Qlib Online Account Simulation and Daily Trading Workflow

Summary

This Qlib documentation explains how to simulate model-driven trading through a persistent account. A configuration specifies a model, a portfolio strategy, and initial cash; the system then generates orders from predictions, executes them under exchange settings, and updates positions and reports. Users can simulate a date range or run the generate, execute, and update steps on each trading date, then compare account performance with a benchmark.

The document describes account files, order and transaction records, configurable execution costs, price assumptions, and examples of performance statistics with and without costs. These examples illustrate the reporting workflow, not a general claim about strategy performance. The simulation’s results depend on the chosen model, strategy, data, exchange assumptions, and timing conventions. Accounts store model and strategy objects alongside positions and reports, and custom models must meet the module’s interface requirements. The guide explains operational setup and evaluation, but does not establish that simulated fills or results will match live trading.

Key ideas

• A configuration connects a prediction model, portfolio strategy, and initial account capital.
• The daily workflow generates orders, executes them under exchange assumptions, and updates account state.
• Performance can be compared with a benchmark and reported with and without transaction costs.
• Results depend on data, timing, strategy, and execution assumptions and may differ from live trading.

Tags

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