Research/Earlier Research
Deep learning
price forecasts.
A summer research project in Professor Weiping Zhang’s group, School of Economics, Fudan University.
- Group
- Professor Weiping Zhang, School of Economics, Fudan
- Period
- Jun – Sep 2023
- Data
- Full A-share database
- Role
- Group project
Schematic · not dataDoes a forecast survive a portfolio?
A price forecast only matters if it changes a decision. The project asked whether deep learning predictions hold up once they drive a portfolio.
Predict, then rebalance monthly.
Using the full A-share database, we implemented deep learning models to predict prices, and ran a backtest of a strategy that adjusts the portfolio monthly according to the predictions.
Forecasts and decisions.
My later independent Kronos evaluation also asks how forecasts translate into portfolio decisions; it uses a separate model and study design.
What this page does not show.
- A group project; no results are shown here.
- The figure is a schematic, not data.