Your First KabuForge Backtest: From Inputs to Results

For your first backtest, check the complete sequence: prepare inputs, validate settings, run the calculation and inspect the results. This page uses fictional synthetic data. Its numbers cannot be read as investment performance from real stock prices.

Being able to explain what you supplied, what was calculated and where outputs were saved prepares you for the next research task. See what backtesting means and costs and turnover for the basic concepts.

Choose your version in Get started and complete its installation steps and checks. If an AI assistant helps with preparation, use the prompt on that page and specify the version and dedicated environment. Run the demo in the same virtual environment used during installation.

The stable offline demo and the main research GUI courses provide evidence for different versions. Screenshots from main do not establish that the stable package installed successfully. Keep installation status separate from operational illustrations and note the source tag or version and environment.

Open the existing offline demo and run its published commands in order from the source root. Choose a fresh output location and keep it separate from previous runs and your own research data. The demo creates configurations and synthetic snapshots.

Check that the configuration files, input snapshots and explicit timeline file have been generated. The inputs remain synthetic at this stage. Familiar-looking security labels or an equity curve do not turn the output into a real-market trading record.

The demo's validation step checks configuration structure and required fields. Once it succeeds, continue with the published history step. History consumes the explicit timeline and creates a report and local SQLite ledger. Track which files are inputs to each operation as you proceed.

If validation fails, inspect file locations, required fields, environment and version. Do not ignore errors and accept the results alone. A valid schema also leaves the economic suitability of the assumptions as a separate research question.

In the output location, check targets, risk checks, order intents and simulated events together. Distinguish planning an order from filling it in the simulation, and inspect the sequence on the timeline. Comparing settings with results makes it easier to follow how conditions affected outputs.

Before comparing reruns, check that synthetic inputs, configuration, timeline, source version and environment match. This can establish reproducibility of the software simulation under the same conditions. It does not establish real-market profitability or historical data availability.

After completing one workflow, preserve the generated inputs and results and record the scope verified with synthetic data. Continue to your first J-Quants research run with your local price file. Real market data is not bundled, so explicitly select, validate and freeze your own input.

With real data, price basis, missing values, availability timing, costs and fill assumptions need their own checks. Do not skip them because the synthetic demo succeeded; review timing and lookahead for each research run.

Source: the public GUI research courses. This article teaches research procedures; it does not recommend securities or trades or guarantee future returns. See the disclaimer.

Explore the backtesting learning path

• What Is Out-of-Sample Evaluation? Handling Periods You Have Already Inspected
• What Daily Bars Cannot Tell You: Highs, Lows and Execution Order
• What is turnover? Distinguish one-sided and two-sided measures