Skip to content

Latest commit

 

History

History
10 lines (6 loc) · 819 Bytes

README.md

File metadata and controls

10 lines (6 loc) · 819 Bytes

Time-series prediction using CNN

This model tries to predict time-series data, that is, the SPY close price from January 1993 to May 2022, by using the CNN architecture.

Based on the sliding-window algorithm, the data are labeled BUY, SELL, and HOLD. If it’s at its peak in that sliding window, then it’s a SELL. If it’s in the troughs, then it’s a BUY. Otherwise, it’s a HOLD.

We combine many technical indicators and intervals to create a B&W image such that each pixel represents a technical indicator at a certain interval.

We use this B&W image to predict at a certain price point should it SELL, BUY, or HOLD

The complete report is available in this link (written in Bahasa Indonesia)