenAPI (fka Swagger). These API specifications provide you with a plug and play map to help you with API discovery, allowing your systems to instantly understand the surface area of an API, the enumerated values, schema, and even the authentication mechanisms in use. Helping reduce API discovery and integrate down from days or hours to minutes, significantly reducing the time it will take to begin using a new API-driven data source. Leaving analysts doing what they do best, understanding how to take these new signals and begin to make sense of how they might be used to evolve investment models and help them become more responsive, and intelligent.
There are thousands of API data sources available today, with many more coming online each day. Exponentially increasing the opportunities for giving investment models the edge they need, however with each API it also means that there will be exponentially more work involved before alpha can be realized. Increasing the need to be able to work with the highest quality APIs, which provide a machine-readable definition, automating the discovery and integration with new data sources. If an API hasn’t properly defined itself and doesn’t possess all the common building blocks of a modern API operation, it is best to move on, as there are often multiple sources for the same data in today’s API economy. Where only the best of breed sources of information, providing the most relevant signals, via a real-time API will survive.