Our algorithm was developed when the founders of vivimetrix, Brian Rossignol and Aaron French, discussed WM - double-bottom, double-top - patterns in financial markets. Delving into the world of Fibonacci and Fibonacci-derived trading ratios was easy for the pair having both studied subjects such as game theory, and advanced mathematics including the Fibonacci sequence, Fibonacci ratios, and the Golden Ratio. Once the ratios were understood, along with all the various types of W and M-shaped patterns – Butterfly, Bat, Crab, Shark and Cypher – it was time to start on the pattern recognition.

This proved to be far more difficult. Designing a single algorithm that both recognizes W and M-shaped patterns in value/time charts while also making sure those patterns conform to specific ratios was no easy task. A lot of back-testing was done on known patterns, both good and bad, and running the algorithm line-by-line to watch each variable until the algorithm was recognizing new WM patterns with a high degree of accuracy.

The idea was then introduced to use an AI approach and run the algorithm on historical market data for every stock in the three major markets for the last 10 years. That is more than 25 million full daily charts (10.3 billion data points, if you’re keeping score)!

Using this methodology, not only could the algorithm learn from historical data, but it could apply that knowledge to current patterns, allowing it to better recognize positive patterns and make more accurate predictions, and it can now continue to learn every day!

The algorithm will continue to learn as it works and it will be taught, soon, to recognize other patterns. It will also be used to predict stocks that are trending, over time (days, weeks, even months), toward being beneficial to monitor.