Straight Non-Linear Thinking

Legend has it that Burton Malkiel once created some fake market prices based on coin flips, and then showed it to a chartist. The chartist did some eye inspection and chanted enthusiastically that now is the perfect time to buy the stock! This visual confusion persuaded Burton about his random-walk hypothesis, so much so that he

Are the markets random?

In this post we want to share a method of thinking instead of a machine-learning technique or quantitative tool. Building a machine-learning algorithm that can systematically profit in the markets is like riding on a long and bumpy road full of surprises and sudden turns.

Straight Non-Linear Thinking:

In this post, we first examine the difference between the baseline accuracies of market-data compared to a random-normal sequence.Then turn to the three-consecutive-drop decision boundary. Combining the observations made in the first two steps, we reformulate the classification problem, and compare the performances of the neural networks before and after this new formulation. The highlights of the findings are listed at the end.

Detecting a Simple Pattern Using Neural Networks (part-2):