Stock Market Prediction using Machine Learning

Using computers to predict stock market futures has long been a big dream for many companies to achieve.  The challenge is down to several key factors thought.

1.  BIG DATA, Market Data for the whole market, … over 13000 stocks over many years has just not been doable with current computers or machine learning algorithms.
2. Seeing the data in all possible dimensions at one time requires massive 100,000 core or more computer networks. Who has these in their office.
3. Things that affect the market are obviously more than just past technical data. It involves politics, News, Trends, Psychology and much other soft information with a big picture unpredictability.

Even with much higher power processing power in your pocket and desktop it still lacks the depth of intelligence to get the whole deep dimensional picture.

Neural Nets.  Or what we call Deep Learning involves 30 or more deep networks and lots of preprogramming to classify the data.  The more you have the better you get. This takes again lots of time and works getting and teaching the classifications of data patterns.  But it has potential.

Boon Logic discovers Pattern Memory

A new technology called Pattern Memory appears to be a breakthrough on the horizon. There are big efforts and billions of dollars going into research on how the human brain can learn and an unlimited number of patterns with instant recall.

The people at Boon Logic have found a way to emulate via algorithm and standard processor the way the human brain represents complex patterns.   They call it Cycles.

The big breakthrough appears to be its speed, it’s speed is not limited to the size of the patterns, just like the human brain.

Refer to BOONLOGIC.COM for more information.

Related to the stock market, Boon Logic has run 10 years of 13000 stocks of trading data and created a pattern set that repeats with regularity.   This can be used to find where current trend fit into and to what probability they would have of occurring again.