centaur_technical_indicators 1.3.0

A collection of 50+ fully configurable technical indicators for financial analysis, powered by Rust and Python. Part of the CRT (Centaur Research & Technologies) ecosystem.

Centaur Technical Indicators

A production-ready Python library providing 50+ technical indicators for financial analysis, built on a high-performance Rust backend.

Part of the CRT (Centaur Research & Technologies) ecosystem.

Looking for the Rust crate? See: ChironMind/CentaurTechnicalIndicators-Rust

Looking for the WASM bindings? See: ChironMind/CentaurTechnicalIndicators-JS

---

🚀 Getting Started

1. Install the package:

pip install centaur_technical_indicators

2. Calculate your first indicator:

import centaur_technical_indicators as cti

prices = [100.2, 100.46, 100.53, 100.38, 100.19]

ma = cti.moving_average.single.moving_average(
    prices,
    "simple"
)
print(f"Simple Moving Average: {ma}")

Expected output:

Simple Moving Average: 100.352

3. Explore more tutorials

• 01 - Using with pandas
• 02 - Using with Plotly
• 03 - Advanced use cases
• 04 - Connecting to an API

---

🛠️ How-To Guides

Task-oriented guides for common problems and advanced scenarios.

• How to pick Bulk vs Single
• How to choose a Constant Model Type
• How to choose a Deviation Model
• How to choose a period
• How to use the McGinley dynamic function variations

---

📚 Reference

The API reference can be found here

Examples

A reference of how to call each function can be found in the tests:

• Reference Examples

Clone and run:

$ source you_venv_location/bin/activate

$ pip3 install -r test_requirements.txt

$ maturin develop

$ pytest .

Library Structure

• Modules based on their analysis areas (moving_average, momentum_indicators, strength_indicators...)
• Most modules have both bulk and single sub-modules:
  • bulk: Compute indicator over rolling periods, returns a list.
  • single: Compute indicator for the entire list, returns a single value.
• Exceptions (asymmetric shape):
  • chart_trends — flat module (no single/bulk sub-modules; functions are top-level).
  • volatility_indicators — volatility_system is bulk only; ulcer_index has both.
  • other_indicators, strength_indicators, and trend_indicators are asymmetric (not every indicator has both variants).
• types used to personalise the technical indicators (moving_average_type, deviation_model, constant_model_type...)

---

🧠 Explanation & Design

Why Centaur Technical Indicators?

• Performance: Rust-powered backend for maximal speed, safety, and low overhead.
• Configurability: Most indicators are highly customizable—tweak calculation methods, periods, or even use medians instead of means.
• Breadth: Covers a wide range of technical indicators out of the box.
• Advanced Use: Designed for users who understand technical analysis and want deep control.

Note: Some features may require background in technical analysis. See Investopedia: Technical Analysis for a primer.

---

📈 Available Indicators

All indicators are grouped and split into modules based on their analysis area.

Candle Indicators

• Ichimoku Cloud, Moving Constant Bands/Envelopes, Donchian Channels, Keltner, Supertrend, McGinley Dynamic Envelopes, McGinley Dynamic Bands

Chart Trends

• Trend break down, overall trends, peak/valley trends, peak favorable move, valley favorable move

Correlation Indicators

• Correlate asset prices

Momentum Indicators

• Chaikin Oscillator, CCI, MACD, Money Flow Index, On Balance Volume, ROC, RSI, Williams %R, Stochastic Oscillator, Slow Stochastic, Slowest Stochastic, Percentage Price Oscillator (PPO), Chande Momentum Oscillator (CMO), Signal Line (MACD signal), McGinley Dynamic Commodity Channel Index, McGinley Dynamic MACD Line

Moving Averages

• McGinley Dynamic, Moving Average

Other Indicators

• ROI, True Range, ATR, Internal Bar Strength, Positivity Indicator

Strength Indicators

• Accumulation/Distribution, PVI, NVI, RVI, Volume Index

Trend Indicators

• Aroon (Up/Down/Oscillator), Parabolic, DM, Volume-Price Trend, TSI

Volatility Indicators

• Ulcer Index, Volatility System

Note on scope: Statistical primitives available in Python's statistics / math modules (the Rust basic_indicators surface) are intentionally not re-bound. Use the standard library directly for those.

---

📊 Performance Benchmarks

Want to know how fast the library runs in real-world scenarios?
We provide detailed, reproducible benchmarks using realistic OHLCV data and a variety of indicators.

Benchmarks summary (Raspberry Pi 5)

All results are produced on a Raspberry Pi 5 (RPi5) and reported as microseconds per call (min/mean/median) and derived ops/sec. Each suite is run in two modes:

• single: run the indicator repeatedly for timing small, per-call latency
• bulk: process larger arrays to measure throughput-oriented workloads

Headline observations from the momentum suite (large 10Y dataset)

• Ultra‑lightweight indicators achieve sub‑microsecond latency per call:
  • ROC single: ~0.11 µs (≈8.72e+06 ops/sec); bulk: ~86 µs (≈1.16e+04 ops/sec)
  • OBV single: ~0.13 µs (≈7.85e+06 ops/sec); bulk: ~130 µs (≈7.7e+03 ops/sec)
• RSI single-call latency ranges roughly 45–115 µs depending on averaging method; bulk ranges ~560–3600 µs
  • Averaging method impact (fast → slow): simple/mean/exponential ≈ median < mode
• Stochastic (fast/slow/slowest) single: ~36–98 µs; bulk: ~109–2600 µs (mode again the slowest)
• CCI families:
  • “standard/mean/median/mode” single calls mostly ~39–155 µs; bulk ~230–3600 µs
  • “ulcer” variant is significantly heavier: single ~6.8–6.9 ms; bulk ~1.0–2.1 ms
• MACD line and signal line single: ~32–80 µs; bulk: ~170–4,000+ µs depending on smoothing and dataset
  • McGinley MACD line single is among the fastest (~32–33 µs); bulk ~300 µs
• Chaikin Oscillator single: ~140–300 µs; bulk: ~500–2,900 µs
• PPO single: ~36–151 µs; bulk: ~175–5,700 µs
• CMO single: ~45 µs; bulk: ~505 µs

These patterns (simple/mean/exponential being fastest; median slightly slower; mode slowest; “ulcer” notably heavy) are consistent across indicator variants and hold across single vs bulk modes.

Small dataset: 1Y daily data Medium dataset: 5Y daily data Large dataset: 10Y daily data

Browse all benchmark tables

• CentaurTechnicalIndicators-Python Benchmarks

(Your results may vary depending on platform and Python environment.)

---

🤝 Contributing

Contributions, bug reports, and feature requests are welcome!

• Open an issue
• Submit a pull request
• See CONTRIBUTING.md for guidelines

---

💬 Community & Support

• Start a discussion
• File issues
• Add your project to the Showcase

---

📰 Release Notes

Latest: See CHANGELOG.md for details.

Full changelog: See Releases for details

---

📄 License

MIT License. See LICENSE.

---

📚 More Documentation

This repository is part of a structured documentation suite:

• 📕 Tutorials: — See here
• 📘 How-To Guides: — See here
• ⏱️ Benchmarks: — See here
• 📙 Explanations: — tech.centaurresearchtechnologies.com
• 📗 Reference: — See here

---