Stock Forecasting
Improvement on Precision score using Perforated Backpropagation™ + HIST
CSI 300 is a stock forecasting dataset
Stock Features
Top 300 stocks in the Chinese A-share market
Opening price, closing price, highest price, lowest price, volume weighted average price and volume
Looks back 60 days to construct a 360-dimensional historical stock data
HIST out of Microsoft research (2022)
Graph-based framework
Mine the concept-oriented shared information between stocks for stock trend forecasting
Highest Annualized Return in Microsoft’s qlib model library
Graph shows test precision on CSI 300 data
Best result shows Increase in precision score by 8%
Average improvement over 5 experiments of 4.2%
Baseline of precision score is 50%
Could be considered a 30% average improvement in decision making