pytafast 0.4.0

Python wrapper for ta-lib using nanobind

pytafast

A high-performance Python wrapper for TA-Lib built with nanobind. Provides 150+ technical analysis functions with pandas/numpy support and async capabilities.

Features

• 🚀 High Performance — C++ bindings via nanobind with GIL release for true parallelism
• 📊 Full TA-Lib Coverage — 150+ indicators including overlaps, momentum, volatility, volume, statistics, cycle indicators, and 61 candlestick patterns
• 🐼 Pandas Native — Seamless support for both numpy.ndarray and pandas.Series (preserves index)
• ⚡ Async Support — All functions available as async via pytafast.aio
• 🔒 GSL Powered Safety — Uses Microsoft GSL (gsl::not_null, gsl::span) to prevent buffer overflows and null pointer dereferences in C++
• 📦 Drop-in Replacement — Same API as ta-lib-python, easy migration

Installation

pip install pytafast

Build from Source

git clone --recursive https://github.com/twn39/pytafast.git
cd pytafast
pip install -v -e .

Quick Start

Basic Usage with NumPy

import numpy as np
import pytafast

# Generate sample price data
np.random.seed(42)
close = np.cumsum(np.random.randn(200)) + 100

# Moving Averages
sma = pytafast.SMA(close, timeperiod=20)
ema = pytafast.EMA(close, timeperiod=12)
wma = pytafast.WMA(close, timeperiod=10)

Momentum Indicators

# RSI — Relative Strength Index
rsi = pytafast.RSI(close, timeperiod=14)

# MACD — returns 3 arrays: macd line, signal line, histogram
macd, signal, hist = pytafast.MACD(close, fastperiod=12, slowperiod=26, signalperiod=9)

# Stochastic Oscillator — requires high, low, close
high = close + np.abs(np.random.randn(200)) * 2
low = close - np.abs(np.random.randn(200)) * 2
slowk, slowd = pytafast.STOCH(high, low, close, fastk_period=5, slowk_period=3, slowd_period=3)

# ADX — Average Directional Index
adx = pytafast.ADX(high, low, close, timeperiod=14)

Bollinger Bands & Volatility

# Bollinger Bands — returns upper, middle, lower bands
upper, middle, lower = pytafast.BBANDS(close, timeperiod=20, nbdevup=2.0, nbdevdn=2.0)

# ATR — Average True Range
atr = pytafast.ATR(high, low, close, timeperiod=14)

Volume Indicators

volume = np.random.random(200) * 1_000_000

# On-Balance Volume
obv = pytafast.OBV(close, volume)

# Chaikin A/D Line
ad = pytafast.AD(high, low, close, volume)

# Money Flow Index
mfi = pytafast.MFI(high, low, close, volume, timeperiod=14)

Statistics & Math

# Linear Regression
linreg = pytafast.LINEARREG(close, timeperiod=14)
slope = pytafast.LINEARREG_SLOPE(close, timeperiod=14)

# Correlation between two series
beta = pytafast.BETA(close, ema, timeperiod=5)
correl = pytafast.CORREL(close, ema, timeperiod=30)

# Standard Deviation
stddev = pytafast.STDDEV(close, timeperiod=20, nbdev=1.0)

Candlestick Pattern Recognition

# Generate OHLC data
open_ = close + np.random.randn(200) * 0.5

# Detect patterns — returns integer array (100=bullish, -100=bearish, 0=none)
engulfing = pytafast.CDLENGULFING(open_, high, low, close)
doji = pytafast.CDLDOJI(open_, high, low, close)
hammer = pytafast.CDLHAMMER(open_, high, low, close)
morning_star = pytafast.CDLMORNINGSTAR(open_, high, low, close, penetration=0.3)

# Find bullish signals
bullish_idx = np.where(engulfing == 100)[0]

Pandas Support

import pandas as pd

# Create a DataFrame of stock prices
df = pd.DataFrame({
    "open": open_,
    "high": high,
    "low": low,
    "close": close,
    "volume": volume,
}, index=pd.date_range("2024-01-01", periods=200, freq="D"))

# All functions accept and return pd.Series, preserving the DatetimeIndex
df["sma_20"] = pytafast.SMA(df["close"], timeperiod=20)
df["rsi_14"] = pytafast.RSI(df["close"], timeperiod=14)
df["atr_14"] = pytafast.ATR(df["high"], df["low"], df["close"], timeperiod=14)

upper, middle, lower = pytafast.BBANDS(df["close"], timeperiod=20)
df["bb_upper"] = upper
df["bb_lower"] = lower

