strategy.py

import pandas as pd
import numpy as np
import math
import compute


# Create column Stance
# Create column Accumulated Close

def SMA_Crossover(df, fast=42, slow=252):

    # create the moving average values and
    # simultaneously append them to new columns in our existing DataFrame.
    df['42d'] = np.round(compute.running_average(df['Adj Close'], windowsize=42), 2)
    df['252d'] = np.round(compute.running_average(df['Adj Close'], windowsize=252), 2)

    # Generate stance: 0, -1, 1.
    df['42-252'] = df['42d'] - df['252d']

    offset = 0
    #add condition to ensure first stance change is never -1.
    df['Stance'] = np.where(df['42-252'] > offset, 1, 0) #42ma being offset amount above 252 value.
    df['Stance'] = np.where(df['42-252'] < offset, -1, df['Stance'])  #offset amount below 252 value
    # print df['Stance'].value_counts()

    # Generate accumulated closing price
    compute.accumulated_close(df)  #instead of everything below:

    # df_ma['Market Returns'] = np.log(df_ma['Adj Close'] / df_ma['Adj Close'].shift(1))
    # df_ma['Strategy'] = df_ma['Market Returns'] * df_ma['Stance'].shift(1)


def RSI(df, lowerCutoff=30, upperCutoff=70, period=14):

    df['RSI'] = pd.Series(compute.RSI(df['Adj Close'], period=period))
    df['Stance'] = 0
    offset = 0
    last_stance = 0
    upper_cutoff_set_once = False

    for index, row in df.iterrows():
        if row['RSI'] > upperCutoff:
            upper_cutoff_set_once = True
            last_stance = 1
        elif upper_cutoff_set_once and row['RSI'] < lowerCutoff:
            last_stance = -1

        df.set_value(index, 'Stance', last_stance)

    compute.accumulated_close(df)


def Bollinger_Band(df):

    # 1. Compute rolling mean
    rolling_mean = compute.rolling_std_mean(df['Adj Close'], window=20)

    # 2. Compute rolling standard deviation
    rolling_std = compute.rolling_std(df['Adj Close'], window=20)

    # 3. Compute upper and lower bands
    upper_band, lower_band = compute.bollinger_bands(rolling_mean, rolling_std)

    df['Stance'] = 0
    last_stance = 0
    upper_cutoff_set_once = False

    upper_band.fillna(method='backfill', inplace=True)
    lower_band.fillna(method='backfill', inplace=True)

    df['Upper Band'] = upper_band
    df['Lower Band'] = lower_band

    for index, row in df.iterrows():

        if row['Adj Close'] > row['Upper Band']:
            upper_cutoff_set_once = True
            last_stance = 1
        elif upper_cutoff_set_once and row['Adj Close'] < row['Lower Band']:
            last_stance = -1

        df.set_value(index, 'Stance', last_stance)

    compute.accumulated_close(df)


def MACD(df):
    print 'Pending.'