RNN and stock price prediction – what and how

Feedforward networks, such as densely connected networks, have no memory and treat each input as a whole. For example, an image input represented as a vector of pixels gets processed by a feedforward network in one single step. But time series data, such as stock prices for the last 10 or 20 days, are better processed with a network with memory; assume the prices of the past 10 days are X₁, X₂, ..., X₁₀, with X₁ being the oldest and X₁₀ the latest, then all 10-day prices can be treated as one sequence input, and when RNN processes such an input, the following steps occur:

1. A specific RNN cell, connected to the first element, X₁, in the sequence, processes X₁ and gets its output, y₁
2. Another RNN cell, connected to the next element, X₂, in the sequence input, uses X₂, as well as the previous output, y₁, to get the next output, y₂
3. The process gets repeated: when using an RNN cell to process the X_i element in the input sequence at time step i, the previous...