Deep Learning Stock Models and Backtest Data Constraints

Summary

This post describes an author’s deep-learning model built with market data and shared through open APIs, alongside links to simulation and backtest examples. It discusses stock selection from large-cap and mid-cap Chinese equity universes and says users can change the stock list. The author reports that one backtest example uses a short period chosen for higher volatility, while another aims to support longer runs by reducing data-read time and increasing request capacity. These descriptions present a model and workflow, but the post provides no performance metrics or enough methodological detail to assess the model’s predictive edge.

The author emphasizes important operating limits: some examples use fixed data and are not suitable for simulated live trading, and the API’s stability and server resources constrain the test period and universe size. The post also gives guidance on keeping the start date within the available history. These caveats make the material more useful as an illustration of data and compute constraints than as evidence that the model generalizes. No risk-adjusted results, benchmark comparison, or out-of-sample evaluation are supplied.

Key ideas

• The post describes a deep-learning model built from market data and shared through APIs.
• Example backtests cover Chinese stock universes, with user-adjustable stock selections.
• Some backtest data is fixed, limiting its suitability for simulated live trading.
• API reliability and server resources constrain the sample period and universe size.
• The document provides no performance metrics or evidence of out-of-sample profitability.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.