utils.py

import pandas as pd
import numpy as np


### Initial Operations
def returns(df):
    """
    close-to-close returns
    """
    return df.close / df.close.shift(1) - 1

def vwap(df):
    """
    volume-weighted average price 
    """
    return (df.volume * df.close) / df.volume 

def adv(df, d):
    """
    adv{d} = average daily dollar volume for the past d days 
    """
    return df.volume.rolling(d).mean()

###
def rank(df):
    """
    Cross-sectional percentile rank.

    :param df:
    :return: 
    """
    return df.rank(pct=True)


def stddev(df, d):
    """
    Rolling standard deviation over the last d days.

    :param df:
    :param d:
    :return:
    """
    return df.rolling(d).std()


def scale(df, a=1):
    """
    rescaled x such that sum(abs(x)) = a (the default is a = 1) 
    :param df:
    :param a:
    :return:
    """
    return df.abs() / df.replace([np.inf, -np.inf], np.nan).abs().sum(skipna=True)

def product(df, d):
    """
    time-series product over the past d days
    :param df: data frame containing prices
    :param d: number of days to look back (rolling window)
    :return: Pandas series 
    """
    return df.rolling(d).apply(np.product)

def decay_linear(df, d):
    """
    weighted moving average over the past d days with linearly decaying
    weights d, d – 1, …, 1 (rescaled to sum up to 1)
    :param df: data frame containing prices
    :param d: number of days to look back (rolling window)
    :return: Pandas series 
    """
    return df.ewm(d).mean()


def delta(df, d):
    """
    today’s value of x minus the value of x d days ago
    """
    return df - df.shift(d)


def corr(x, y, d):
    """
    time-serial correlation of x and y for the past d days 
    """
    return x.rolling(d).corr(y)


def cov(x, y, d):
    """
    time-serial covariance of x and y for the past d days 
    """
    return x.rolling(d).cov(y)

def delay(df, d):
    """
    value of x d days ago
    """
    return df.shift(d)

### Time-Series Operations
def ts_max(df, d=10):
    """
    The rolling max over the last d days. 

    :param df: data frame containing prices
    :param d: number of days to look back (rolling window)
    :return: Pandas series
    """
    return df.rolling(d).max()


def ts_min(df, d=10):
    """
    The rolling min over the last d days. 

    :param df: data frame containing prices
    :param d: number of days to look back (rolling window)
    :return: Pandas series
    """
    return df.rolling(d).min()


def ts_argmax(df, d):
    """
    Gets the day, ts_max(x, d) occured on.

    :param df: dataframe
    :param d: number of days to look back (rolling window)
    :return: Pandas Series
    """
    return df.rolling(d).apply(np.argmax).add(1)


def ts_argmin(df, d):
    """
    Gets the day, ts_min(x, d) occured on.

    :param df: dataframe
    :param d: number of days to look back (rolling window)
    :return: Pandas Series
    """
    return df.rolling(d).apply(np.argmin).add(1)


def ts_rank(df, d):
    """
    time-series rank in the past d days
    
    :param df: dataframe
    :param d: number of days to look back (rolling window)
    :return: Pandas Series
    """
    return df.rolling(d).apply(lambda x: pd.Series(x).rank(pct=True).iloc[-1])

def ts_sum(df, d):
    """
    time-series sum over the past d days
    """
    return df.rolling(d).sum()