wael-lib 1.2.0

Technical Analysis, Data Handling & Sequence Utilities for Trading Applications

wael-lib

Technical Analysis, Data Handling & Sequence Utilities for Trading Applications

---

Installation

pip install wael-lib

Quick Start

import pandas as pd
from wael_lib import TechAnalysis, WDataHandler, SequenceCreator, SequenceScaler

ta = TechAnalysis()
dh = WDataHandler()

df = pd.read_csv("OHLCV.csv")

# Indicators
df["SMA20"]  = ta.sma(df["Close"], 20)
df["RSI14"]  = ta.rsi(df["Close"], 14)
macd, signal, hist = ta.macd(df["Close"])
df["PriceRange"] = ta.categorize_price_ranges(df["Close"])

# Clean outliers
df["Volume"] = dh.handle_outliers(df, "Volume")

# Build LSTM sequences
seq = SequenceCreator(window=30, horizon=1)
X_seq, y_seq = seq.create_sequences(df[["Close", "Volume"]], df["Target"])
scaler = SequenceScaler()
X_scaled = scaler.fit_transform(X_seq)

---

Classes

Class	Purpose
TechAnalysis	30+ indicators, trend/structure detection, candlestick patterns
WDataHandler	IQR outlier clipping, empirical CDF
SequenceCreator	Sliding-window sequences for LSTM / GRU
SequenceScaler	MinMaxScaler for 3-D tensors (sklearn-compatible)
MySequenceScaler	StandardScaler for 3-D tensors (sklearn-compatible)

---

API Reference

TechAnalysis

Moving Averages

Method	Signature	Description
sma	sma(source, length) → pd.Series	Simple Moving Average
ema	ema(source, length) → pd.Series	Exponential Moving Average
wma	wma(source, length) → pd.Series	Weighted Moving Average
tema	tema(source, length) → pd.Series	Triple Exponential Moving Average

Oscillators & Momentum

Method	Signature	Description
rsi	rsi(source, length) → pd.Series	Relative Strength Index (0–100)
stoch	stoch(high, low, close, k_length, d_length) → (pd.Series, pd.Series)	Stochastic Oscillator (%K, %D)
mfi	mfi(high, low, close, volume, length) → pd.Series	Money Flow Index (0–100)
adx	adx(high, low, close, length) → pd.Series	Average Directional Index
macd	macd(source, fast=12, slow=26, signal=9) → (line, signal, hist)	MACD (3 series)
tema_macd	tema_macd(source, fast=12, slow=24, signal=9) → (line, signal, hist)	TEMA-based MACD (reduced lag)
PriceExtOsc	PriceExtOsc(high, low, length) → pd.Series	Price Extension Oscillator (0–100)
accumulation	accumulation(df, length=14) → pd.Series	Accumulation/Distribution
acc_osc	acc_osc(df, length=14) → pd.Series	Accumulation Oscillator (0–100)
extremes	extremes(source, length=14) → pd.Series	Min-max normalisation (0–100)
categorize_price_ranges	categorize_price_ranges(source) → pd.Series	Categorize prices into descriptive ranges (Very Low, Low, Moderate, High, Very High)

Volatility

Method	Signature	Description
atr	atr(high, low, close, length) → pd.Series	Average True Range
stdev	stdev(source, length) → pd.Series	Rolling Standard Deviation

Utility Functions

Method	Signature	Description
highest	highest(source, length) → pd.Series	Rolling maximum
lowest	lowest(source, length) → pd.Series	Rolling minimum
change	change(source, length) → pd.Series	Period-over-period difference
valuewhen	valuewhen(condition, value, occurrence) → pd.Series	Value at last condition=True
barssince	barssince(condition) → pd.Series	Bars since last condition=True
math_max	math_max(a, b) → pd.Series	Element-wise maximum
math_min	math_min(a, b) → pd.Series	Element-wise minimum
math_round	math_round(source) → pd.Series	Round to nearest integer
math_abs	math_abs(source) → pd.Series	Absolute value
math_sum	math_sum(source, length) → pd.Series	Rolling sum

Pivot Points

Method	Signature	Description
pivot_high	pivot_high(high, left, right, fill_pivots=True) → pd.Series	Detect swing highs
pivot_low	pivot_low(low, left, right, fill_pivots=True) → pd.Series	Detect swing lows

Trend & Structure Detection

Method	Signature	Description
rw_top	rw_top(data, curr_index, order) → bool	Local top check
rw_bottom	rw_bottom(data, curr_index, order) → bool	Local bottom check
rw_extremes	rw_extremes(data, order) → (tops, bottoms)	All local extremes
detect_trend	detect_trend(df, row_index, tops, bottoms) → str	Uptrend / Downtrend / Sideways
detect_bos_choch	detect_bos_choch(df, row_index, tops, bottoms) → str	Break of Structure or Change of Character
detect_formations	detect_formations(df, row_index, tops, bottoms) → str	Chart formations (H&S, Double Top, Triangles, …)
detect_candlestick_pattern	detect_candlestick_pattern(df, row_index) → str	12 candlestick patterns (Engulfing, Hammer, Doji, …)

---

WDataHandler

Method	Signature	Description
handle_outliers	handle_outliers(df, feature) → pd.Series	IQR-based outlier clipping
calculate_ECDF	calculate_ECDF(data, percent=False) → (sorted, ecdf)	Empirical CDF (survival form)

Example — Outlier Clipping:

dh = WDataHandler()
df["Volume"] = dh.handle_outliers(df, "Volume")

Example — ECDF:

sorted_vals, ecdf = dh.calculate_ECDF(df["Close"].values)

---

SequenceCreator

Create sliding-window sequences for LSTM / GRU models.

Method	Signature	Description
__init__	SequenceCreator(window=10, horizon=1)	Set window and horizon
create_sequences	create_sequences(X, y) → (X_seq, y_seq)	Input/target pairs (batch, timesteps, features)
create_X_sequences	create_X_sequences(X) → X_seq	Input-only sequences for inference

Example:

seq = SequenceCreator(window=30, horizon=1)
X_seq, y_seq = seq.create_sequences(df[["Close", "Volume"]], df["Target"])
# X_seq.shape → (n_samples - 30, 30, 2)

# For prediction on new data
X_pred = seq.create_X_sequences(new_data[["Close", "Volume"]])
predictions = model.predict(X_pred)

---

SequenceScaler

Scikit-learn compatible MinMaxScaler wrapper for 3-D tensors.

Method	Signature	Description
fit	fit(X, y=None) → self	Learn min/max from 3-D data
transform	transform(X) → np.ndarray	Scale to [0, 1]
fit_transform	fit_transform(X, y=None) → np.ndarray	Fit then transform

Example:

scaler = SequenceScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled  = scaler.transform(X_test)

---

MySequenceScaler

Same API as SequenceScaler but uses StandardScaler (zero-mean, unit-variance).

Example:

scaler = MySequenceScaler()
X_train_scaled = scaler.fit_transform(X_train)

---

Dependencies

Package	Version
pandas	>= 2.0.0
numpy	>= 1.24.0
scikit-learn	>= 1.3.0

License

MIT — see LICENSE for details.

Author

Wael Fouda — GitHub