A neural network takes in a series of inputs, or "features", that are the columns of a dataset. In this example, the features are the price predictions of a series of technical indicators.
The most basic technical indicator, and the cornerstone of many predictive strategies, is the
moving average . This is simply the average of the security's price for the last n days.
In this graph, n = 30:
Bollinger Bands are another popular indicator. These combine the security's standard deviation (σ) and
its moving average. Bollinger bands reflect price changes- the closer the prices are to the upper band, the more
overbought the market, and vice versa. The distance between the two bands is a measure of volatility.
In this graph, σ = 2 .