← All tools · Guides

Local Python and SQLite for an after-close China A-share review

Local review series, part 1. The bars stay on your PC, the stack costs nothing, and the later notes build on this file.

How to choose: Stay on the computer you already use for review when the job only has to run after the close. The stack costs nothing, the bars never leave the machine, and they sit next to the charting app you already open. The cost is that the PC has to be awake. A sleeping laptop will not run the later tasks. If the machine must stay up overnight, start with the cloud server note instead. For a daily review after the close, local is enough. On Windows, install Python 3.12 or 3.13 from python.org and tick Add python.exe to PATH on the first screen. Skip the Microsoft Store build. Its virtual environments misbehave. On macOS, Homebrew is enough: brew install python. Confirm with python --version. If pip is slow from mainland China, point it at the Tsinghua index the Chinese page uses. Then create D:/quant/.venv so this stack does not share packages with anything else, activate it, and install akshare and pandas into it. SQLite needs no separate server. Python's sqlite3 module opens one file, D:/quant/astock.db. DB Browser for SQLite is the free viewer when you want to see the rows. The first table is daily: code, date, open, high, low, close, volume, with a primary key on code and date, so the same bar cannot be stored twice. Prove the path by inserting 600519 on 2025-08-29 and printing that row. This page does not download a watchlist, does not schedule a task, and does not call a model. Those are later notes. It is not the tool-order guide and not the site FAQ.

Who it is for: Someone setting up a China A-share review on their own PC, who wants Python, a virtual environment, and an empty SQLite file before any data script runs. What you need from this page is the PATH checkbox, the venv, the directory layout, and a table that rejects a duplicate bar.

Who it is not for: Someone who wants the machine on all night, the whole A-share universe, or a commercial data contract. A machine that must stay awake starts with the cloud server note. Filling the table from a short watchlist is the next note. Pulling every listed name is the AKShare database walkthrough. This page is also not the tool-order guide and not the site FAQ. Do not put the database file in OneDrive or any synced folder. The sync client locks the file and writes fail.

The same walkthrough in Chinese is a local Python environment and the first SQLite file. A machine that stays on, including the first SSH login, is opening a cloud server. The same article in Chinese is the Chinese cloud server note. That note is not this article. The next part of this series, a short watchlist written into the table you create here, is the local watchlist sync. The same article in Chinese is the Chinese watchlist sync. That article is not this one either. Storing the whole market is AKShare into SQLite. That walkthrough is a different job.

Facts below follow the Chinese walkthrough. The commands were not re-run for this English page. This is not investment advice. Software versions follow the publisher's current release.

01 · Local or cloud

Use the computer in front of you when all three of these are true:

• The job runs after the close. You are not asking a laptop to wake itself at 2 a.m.
• The bars should stay on that machine. Nothing in this series uploads the database.
• The budget is zero. Python, SQLite, and DB Browser for SQLite are free.

The tradeoff is concrete. The PC has to be on, and sleep interrupts the later scheduled tasks. A review you run yourself after the close is unaffected. If you want the machine up when the lid is shut, leave this path and start with opening a cloud server.

02 · Install Python, and do not skip the checkbox

• Windows. Download the 3.12 or 3.13 installer from python.org. On the first screen, tick Add python.exe to PATH. If that box is clear, every later python command fails to resolve. Skip the Microsoft Store build. Its virtual environments misbehave.
• macOS. Homebrew: brew install python.

Open a terminal and check the version. If downloads from PyPI stall on a connection in mainland China, set the index the Chinese page uses. Then give this project its own environment:

python --version          # 3.12 or newer means the install worked

pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple

# A project-local environment, so this stack does not share packages
cd D:/quant && python -m venv .venv
# Windows:        .venv\Scripts\activate
# macOS:          source .venv/bin/activate
pip install akshare pandas

The prompt should show (.venv) before the install. akshare and pandas are not used on this page. The next note imports both.

03 · SQLite: a database you do not install

SQLite is one file. Python already includes the driver, so there is no service to start and no port to open. For a personal price history, that is the right default. Add a viewer if you want to look at the rows without writing SQL:

• DB Browser for SQLite, free and open source. Double-click the .db file and browse the table.
• From the terminal, import sqlite3 is enough for this page.

04 · The project directory and the first table

D:/quant/                    # project root
├── .venv/                   # virtual environment
├── scripts/                 # scripts for the later notes
├── logs/                    # one log per scheduled job
├── briefs/                  # archived review notes (part 3)
└── astock.db                # the market database, one SQLite file

import sqlite3
conn = sqlite3.connect("D:/quant/astock.db")
conn.execute("""
CREATE TABLE IF NOT EXISTS daily(
  code TEXT, date TEXT,
  open REAL, high REAL, low REAL, close REAL, volume REAL,
  PRIMARY KEY(code, date)
)""")
conn.close()

PRIMARY KEY(code, date) allows one bar per name per day. A repeated insert raises an error. That is the guard the next note depends on when it skips a row it already stored. CREATE TABLE IF NOT EXISTS means you can run the snippet again without dropping the file.

05 · Check: ten lines through the whole path

import sqlite3
conn = sqlite3.connect("D:/quant/astock.db")
conn.execute("INSERT INTO daily VALUES('600519','2025-08-29',1488,1498,1476,1490,23345)")
print(conn.execute("SELECT * FROM daily").fetchall())
conn.commit(); conn.close()
# The printed row is the one just inserted. Python and SQLite are connected.

Run it with the virtual environment active. On Windows, python -X utf8 avoids a garbled console. Open astock.db in DB Browser for SQLite if you want to see the same row outside the terminal. The numbers are a sample row from the Chinese page, not a live quote.

06 · Pitfalls

• Spaces or non-ASCII characters in the path. Some libraries handle them badly. Keep the project root short and plain, such as D:/quant.
• The .db file inside OneDrive or another synced folder. The sync client locks the file and writes fail. Leave the database in a local directory.
• The virtual environment was not activated. No module named akshare almost always means the prompt is missing (.venv).
• Garbled text in the Windows console. Run the script as python -X utf8.