Orange sphere ML module is trying to predict the future stock price based on its past quotes (a form of time series analysis).
The main idea behind this is a Technical analysis. Technical analysis is a trading discipline employed to evaluate investments and identify trading opportunities by analyzing statistical trends gathered from trading activity, such as price movement and volume, patterns of price movements, different types of indicators and various other analytical charting techniques.
Many traders use technical analysis in investment decisions and some of them like Larry Williams are successful. So we accept this success of some traders as evidence of the efficiency of technical analysis.
But how we can apply ML to technical analysis?
LeCun and his colleagues introduced Convolutional neural networks (CNN) in 1995 (LeCun et al., 1995; Gardner & Dorling, 1998). The name “Convolutional neural network” indicates that the network employs a mathematical operation called convolution. The operation of a convolutional neural network is usually interpreted as a transition from specific features of the image to more abstract details, and then to even more abstract details, up to highlighting concepts of a high level. This transition means applying some sort of filtering to get classifications signals. And thanks to backpropagation the network self-adjusts and generates the necessary hierarchy of abstract features (sequence of feature maps), filtering unimportant details and highlighting the essential.
On the picture below you can find basic 1-dimensional Convolutional neural network Architecture:
We are going to build Orange Sphere ML architecture based on CNN. So, we need some data inputs. Then CNN will process those inputs trough it’s deep neural network hidden layers and will create an output. As an input, we will use data from Yahoo finance. And as an output from Orange Sphere ML, we will get a prediction of stock price movements for a 3 – 6 months period.
Orange Sphere ML is under testing process now.