← All tools · Guides

How to choose a quant trading platform: QMT, PTrade, JoinQuant, RiceQuant, MyQuant, BigQuant

Decide where the code runs, and how it reaches your account.

How to choose: name where the code runs, and whether it must send orders on your own stock account. Research and backtests fit JoinQuant, RiceQuant, or BigQuant, or the MyQuant desktop terminal. Automatic orders on your own account go through QMT or PTrade. Opening thresholds and fees are per the broker's own disclosures. This page does not publish a number for them.

Who it is for: someone choosing a China quant trading platform, either a hosted site for research and backtests, or a broker terminal after they open an account so a program can use their own stock account.

Who it is not for: someone comparing open-source backtesting frameworks, trading bots, or the tool order of data, then a backtest, then paper trading. Those are other articles. This page also does not stand in for a broker's account-minimum and fee schedule.

The same comparison in Chinese is how to choose a quant trading platform (Chinese). The open-source tool order, data then backtest then an agent, is a different article: how to choose AI quant tools. The nine-framework backtest comparison is how to choose a backtesting framework. Neither of those is a language version of this page. The shorter backtrader note is a separate article. The QMT versus PTrade deep dive is how to choose between QMT and PTrade, and it is not a language version of this page either. A charting site for watchlists, a screener, and alerts is the TradingView setup. It does not send orders on a broker account, and it is not a language version of this page. Opening the Ubuntu machine a crawler runs on is opening a cloud server. Conda and cron on that machine are the cloud quant desk. A two-week ledger after a backtest is paper trading. The order of what to learn is the learning path. None of those sends orders, and none is a language version of this page. AI assistants, general models, platform factors, and open-source research projects are how to choose AI stock-picking tools. That page is not a language version of this one.

This site has no commercial relationship with the platforms or brokers below, and it does not offer account-opening links. Features and prices move. The table was checked against each vendor site on 15 Oct 2026. BigQuant plan prices were rechecked on 21 Oct 2026. Opening conditions and fees are per the broker's own disclosures.

01 · Six platforms, one table

02 · Broker terminals that can send an order: QMT and PTrade

These two are the ones that can actually send an order. They are not sold to you as a download from a shop. A broker buys the terminal and offers it to its own clients, so the first step is always an account at a broker that provides it. Which brokers offer it, whether there is an asset threshold, and whether there is a fee, differ by broker and sometimes by branch. That is per the broker's own disclosures. Online claims of the form "you need this much in assets" were not verified broker by broker, and they are not repeated here.

QMT is the broker-client terminal from ThinkTrader. People writing Python usually use MiniQMT plus the xtquant library. The ThinkTrader docs describe two modules, market data and trading, and they say the MiniQMT client has to be running first. The program runs on your computer. You can use your own Python and third-party libraries. The computer has to stay on, and the network has to stay up.

PTrade is Hundsun's quant trading terminal. Hundsun does not publish a public manual. The following points come from a third-party compilation, the PTrade API docs: the strategy is hosted in the broker's machine room and keeps running when your computer is off; that room is an internal network, so it generally cannot reach the public internet, and you cannot pip-install libraries, only use the built-in set; some brokers' live editions do not support an intraday backtest. The function names are close to JoinQuant, such as initialize and handle_data. The limits that apply to you are the version your broker actually runs.

Between the two: choose QMT if you need your own libraries or your own data. Choose PTrade if you do not want a computer left on and the strategy does not need outside data. MiniQMT, the application steps, whether a model API can be called, and the programmatic-trading report are how to choose between QMT and PTrade. The same article in Chinese is QMT and PTrade (Chinese). That deep dive is not a language version of this page.

03 · Hosted research sites: JoinQuant, RiceQuant, BigQuant

You write in the browser. The vendor supplies the data. There is no local install, which is why these sites suit a first strategy. The cost of that convenience is that the code runs on the vendor's servers, under the vendor's CPU, memory, bandwidth, and timeout limits.

• JoinQuant: strategies are Python. The official docs cover backtests and simulated trading, and they say a strategy runs in an isolated process with CPU, memory, bandwidth, and timeout limits. A platform membership and the local data service JQData are separate products.
• RiceQuant: besides the hosted site there is RQSDK, installed locally (RQData for data, RQAlpha Plus for backtests, and related pieces). Using it means applying for a trial. The vendor says it also publishes skill files for AI tools such as Claude Code. Its open-source backtest framework RQAlpha is limited to non-commercial use.
• BigQuant: positioned as an AI quant site, with a factor library, visual blocks, and Python. Simulated trading is included. Live trading is a separate application.