Async Support

import asyncio
import pytafast

async def compute_indicators(close, high, low, volume):
    """Compute multiple indicators concurrently."""
    sma, rsi, macd_result, atr = await asyncio.gather(
        pytafast.aio.SMA(close, timeperiod=20),
        pytafast.aio.RSI(close, timeperiod=14),
        pytafast.aio.MACD(close),
        pytafast.aio.ATR(high, low, close, timeperiod=14),
    )
    macd, signal, hist = macd_result
    return sma, rsi, macd, atr

# asyncio.run(compute_indicators(close, high, low, volume))

Cycle Indicators

# Hilbert Transform indicators for cycle analysis
ht_period = pytafast.HT_DCPERIOD(close)
ht_trendline = pytafast.HT_TRENDLINE(close)
sine, leadsine = pytafast.HT_SINE(close)
trend_mode = pytafast.HT_TRENDMODE(close)  # 1 = trend, 0 = cycle

API Reference

All functions accept numpy.ndarray (float64) or pandas.Series. When a pandas.Series is provided, the function returns a pandas.Series with the same index.

Overlap Studies

Function	Signature	Description
SMA	SMA(real, timeperiod=30)	Simple Moving Average
EMA	EMA(real, timeperiod=30)	Exponential Moving Average
WMA	WMA(real, timeperiod=30)	Weighted Moving Average
DEMA	DEMA(real, timeperiod=30)	Double Exponential Moving Average
TEMA	TEMA(real, timeperiod=30)	Triple Exponential Moving Average
TRIMA	TRIMA(real, timeperiod=30)	Triangular Moving Average
KAMA	KAMA(real, timeperiod=30)	Kaufman Adaptive Moving Average
MAMA	MAMA(real, fastlimit=0.5, slowlimit=0.05)	MESA Adaptive Moving Average. Returns (mama, fama)
T3	T3(real, timeperiod=5, vfactor=0.7)	Triple Exponential Moving Average (T3)
BBANDS	BBANDS(real, timeperiod=5, nbdevup=2.0, nbdevdn=2.0, matype=0)	Bollinger Bands. Returns (upper, middle, lower)
MA	MA(real, timeperiod=30, matype=0)	Moving Average (generic)
SAR	SAR(high, low, acceleration=0.02, maximum=0.2)	Parabolic SAR
MIDPOINT	MIDPOINT(real, timeperiod=14)	Midpoint over period
MIDPRICE	MIDPRICE(high, low, timeperiod=14)	Midprice over period

Momentum Indicators

Function	Signature	Description
RSI	RSI(real, timeperiod=14)	Relative Strength Index
MACD	MACD(real, fastperiod=12, slowperiod=26, signalperiod=9)	Returns (macd, signal, hist)
MACDEXT	MACDEXT(real, fastperiod=12, fastmatype=0, slowperiod=26, slowmatype=0, signalperiod=9, signalmatype=0)	MACD with controllable MA types. Returns (macd, signal, hist)
MACDFIX	MACDFIX(real, signalperiod=9)	MACD Fix 12/26. Returns (macd, signal, hist)
STOCH	STOCH(high, low, close, fastk_period=5, slowk_period=3, slowk_matype=0, slowd_period=3, slowd_matype=0)	Returns (slowk, slowd)
STOCHF	STOCHF(high, low, close, fastk_period=5, fastd_period=3, fastd_matype=0)	Returns (fastk, fastd)
STOCHRSI	STOCHRSI(real, timeperiod=14, fastk_period=5, fastd_period=3, fastd_matype=0)	Returns (fastk, fastd)
ADX	ADX(high, low, close, timeperiod=14)	Average Directional Index
ADXR	ADXR(high, low, close, timeperiod=14)	Average Directional Index Rating
CCI	CCI(high, low, close, timeperiod=14)	Commodity Channel Index
MFI	MFI(high, low, close, volume, timeperiod=14)	Money Flow Index
ROC	ROC(real, timeperiod=10)	Rate of change : ((price/prevPrice)-1)*100
ROCP	ROCP(real, timeperiod=10)	Rate of change Percentage: (price-prevPrice)/prevPrice
ROCR	ROCR(real, timeperiod=10)	Rate of change ratio: (price/prevPrice)
ROCR100	ROCR100(real, timeperiod=10)	Rate of change ratio 100: (price/prevPrice)*100
MOM	MOM(real, timeperiod=10)	Momentum
WILLR	WILLR(high, low, close, timeperiod=14)	Williams' %R
CMO	CMO(real, timeperiod=14)	Chande Momentum Oscillator
DX	DX(high, low, close, timeperiod=14)	Directional Movement Index
MINUS_DI	MINUS_DI(high, low, close, timeperiod=14)	Minus Directional Indicator
MINUS_DM	MINUS_DM(high, low, timeperiod=14)	Minus Directional Movement
PLUS_DI	PLUS_DI(high, low, close, timeperiod=14)	Plus Directional Indicator
PLUS_DM	PLUS_DM(high, low, timeperiod=14)	Plus Directional Movement
APO	APO(real, fastperiod=12, slowperiod=26, matype=0)	Absolute Price Oscillator
PPO	PPO(real, fastperiod=12, slowperiod=26, matype=0)	Percentage Price Oscillator
AROON	AROON(high, low, timeperiod=14)	Returns (aroondown, aroonup)
AROONOSC	AROONOSC(high, low, timeperiod=14)	Aroon Oscillator
ULTOSC	ULTOSC(high, low, close, timeperiod1=7, timeperiod2=14, timeperiod3=28)	Ultimate Oscillator
BOP	BOP(open, high, low, close)	Balance Of Power
TRIX	TRIX(real, timeperiod=30)	1-day Rate-Of-Change (ROC) of a Triple Smooth EMA

