Nostradamus Stock Predictions Configuration

Configuration for developing the Nostradamus stock prediction system using Aider with AI-powered architecture mode, automated testing, and code quality checks.

What This Configuration Does

This Aider setup optimizes development for a Go-based stock prediction application by:

Instructions

When working on the Nostradamus stock prediction system:

1. **Model Selection**

- Primary model: `openrouter/openai/o3-mini-high` for complex reasoning and predictions logic

- Editor model: `openrouter/anthropic/claude-3.5-sonnet` for code refactoring and structural changes

- Weak model: `openrouter/anthropic/claude-3-5-haiku` for quick, simple tasks

2. **Architect Mode**

- Architect mode is ENABLED

- Use it to analyze the codebase structure before making changes

- Ask for architectural guidance when adding new features

- Let Aider suggest the best files to modify for each task

3. **Testing Workflow**

- Auto-test is ENABLED

- All changes trigger: `go test cmd/main_test.go`

- Fix any test failures before proceeding

- Add new tests when implementing features

4. **Code Quality**

- Auto-lint is ENABLED

- All changes run: `golangci-lint run`

- Address linting issues immediately

- Follow Go best practices and conventions

5. **Version Control**

- Auto-commits are ENABLED

- Each successful change creates a commit

- Commits include descriptive messages based on changes

- Review commit history regularly

6. **Conventions**

- ALWAYS read `CONVENTIONS.xml` before making changes

- Follow the project's established patterns

- Maintain consistency with existing code style

Example Usage

**Adding a new prediction algorithm:**

```

I need to add a moving average prediction algorithm to the system

```

Aider will:

1. Analyze the codebase structure (architect mode)

2. Identify relevant files to modify

3. Implement the algorithm following CONVENTIONS.xml

4. Run `go test cmd/main_test.go` automatically

5. Run `golangci-lint run` to check code quality

6. Commit the changes if all checks pass

**Refactoring existing code:**

```

Refactor the prediction scoring logic to be more modular

```

Aider will use Claude 3.5 Sonnet (editor model) to restructure the code while maintaining test coverage and code quality.

Important Notes