qteasy Quantitative Investment Toolkit

QTEASY is a quantitative trading strategy development toolkit designed for full-lifecycle management. Its core capabilities include:

• Financial Data Management: Fetching data from multiple channels, cleaning it, and storing it locally in a unified format. It uses DataType objects to provide structured access to complex information like adjusted prices and index constituents.
• Strategy Development: Defining strategies using the BaseStrategy class. It supports a 'building block' approach where complex strategies are composed from simple ones, and includes over 70 built-in strategies (technical indicators, moving averages, breakouts, reversals, etc.).
• Backtesting and Evaluation: Using the Operator class to run strategies against historical data following real market rhythms. It provides multi-dimensional evaluation, trading reports, and result charts.
• Optimization: Utilizing algorithms such as Simulated Annealing, Genetic Algorithms, and Bayesian Optimization to find optimal parameters in large search spaces.
• Simulated Live Trading: Running strategies against real-time market data to track logs, holdings, and account changes, ensuring the same logic used in backtesting is used in live environments to prevent 'look-ahead bias' (future functions).

qteasy is a quantitative investment toolkit designed for three primary workflows:

1. Financial Data Management:

   • Multi-channel data acquisition and cleaning with unified local storage.
   • Structured data management using DataType objects, allowing easy access to complex information like adjusted prices and index constituents.
   • Built-in visualization and statistical analysis for financial data.

2. Strategy Creation:

   • Define strategies using the BaseStrategy class for clear, intuitive logic.
   • Access over 70 built-in strategies or compose complex strategies using a 'building block' mechanism (strategy mixing and grouping).
   • Built-in protection against 'look-ahead bias' (future functions) and data leakage through encapsulated data input methods.
   • Unified logic for both backtesting and live trading.

3. Backtesting, Optimization, and Trading:

   • Manage strategy execution via the Operator class, which simulates real market rhythms and generates comprehensive performance reports.
   • Optimize strategy parameters using algorithms like Simulated Annealing, Genetic Algorithms, and Bayesian Optimization.
   • Support for simulated automated trading by tracking real-time market data, logs, holdings, and account changes.
   • Future support for real-world automated trading via QMT interfaces.

In qteasy, process data (prefixed with proc.*) refers to data that changes based on the execution path of a backtest or live trading. Unlike static historical data, process data cannot be pre-generated and must be maintained by the Backtester or Trader during runtime.

Use process data when your strategy depends on:

• Account state: Current/historical cash (proc.own_cash), available cash (proc.available_cash), or total assets (proc.total_value).
• Positions: Current holdings (proc.own_amounts), sellable quantities (proc.available_amounts), or position market value (proc.position_value).
• Trade results: Actual executed quantities (proc.trade_records), transaction costs (proc.trade_cost), or execution prices (proc.trade_price).

Key Concept: No Look-ahead Bias qteasy guarantees that when generating a signal at step $k$, the strategy cannot see the execution results of step $k$. It can only see completed historical data. For example, trade records are only visible up to step $k-1$.

The RiskManager acts as a pre-submission checkpoint. It evaluates an OrderIntent against an AccountSnapshot using a chain of rules. If any rule fails, it produces a RiskDecision (Reject) containing an English reason and a rule_id.

Configuration

To enable risk management, you must instantiate a RiskManager with a list of rules in Python and pass it to your Trader during setup:

# Example conceptual setup
risk_manager = RiskManager([rule1, rule2, ...])
trader = Trader(..., risk_manager=risk_manager)

If no RiskManager is provided, the local rule chain is skipped, and orders are sent directly to the broker.

Common Rule Scenarios

Troubleshooting Rule Rejections

If the CLI displays: Order rejected by risk rule [MAX_ORDER_QTY]: order quantity exceeds limit

1. Identify the rule_id (e.g., MAX_ORDER_QTY).
2. Search risk_log for <RISK REJECTED> to see the specific details.
3. Adjust your rule thresholds or the order parameters in your strategy.

Visualization in qteasy is decoupled from the execution engine using the HistoryPanel abstraction.

• HistoryPanel: A three-dimensional data container (shares × hdates × htypes) that serves as the standard return type for get_history_data and the input for plotting.
• Unified Plotting: Use HistoryPanel.plot(...) as the primary entry point. It automatically selects appropriate chart types (K-line, Volume, MACD, Line, etc.) based on the data htypes and shares.
• Backends: The system generates a backend-agnostic chart specification. This allows the same logic to render via matplotlib (for static images/reports) or Plotly (for interactive web views) while maintaining consistent styling and semantics.
• High-level APIs: Functions like qt.candle(...) are convenience wrappers that handle the workflow of: Fetch Data $\rightarrow$ Build HistoryPanel $\rightarrow$ Call hp.plot().