Hexital - Incremental Technical Analysis Library

---

Documentation: https://merlinr.github.io/Hexital/

Source Code: https://github.com/MerlinR/Hexital

---

Hexital is a fast, zero-dependency Python library for technical analysis. It computes indicators incrementally — append a candle, get the new reading — instead of recalculating the entire series each time.

• Fast — built for live feeds and append-one-candle workflows
• Easy — dicts, lists, or Candle objects as input
• Versatile — indicators, patterns, candlestick transforms, analysis helpers
• Lightweight — no pandas or numpy required at runtime

Beta: Breaking changes are still possible. See the Release Notes.

---

Installation

pip install hexital

Development branch:

pip install git+https://github.com/merlinr/hexital.git@development

---

Choose your path

I want to…	Use	Guide
Compute one indicator on a live feed	EMA(...).append()	Quick Start
Run several indicators on one candle stream	Hexital(...)	Strategies
Load candles from CSV, Pandas, timestamps	Candle.from_dicts() etc.	Candles
Build 5m bars from 1m data	timeframe= on indicator + label on candles	Candles · Features
Check crossovers, rising/falling	hexital.analysis	Analysis
Heikin-Ashi or other candle transforms	candlestick=	Candlesticks
Write my own indicator	subclass Indicator	Custom indicators
Browse what's built in	catalogues	Indicators · Patterns

New here? Quick Start walks through the examples below step by step.

---

Getting started

Single indicator

from hexital import EMA, Candle

candles = Candle.from_dicts([
    {"open": 17213, "high": 2395, "low": 7813, "close": 3615, "volume": 19661},
    {"open": 1301, "high": 3007, "low": 11626, "close": 19048, "volume": 28909},
])

ema = EMA(candles=candles, period=3)
ema.calculate()
print(ema.reading())  # 8408.7552

# Append updates the reading automatically
ema.append(Candle.from_dict({"open": 19723, "high": 4837, "low": 11631, "close": 6231, "volume": 38993}))
print(ema.reading())  # 7319.8776

Hexital — multiple indicators, one candle stream

Use [Hexital][hexital.core.hexital.Hexital] when a strategy needs several indicators fed from the same candles:

from hexital import EMA, WMA, Candle, Hexital

candles = Candle.from_dicts([
    {"open": 17213, "high": 2395, "low": 7813, "close": 3615, "volume": 19661},
    {"open": 1301, "high": 3007, "low": 11626, "close": 19048, "volume": 28909},
    {"open": 12615, "high": 923, "low": 7318, "close": 1351, "volume": 33765},
])

strategy = Hexital("Demo Strat", candles, [
    WMA(name="WMA", period=8),
    EMA(period=3),
])
strategy.calculate()

print(strategy.reading("EMA_3"))  # 8408.7552
print(strategy.reading("WMA"))    # 9316.4722

strategy.append(Candle.from_dict({"open": 19723, "high": 4837, "low": 11631, "close": 6231, "volume": 38993}))
print(strategy.reading("EMA_3"))  # 7319.8776
print(strategy.reading("WMA"))    # 8934.9722

Named indicators keep stable keys (WMA). Unnamed indicators get generated names from type and settings (EMA_3 = EMA with period 3). Nested dict fields use : at lookup time (e.g. MACD_12_26_9:signal).

---

What's included

Indicators

40+ incremental indicators for common strategies. Full reference: indicator catalogue.

ADX · AO · Amorph · AROON · ATR · BBANDS / BandWidth · CCI · ChandelierExit / CKSP · CMF · CMO · COPC · Counter · DEMA · Donchian · EMA · Fisher · HL / HLA / HLCA · HMA · Ichimoku · JMA · KAMA · KC · KST · LinearRegression / RegressionSlope / RegressionChannel · MACD · MFI · MOP · NATR · OBV · PPO · PSAR · PivotPoints · RMA · ROC · RSI · RVI · SMA · Squeeze / SqueezePro · STDEV / STDEVT · STOCH · Supertrend · TEMA · TR · TRIX · TSI · UO · Vortex · VWAP · VWMA · WillR · WMA · ZScore

Candlestick patterns

Pattern detection on candle sequences — full catalogue.

doji · dojistar · hammer · inverted_hammer

Candlestick types

Transform incoming candles before indicators run (e.g. Heikin-Ashi) — catalogue.

HeikinAshi

Movements

Pine Script–style helpers for indicator behaviour over time — full catalogue.

positive / negative · rising / falling · mean_rising / mean_falling · highest / lowest · highestbar / lowestbar · cross / crossover / crossunder · value_range

from hexital.analysis import cross, rising

rising(ema, "EMA_3", length=8)
cross(strategy, "EMA_3", "WMA")

---

Testing & performance

Accuracy

Every built-in indicator is unit tested against Pandas-TA as a source of truth. Values are compared with a small tolerance where floating-point or formula differences apply.

When to use Hexital vs Pandas-TA

Use case	Better fit
Live / streaming — append one candle at a time	Hexital
Large bulk backtest — load full history once, vectorise	Pandas-TA

Hexital only calculates missing readings on append (O(1) per update). Libraries built on pandas typically recompute or reshape the full frame on each append, which gets slower as history grows.

In internal benchmarks, Hexital stays roughly flat as candle count increases during incremental updates, while Pandas-TA time grows with series length. For bulk calculation on large static datasets, Pandas-TA is often faster.

More detail and charts: Features.

---

Learn more

• Features — chaining, custom indicators, multi-timeframe, benchmarks
• API Reference
• Design & alternatives — vs Pandas-TA and TALipp

---

License

MIT — see LICENSE.