User Guide

Welcome to the Rhoa User Guide. This section provides in-depth conceptual documentation to help you understand and effectively use Rhoa’s features.

Guides

Overview

Rhoa is designed to streamline technical analysis and machine learning workflows for financial time series data. It extends pandas with specialized accessors that integrate seamlessly with your existing data science pipelines.

Core Philosophy

Pandas-Native Integration

Rhoa extends pandas rather than replacing it. Use familiar pandas syntax with powerful financial analysis capabilities.

Sensible Defaults

Common technical indicators work out of the box with industry-standard parameters. Advanced users can customize every aspect.

ML-First Design

The target generation system is built specifically for machine learning, with automatic optimization and proper validation handling.

Production-Ready

Type hints, comprehensive tests, and clear error messages make Rhoa suitable for production environments.

Who This Guide Is For

Beginners

If you’re new to technical analysis or machine learning, start with Basic Concepts to understand core concepts.

Data Scientists

Experienced with ML but new to finance? Jump to Targets Guide to learn how Rhoa’s target generation works.

Traders

Familiar with technical indicators? See Indicators Guide for implementation details and best practices.

Everyone

The Visualization Guide shows how to create publication-quality charts and evaluate model performance.

Getting Started

If you haven’t installed Rhoa yet, see the Installation guide.

For a quick introduction, check out the Quick Start tutorial.

For hands-on examples, browse the Examples section.

Module Organization

Rhoa is organized into focused modules:

indicators

Technical analysis indicators accessible via .indicators accessor on pandas DataFrame and Series. Includes trend, momentum, volatility, and oscillator indicators.

targets

ML target generation with automatic optimization. Creates binary classification targets optimized for your data and trading strategy.

plots

Visualization tools for model evaluation. Plot predictions, confusion matrices, and performance analysis.

data

Data utilities for importing from various sources (currently Google Sheets, with more coming).

strategy (planned)

Backtesting and strategy evaluation framework.

preprocess (planned)

Data cleaning, normalization, and feature engineering utilities.

Best Practices

Always Import Rhoa

The accessors only become available after importing:

import rhoa  # Required to register accessors
Handle NaN Values

Technical indicators create NaN for initial periods. Always handle these:

df = df.dropna()  # Or use appropriate forward/backward fill
Use Time-Based Splits

For financial data, always use time-based train/test splits, never random splits:

# CORRECT
split_idx = int(len(df) * 0.8)
train, test = df[:split_idx], df[split_idx:]

# WRONG
train, test = train_test_split(df)  # Don't do this!
Save Metadata

When generating targets, always save the metadata for reproducibility:

targets, meta = generate_target_combinations(df, mode='auto')

import json
with open('target_meta.json', 'w') as f:
    json.dump(meta, f)
Validate Out-of-Sample

Never validate on data used for target generation or training:

# Generate targets on training data only
train_targets, meta = generate_target_combinations(train_df, mode='auto')

# Apply same parameters to test data
# (using meta parameters)

Common Patterns

Multi-Indicator Analysis

# Calculate multiple indicators
df['SMA_50'] = df.rhoa.indicators.sma(50)
df['RSI'] = df.rhoa.indicators.rsi(14)
macd = df.rhoa.indicators.macd()
df['MACD'] = macd['macd']

Feature Engineering

# Create features for ML
df['SMA_20'] = df.rhoa.indicators.sma(20)
df['Returns'] = df['Close'].pct_change()
df['Volatility'] = df.rhoa.indicators.ewmstd(span=20)

Target Generation

from rhoa.targets import generate_target_combinations

targets, meta = generate_target_combinations(
    df, mode='auto', target_class_balance=0.4
)

Model Evaluation

# Visualize predictions
fig = df.rhoa.plots.signal(
    y_pred=predictions,
    y_true=ground_truth,
    date_col='Date',
    price_col='Close'
)

Next Steps

Continue to the individual guides:

Need Help?