ich bin Python-Anfänger und habe mir eine Funktion geschrieben. Innerhalb dieser Funktion habe ich einen DataFrame names "Portfolio kreiert und würde diesen nun gerne außerhalb der Funktion nutzen (bspw. für Plots).
Mit return portfolio funktioniert das ganze nicht. Habt ihr eine andere Lösung?
Code: Alles auswählen
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
# Import a data source - FSE-Data with Index 'Date'
all_close_prices = pd.read_csv('FSE_daily_close.csv')
all_close_prices = all_close_prices.set_index('Date')
# Fill NaN Values with the last available stock price - except for Zalando
all_close_prices = all_close_prices.fillna(method='ffill')
# Import ticker symbols
ticker_list = list(all_close_prices)
# Zalando 'FSE/ZO1_X' (position row 99) - doesn't begin in 2004
# Drop Zalando
all_close_prices.drop('FSE/ZO1_X', axis=1)
# Also from the ticker list
ticker_list.remove('FSE/ZO1_X')
# Create an empty signal dataframe with datetime index equivalent to the stocks
signals = pd.DataFrame(index=all_close_prices.index)
def ma_strategy(ticker, long_window, short_window):
# Calculate the moving avergaes
moving_avg_long = all_close_prices.rolling(window=long_window, min_periods=1).mean()
moving_avg_short = all_close_prices.rolling(window=short_window, min_periods=1).mean()
moving_avg_short = moving_avg_short
moving_avg_long = moving_avg_long
# Add the two MAs for the stocks in the ticker_list to the signals dataframe
for i in ticker_list:
signals['moving_avg_short_' + i] = moving_avg_short[i]
signals['moving_avg_long_' + i] = moving_avg_long[i]
# Set up the signals
for i in ticker_list:
signals['signal_' + i] = np.where(signals['moving_avg_short_' + i] > signals['moving_avg_long_' + i], 1, 0)
signals['positions_' + i] = signals['signal_' + i].diff(periods=1)
#Backtest
initial_capital = float(100000)
# Create a DataFrame `positions` with index of signals
positions = pd.DataFrame(index=all_close_prices)
# Create a new column in the positions DataFrame
# On the days that the signal is 1 (short moving average crosses the long moving average, you’ll buy a 100 shares.
# The days on which the signal is 0, the final result will be 0 as a result of the operation 100*signals['signal']
positions = 100 * signals[['signal_' + ticker]]
# Store the portfolio value owned with the stock
# DataFrame.multiply(other, axis='columns', fill_value=None) - Multiplication of dataframe and other, element-wise
# Store the difference in shares owned - same like position column in signals
pos_diff = positions.diff()
# Add `holdings` to portfolio
portfolio = pd.DataFrame(index=all_close_prices.index)
portfolio['holdings'] = (positions.multiply(all_close_prices[ticker], axis=0)).sum(axis=1)
# Add `cash` to portfolio
portfolio['cash'] = initial_capital - (pos_diff.multiply(all_close_prices[ticker], axis=0)).sum(
axis=1).cumsum()
# Add `total` to portfolio
portfolio['total'] = portfolio['cash'] + portfolio['holdings']
# Add `returns` to portfolio
portfolio['return'] = portfolio['total'].pct_change()
portfolio['return_cum'] = portfolio['total'].pct_change().cumsum()
return portfolio
# Visualize the total value of the portfolio
portfolio_value = plt.figure(figsize=(12, 8))
ax1 = portfolio_value.add_subplot(1, 1, 1, ylabel='Portfolio value in $')
# Plot the equity curve in dollars
portfolio['total'].plot(ax=ax1, lw=2.)