Volatility Indicators

Function	Signature	Description
ATR	ATR(high, low, close, timeperiod=14)	Average True Range
NATR	NATR(high, low, close, timeperiod=14)	Normalized Average True Range
TRANGE	TRANGE(high, low, close)	True Range

Volume Indicators

Function	Signature	Description
OBV	OBV(real, volume)	On Balance Volume
AD	AD(high, low, close, volume)	Chaikin A/D Line
ADOSC	ADOSC(high, low, close, volume, fastperiod=3, slowperiod=10)	Chaikin A/D Oscillator

Price Transform

Function	Signature	Description
AVGPRICE	AVGPRICE(open, high, low, close)	Average Price
MEDPRICE	MEDPRICE(high, low)	Median Price
TYPPRICE	TYPPRICE(high, low, close)	Typical Price
WCLPRICE	WCLPRICE(high, low, close)	Weighted Close Price

Statistics

Function	Signature	Description
STDDEV	STDDEV(real, timeperiod=5, nbdev=1.0)	Standard Deviation
VAR	VAR(real, timeperiod=5, nbdev=1.0)	Variance
BETA	BETA(real0, real1, timeperiod=5)	Beta
CORREL	CORREL(real0, real1, timeperiod=30)	Pearson's Correlation Coefficient (r)
LINEARREG	LINEARREG(real, timeperiod=14)	Linear Regression
LINEARREG_ANGLE	LINEARREG_ANGLE(real, timeperiod=14)	Linear Regression Angle
LINEARREG_INTERCEPT	LINEARREG_INTERCEPT(real, timeperiod=14)	Linear Regression Intercept
LINEARREG_SLOPE	LINEARREG_SLOPE(real, timeperiod=14)	Linear Regression Slope
TSF	TSF(real, timeperiod=14)	Time Series Forecast
AVGDEV	AVGDEV(real, timeperiod=14)	Average Deviation
MAX	MAX(real, timeperiod=30)	Highest value over a specified period
MIN	MIN(real, timeperiod=30)	Lowest value over a specified period
SUM	SUM(real, timeperiod=30)	Summation
MINMAX	MINMAX(real, timeperiod=30)	Returns (min, max)
MINMAXINDEX	MINMAXINDEX(real, timeperiod=30)	Returns (minidx, maxidx)

Cycle Indicators

Function	Signature	Description
HT_DCPERIOD	HT_DCPERIOD(real)	Hilbert Transform - Dominant Cycle Period
HT_DCPHASE	HT_DCPHASE(real)	Hilbert Transform - Dominant Cycle Phase
HT_PHASOR	HT_PHASOR(real)	Hilbert Transform - Phasor Components. Returns (inphase, quadrature)
HT_SINE	HT_SINE(real)	Hilbert Transform - SineWave. Returns (sine, leadsine)
HT_TRENDLINE	HT_TRENDLINE(real)	Hilbert Transform - Instantaneous Trendline
HT_TRENDMODE	HT_TRENDMODE(real)	Hilbert Transform - Trend vs Cycle Mode

Math Operators

Function	Signature	Description
ADD	ADD(real0, real1)	Vector Arithmetic Addition
SUB	SUB(real0, real1)	Vector Arithmetic Subtraction
MULT	MULT(real0, real1)	Vector Arithmetic Multiplication
DIV	DIV(real0, real1)	Vector Arithmetic Division

Math Transforms

All take a single real array: ACOS, ASIN, ATAN, CEIL, COS, COSH, EXP, FLOOR, LN, LOG10, SIN, SINH, SQRT, TAN, TANH

Candlestick Patterns

All candlestick functions accept (open, high, low, close) and return an integer array (100, -100, or 0).