04 · A desktop terminal: MyQuant

MyQuant describes a free terminal for strategy research, backtests, trading, and analysis. It supports Windows, and strategies can be written in Python, C++, C#, or Matlab. Whether it can reach a live account is per the vendor and the broker you use.

05 · If you do not want to be locked to one vendor

Code written for the platforms above mostly runs only in that vendor's environment. Moving it means a rewrite. To keep data, backtests, and orders on machines you control, use an open-source framework. vn.py connects several China trading interfaces and also reads ThinkTrader research data. Qlib fits machine-learning research. The install and first model are in the Qlib walkthrough. The nine-framework comparison is how to choose a backtesting framework. AI research projects are in open-source AI quant projects. The same list in Chinese is open-source AI quant projects. That list is not an English alternate of this page.

06 · What you can do without paying

Research and a backtest can cost nothing. A program that sends orders on your account generally cannot skip a broker. The notes below follow what each vendor's site said. Prices were checked on 21 Oct 2026 and can change. Use the vendor's own page. Broker account thresholds and fees are a different question, and they stay per the broker's own disclosures.

• Open source, no licence fee: AKShare for data, and backtrader, Qlib, or vn.py for the backtest. AKShare, Qlib, and vn.py are MIT. backtrader is GPL-3.0. You install the stack yourself, and a free data source can break. The choice between Tushare and AKShare is how to choose between Tushare and AKShare. The same comparison in Chinese is Tushare and AKShare (Chinese).
• MyQuant: the vendor site describes a free terminal for research, backtests, trading, analysis, and management. Register, then download. Windows only.
• BigQuant: a free community edition includes basic data and backtest capacity, plus a strategy-deployment server that the vendor has offered free for a limited time. On the official price page, the 21 Oct 2026 check showed a standard plan at 1,099 yuan a year and a flagship plan at 5,499 yuan a year. The live-trading server was only on the flagship plan. Those figures are the vendor's published plan prices, not a broker account minimum.
• RiceQuant: the local RQSDK suite needs a commercial licence under the vendor's docs. You can apply for a trial first. The same docs describe a 1 GB daily data quota on a trial account. Pricing for the hosted site was not reconfirmed on this check. Use the vendor's site.
• JoinQuant: registration is enough to backtest in the browser. A paid membership also exists. Which features require it is per the vendor's site. On this check the site blocked the region we used, so that split was not reconfirmed.
• QMT and PTrade: the broker provides them. Whether they charge, and whether there is an asset threshold, is per the broker's own disclosures.

Free is not the same as costless. Bad data, or a backtest that ignores fees and slippage, will make a live account look worse than the backtest. Get a strategy running on a free setup before you pay for data or for live trading.

07 · Ask this before you open the account

• Threshold and fees: whether there is an asset threshold, whether there is a monthly fee, and whether commission changes because you turned the terminal on. The answer is per the broker's own disclosures.
• What you can trade: stocks, ETFs, convertible bonds, margin, and futures are not always all available.
• Where the strategy sits: a hosted terminal such as PTrade stores the code in the broker's machine room. A hosted research site stores it on the vendor's servers. If that bothers you, pick something that runs locally.
• Limits: quote rate, order rate, whether the runtime can reach the internet, and whether you can install libraries.
• When an order is wrong: the loss is yours. Run the same rule in a simulator first. The walkthrough is paper trading after a backtest. The same article in Chinese is paper trading.

08 · How to choose, by situation

• Just starting: JoinQuant, RiceQuant, or BigQuant. Get one very small strategy through a backtest.
• You do not want to pay yet: AKShare plus backtrader or Qlib, or BigQuant's community edition, or the MyQuant terminal.
• The strategy has been checked, you want automatic orders, and you will run your own environment: QMT (MiniQMT).
• You want the strategy to keep running, and you do not want to mind a computer: PTrade.
• You write software and want the stack on machines you control: vn.py plus a broker interface.
• You are not sure which kind of tool you need: that is a different article, open-source tools by job, and the order around it is how to choose AI quant tools.

This is not investment advice. Programmatic trading can lose money. A backtest is not a statement about future returns. Features, prices, and opening conditions follow each vendor and the broker's own disclosures.