GitHub

Architecture

This page provides an overview of OnlineTechnicalIndicators.jl's architecture, including module organization and type hierarchy.

Module Overview

OnlineTechnicalIndicators.jl is organized into several submodules, each with a specific responsibility:

OnlineTechnicalIndicators (main module)
├── Candlesticks     # OHLCV data structures
├── Internals        # Internal utility functions
├── Indicators       # Technical indicators (60+)
├── Patterns         # Candlestick pattern recognition
├── Wrappers         # Indicator wrappers (Smoother, DAGWrapper)
├── Factories        # Factory functions (MovingAverage)
└── SampleData       # Sample OHLCV data for testing

Module Relationships

flowchart TB subgraph OnlineTechnicalIndicators Main[OnlineTechnicalIndicators.jl] subgraph Submodules Candlesticks[Candlesticks] Internals[Internals] Indicators[Indicators] Patterns[Patterns] Wrappers[Wrappers] Factories[Factories] SampleData[SampleData] end subgraph BaseTypes["Abstract Types"] TI[TechnicalIndicator] TISO[TechnicalIndicatorSingleOutput] TIMO[TechnicalIndicatorMultiOutput] MAI[MovingAverageIndicator] end end subgraph External["External Dependencies"] OSB[OnlineStatsBase.jl] OSC[OnlineStatsChains.jl] Tables[Tables.jl] end Main --> Submodules TI --> OSB BaseTypes --> TI TISO --> TI TIMO --> TI MAI --> TISO Indicators --> BaseTypes Patterns --> BaseTypes Wrappers --> Indicators Factories --> Indicators

Type Hierarchy

All technical indicators inherit from a common type hierarchy rooted in OnlineStatsBase.OnlineStat.

Abstract Types

# From OnlineStatsBase
abstract type OnlineStat{T} end

# Base type for all technical indicators
abstract type TechnicalIndicator{T} <: OnlineStat{T} end

# Single output indicators (e.g., SMA, RSI)
abstract type TechnicalIndicatorSingleOutput{T} <: TechnicalIndicator{T} end

# Multiple output indicators (e.g., MACD, BB)
abstract type TechnicalIndicatorMultiOutput{T} <: TechnicalIndicator{T} end

# Moving average indicators
abstract type MovingAverageIndicator{T} <: TechnicalIndicatorSingleOutput{T} end

Type Hierarchy Diagram

classDiagram class OnlineStat~T~ { <<OnlineStatsBase>> } class TechnicalIndicator~T~ { <<abstract>> +value +n::Int } class TechnicalIndicatorSingleOutput~T~ { <<abstract>> +value::Union Missing T } class TechnicalIndicatorMultiOutput~T~ { <<abstract>> +value::Union Missing Val } class MovingAverageIndicator~T~ { <<abstract>> } OnlineStat <|-- TechnicalIndicator TechnicalIndicator <|-- TechnicalIndicatorSingleOutput TechnicalIndicator <|-- TechnicalIndicatorMultiOutput TechnicalIndicatorSingleOutput <|-- MovingAverageIndicator

Submodule Details

Candlesticks

Provides OHLCV (Open, High, Low, Close, Volume) data structures.

Exports:

  • OHLCV: Candlestick data structure with optional timestamp
  • OHLCVFactory: Factory for batch OHLCV creation from vectors
  • ValueExtractor: Module with extraction functions (extract_open, extract_close, etc.)
using OnlineTechnicalIndicators.Candlesticks

# Create a single candlestick
candle = OHLCV(100.0, 105.0, 98.0, 103.0; volume=1000.0, time=Date(2024,1,1))

# Create from vectors
factory = OHLCVFactory(opens, highs, lows, closes; volume=volumes)
candles = collect(factory)

Internals

Internal utilities for indicator implementation. While exported for advanced use cases, the API may change between minor versions.

Type Queries:

  • is_multi_input(T): Check if indicator requires OHLCV input
  • is_multi_output(T): Check if indicator produces multiple values
  • expected_return_type(ind): Get return type of an indicator

Value Utilities:

  • has_output_value(ind): Check if indicator has valid output
  • has_valid_values(buf, window): Check circular buffer validity
  • is_valid(x): Check if value is not missing

Calculation Functions:

  • _fit!: Internal fit function for OnlineStatsBase integration
  • _calculate_new_value: Calculate value from internal state
  • _calculate_new_value_only_from_incoming_data: Calculate from incoming data only

Indicators

Contains 60+ technical indicators organized by input/output type. See Indicators support for the complete list.

Categories:

  • SISO (Single Input, Single Output): SMA, EMA, RSI, etc.
  • SIMO (Single Input, Multiple Output): BB, MACD, StochRSI, KST
  • MISO (Multiple Input, Single Output): ATR, OBV, MFI, etc.
  • MIMO (Multiple Input, Multiple Output): Stoch, ADX, SuperTrend, etc.

Patterns

Contains 13 candlestick pattern detectors. See Candlestick Pattern Recognition for details.

Categories:

  • Single Candle: Doji, Hammer, ShootingStar, Marubozu, SpinningTop
  • Two Candle: Engulfing, Harami, PiercingDarkCloud, Tweezer
  • Three Candle: Star, ThreeSoldiersCrows, ThreeInside
  • Composite: CandlestickPatternDetector (detects all patterns)

Wrappers

Provides wrapper types for composing indicators.

Exports:

  • Smoother: Generic wrapper that applies a moving average to any indicator's output
  • DAGWrapper: Wrapper for StatDAG integration with fit! infrastructure
using OnlineTechnicalIndicators.Wrappers

# Create a smoother (applies SMA to TrueRange output)
smoother = Smoother(TrueRange; period=14, ma=SMA)

for candle in ohlcv_data
    fit!(smoother, candle)
end

Factories

Factory functions for creating indicator instances.

Exports:

  • MovingAverage: Factory for creating typed moving average indicators
  • MAFactory: Deprecated alias for MovingAverage
using OnlineTechnicalIndicators.Factories

factory = MovingAverage(Float64)
ma = factory(SMA, period=10)  # Creates SMA{Float64}(period=10)

SampleData

Sample OHLCV data for testing and examples.

Exports:

  • OPEN_TMPL, HIGH_TMPL, LOW_TMPL, CLOSE_TMPL, VOLUME_TMPL: Price/volume vectors
  • V_OHLCV: Vector of OHLCV candlesticks
  • TAB_OHLCV: Tables.jl-compatible sample data
using OnlineTechnicalIndicators.SampleData

candles = SampleData.V_OHLCV
prices = SampleData.CLOSE_TMPL

External Dependencies

OnlineStatsBase.jl

Provides the foundational OnlineStat abstract type and core utilities:

  • fit!: Update a statistic with new data
  • value: Get the current value
  • nobs: Get the number of observations
  • CircBuff: Circular buffer for rolling windows

OnlineStatsChains.jl

Provides StatDAG for composing indicators into directed acyclic graphs with automatic value propagation. Used internally by composed indicators like DEMA, TEMA, T3, and TRIX.

Tables.jl

Enables integration with any Tables.jl-compatible data source (DataFrames, CSV, etc.).

See Also