Standard patterns: CDL2CROWS, CDL3BLACKCROWS, CDL3INSIDE, CDL3LINESTRIKE, CDL3OUTSIDE, CDL3STARSINSOUTH, CDL3WHITESOLDIERS, CDLADVANCEBLOCK, CDLBELTHOLD, CDLBREAKAWAY, CDLCLOSINGMARUBOZU, CDLCONCEALBABYSWALL, CDLCOUNTERATTACK, CDLDOJI, CDLDOJISTAR, CDLDRAGONFLYDOJI, CDLENGULFING, CDLGAPSIDESIDEWHITE, CDLGRAVESTONEDOJI, CDLHAMMER, CDLHANGINGMAN, CDLHARAMI, CDLHARAMICROSS, CDLHIGHWAVE, CDLHIKKAKE, CDLHIKKAKEMOD, CDLHOMINGPIGEON, CDLIDENTICAL3CROWS, CDLINNECK, CDLINVERTEDHAMMER, CDLKICKING, CDLKICKINGBYLENGTH, CDLLADDERBOTTOM, CDLLONGLEGGEDDOJI, CDLLONGLINE, CDLMARUBOZU, CDLMATCHINGLOW, CDLONNECK, CDLPIERCING, CDLRICKSHAWMAN, CDLRISEFALL3METHODS, CDLSEPARATINGLINES, CDLSHOOTINGSTAR, CDLSHORTLINE, CDLSPINNINGTOP, CDLSTALLEDPATTERN, CDLSTICKSANDWICH, CDLTAKURI, CDLTASUKIGAP, CDLTHRUSTING, CDLTRISTAR, CDLUNIQUE3RIVER, CDLUPSIDEGAP2CROWS, CDLXSIDEGAP3METHODS

Patterns with optional penetration parameter: CDLABANDONEDBABY(..., penetration=0.3), CDLDARKCLOUDCOVER(..., penetration=0.5), CDLEVENINGDOJISTAR(..., penetration=0.3), CDLEVENINGSTAR(..., penetration=0.3), CDLMATHOLD(..., penetration=0.5), CDLMORNINGDOJISTAR(..., penetration=0.3), CDLMORNINGSTAR(..., penetration=0.3)

Performance

pytafast achieves superior performance compared to the official ta-lib-python in Pandas, multi-threaded, and complex calculation scenarios. Benchmarked on 100,000 data points:

Multi-threading Throughput

Thanks to efficient GIL release in C++, pytafast scales near-linearly with CPU cores. Benchmark results in a 4-thread concurrent environment:

Benchmark (4 Threads)	Official Wrapper	pytafast	Throughput Increase
SMA Concurrency	20.1 ms	6.3 ms	3.15x
RSI Concurrency	56.5 ms	16.3 ms	3.46x
MACD Concurrency	75.0 ms	20.7 ms	3.62x

Single-call Efficiency

Category	Representative Indicators	Avg Speedup
Pandas Integration	SMA, EMA, MOM, RSI (Series input)	1.05x - 1.30x
Complex Indicators	ADX, ADXR, ULTOSC, MAMA	1.08x - 1.12x
Overlap Studies	SMA, EMA, BBANDS, KAMA, T3	1.02x
Momentum	RSI, MACD, STOCH, CCI, ROC	1.03x
Statistics & Math	LINEARREG, STDDEV, BETA, ADD	1.04x
Overall Consistency	All 150+ functions	100% Match

Key Highlights:

• MOM (Pandas): pytafast is 30% faster than official wrapper.
• ADX (NumPy): pytafast is 11% faster due to lower binding overhead.
• Safety: Integrated Microsoft GSL ensures robust memory management for large-scale financial data.

Recommendation: Use pytafast for multi-threaded backtesting or high-concurrency web servers to unlock full CPU potential.

API Compatibility

pytafast follows the same function signatures as the official TA-Lib Python wrapper, making it a drop-in replacement in most cases.

License

This project is licensed under the MIT License - see the LICENSE file for details.

This project includes and statically links to the TA-Lib C library, which is distributed under a BSD License. Copyright (c) 1999-2025, Mario Fortier. See third_party/ta-lib/LICENSE for details.