HTM for Stocks available in the App Store
Numenta Newsletter — February 18, 2016
I’m pleased to announce that we have released an iOS version of HTM for Stocks, a demonstration application that enables anybody to experience HTM technology firsthand. Previously available for Android only, HTM for Stocks now can be found in the App Store.
HTM for Stocks continuously monitors stock price, stock volume and Twitter volume for roughly 200 publicly traded companies and alerts you in real time when something unusual is happening. Results are displayed from most to least anomalous, so you can determine what is important and where to focus.
This application not only makes detecting securities anomalies easy, but it also highlights the value of HTM. For example, you don’t have to know what you’re looking for to find these real-time anomalies. HTM starts learning hundreds of models immediately. In this case, it creates a model for three data streams per stock: stock price, stock volume and Twitter volume. The application continuously compares a prediction of the next data point with the actual next point and, if the overall behavior is unexpected, it reports an anomaly. No human intervention is required. You don’t have to tune it or define parameters. You don’t need to know anything about the stocks. Because HTM for Stocks continuously learns, it adapts well to changes. For example, if a stock price spikes but then continues to stay at that level, HTM for Stocks learns the higher price is a new normal.
Perhaps one of the nicest benefits of this application is that it displays stocks from most to least anomalous activity. This ability to cut through vast amounts of noisy data and surface only the companies where something unusual is happening is extremely valuable and time-saving. In this application, we’re monitoring stocks, but you could use the same principles to find anomalous behavior in many types of data streams:
- Machine sensor data - to do preventative maintenance and know which machines need repairs
- Human behavior metrics – to identify suspicious activity
- GPS coordinates – to track vehicles and find deviations from an expected route
Like all of our sample applications, we’ve made the source code available, and we hope to see developers create derivative products. If you are interested in developing HTM for Stocks into a commercial product, you can contact us at firstname.lastname@example.org.
In the meantime, we hope that our readers who are also iPhone users will download HTM for Stocks and give it a try.