Implementation:TA Lib Ta lib python OHLC Input Pattern

Knowledge Sources	ta-lib-python
Domains	Data_Preparation, Technical_Analysis
Last Updated	2026-02-09 22:00 GMT

Overview

Concrete pattern for preparing four separate OHLC numpy arrays as input to candlestick pattern recognition functions.

Description

This is a Pattern Doc documenting the data preparation pattern for CDL functions. Each CDL function requires four positional float64 numpy arrays in the order: open, high, low, close. The Cython layer validates array lengths match via check_length4 and converts to float64 via check_array.

Usage

Prepare four equal-length float64 arrays before calling any CDL pattern function.

Code Reference

Source Location

• Repository: ta-lib-python
• File: tests/conftest.py (ford_2012 fixture provides canonical OHLCV test data)

Signature

# OHLC data preparation patterns
import numpy as np

# Pattern 1: From separate arrays
open_prices = np.array([44.0, 44.3, 44.1], dtype=float)
high_prices = np.array([44.2, 44.5, 44.4], dtype=float)
low_prices  = np.array([43.9, 44.0, 43.8], dtype=float)
close_prices = np.array([44.1, 44.3, 43.9], dtype=float)

# Pattern 2: From pandas DataFrame
import pandas as pd
df = pd.read_csv('prices.csv')
open_prices = df['open']    # pd.Series, auto-converted by _wrapper
high_prices = df['high']
low_prices  = df['low']
close_prices = df['close']

Import

import numpy as np
import talib

I/O Contract

Inputs

Name	Type	Required	Description
open	np.ndarray (float64)	Yes	Opening prices
high	np.ndarray (float64)	Yes	High prices
low	np.ndarray (float64)	Yes	Low prices
close	np.ndarray (float64)	Yes	Closing prices

Outputs

Name	Type	Description
prepared_arrays	4x np.ndarray (float64)	Four validated, equal-length float64 arrays

Usage Examples

Prepare OHLC Data

import numpy as np
import talib

# Generate sample OHLC data
n = 100
close = np.cumsum(np.random.randn(n) * 0.5) + 100
open_p = close + np.random.randn(n) * 0.2
high = np.maximum(open_p, close) + np.abs(np.random.randn(n) * 0.3)
low = np.minimum(open_p, close) - np.abs(np.random.randn(n) * 0.3)

# Pass to any CDL function
doji = talib.CDLDOJI(open_p, high, low, close)

Related Pages

Implements Principle

• Principle:TA_Lib_Ta_lib_python_OHLC_Data_Preparation