def plot_candlestick() -> None:
"""Plot a candlestick chart using data from Quandl."""
import matplotlib.dates as mdates
from mpl_finance import candlestick_ohlc
quandl.read_key()
# Get data of ABN Amro
df = quandl.get('EURONEXT/ABN',
start_date='2018-07-01',
end_date='2018-07-31')
# Convert index to mpl format date and extract open, high, low, and close
df['Date'] = df.index.map(mdates.date2num)
df_ohlc = df[['Date', 'Open', 'High', 'Low', 'Last']]
print(STR_FMT.format('df_ohlc.head()', df_ohlc.head()))
print(STR_FMT.format('df_ohlc.tail()', df_ohlc.tail()))
# Plot
fig, ax = plt.subplots(figsize=(8, 4))
formatter = mdates.DateFormatter('%Y-%m-%d')
ax.xaxis.set_major_formatter(formatter)
candlestick_ohlc(ax,
df_ohlc.values,
width=0.8,
colorup='green',
colordown='red')
# Save figure
plt.subplots_adjust(hspace=0.75)
fig.savefig(IMGDIR + 'candlestick.png', bbox_inches='tight')
def ema() -> None:
"""Exponential moving average of a time series."""
quandl.read_key()
# Get data of ABN Amro
df = quandl.get('EURONEXT/ABN', column_index=4)
print(STR_FMT.format('df.head()', df.head()))
print(STR_FMT.format('df.tail()', df.tail()))
# Fill in missing values on a daily basis
df_filled = df.asfreq('D', method='ffill')
df_last = df_filled['Last']
# Calculate the EMA with a decay spanning 5-days and 30-days
# The ewm() method provides exponential weighted functions
help(df_last.ewm)
series_short = df_last.ewm(span=5).mean()
series_long = df_last.ewm(span=30).mean()
# Plot the long and short window EMAs
df_sma = pd.DataFrame(columns=['short', 'long'])
df_sma['short'] = series_short
df_sma['long'] = series_long
df_sma.plot(figsize=(12, 8))
plt.savefig(IMGDIR + 'ema.png', bbox_inches='tight')
def plot_time_series() -> None:
"""Plot a time series using data from Quandl."""
quandl.read_key()
# Get data of ABN Amro
df = quandl.get('EURONEXT/ABN')
print(STR_FMT.format('df.head()', df.head()))
print(STR_FMT.format('df.tail()', df.tail()))
df.plot()
plt.savefig(IMGDIR + 'dataset.png', bbox_inches='tight')
# Extract the daily closing price and volume
prices = df['Last']
volumes = df['Volume']
print(STR_FMT.format('prices.head()', prices.head()))
print(STR_FMT.format('volumes.tail()', volumes.tail()))
print(STR_FMT.format('type(volumes)', type(volumes)))
# Plot the prices and volumes
# Top plot consisting of daily closing price
top = plt.subplot2grid((4, 4), (0, 0), rowspan=3, colspan=4)
top.plot(prices.index, prices, label='Last')
plt.title('ABN Last Price from {low} - {high}'.format(
low=prices.index[0].year, high=prices.index[-1].year))
plt.legend(loc=2)
# The bottom plot consisting of daily trading volume
bottom = plt.subplot2grid((4, 4), (3, 0), rowspan=1, colspan=4)
bottom.bar(volumes.index, volumes)
plt.title('ABN Daily Trading Volume')
# Save figure
plt.gcf().set_size_inches(12, 8)
plt.subplots_adjust(hspace=0.75)
plt.savefig(IMGDIR + 'time_series.png', bbox_inches='tight')
def plot_correlation() -> None:
"""Visualise the correlation between two datasets."""
quandl.read_key()
# Get data for ABN Amro, Banco Santander, and Kas Bank
df = quandl.get(['EURONEXT/ABN', 'EURONEXT/SANTA'],
column_index=4,
start_date='2016-01-01',
end_date='2017-12-31')
print(STR_FMT.format('df.head()', df.head()))
print(STR_FMT.format('df.tail()', df.tail()))
print(STR_FMT.format('df.describe()', df.describe()))
# Compute the daily changes and window size for rolling
df_filled = df.asfreq('D', method='ffill')
daily_changes = df_filled.pct_change()
abn_returns = daily_changes['EURONEXT/ABN - Last']
santa_returns = daily_changes['EURONEXT/SANTA - Last']
window = len(df_filled.index) // 2
print(STR_FMT.format('window', window))
print(STR_FMT.format('abn_returns.describe()', abn_returns.describe()))
print(STR_FMT.format('santa_returns.describe()', santa_returns.describe()))
# Compute the correlation with a rolling window
df_corrs = abn_returns\
.rolling(window=window, min_periods=window)\
.corr(other=santa_returns)\
.dropna()
df_corrs.plot(figsize=(12, 8))
plt.savefig(IMGDIR + 'correlation.png', bbox_inches='tight')
def qq_plot() -> None:
"""
Create a Q-Q (quantile-quantile) plot.
Notes
----------
This is a graphical method for comparing two probability distributions by
plotting their quantiles against each other, for example to test if the
daily changes of a stock are distributed normally.
"""
from scipy import stats
quandl.read_key()
# Get the daily changes data
df = quandl.get('EURONEXT/ABN', column_index=4)
daily_changes = df.pct_change(periods=1).dropna()
print(STR_FMT.format('daily_changes.describe()', daily_changes.describe()))
# Create the Q-Q plot against a normal distribution
# Note that stats.probplot is the same as a Q-Q plot, however probabilities
# are shown in the scale of the theoretical distribution (x-axis) and the
# y-axis contains unscaled quantiles of the sample data.
fig, ax = plt.subplots(figsize=(8, 4))
help(stats.probplot)
stats.probplot(daily_changes['Last'], dist='norm', plot=ax)
fig.savefig(IMGDIR + 'qq_plot.png', bbox_inches='tight')
def time_series_analytics() -> None:
"""Visualise some statistical proprties of time series."""
quandl.read_key()
# Get data of ABN Amro
df = quandl.get('EURONEXT/ABN', column_index=4)
print(STR_FMT.format('df.head()', df.head()))
print(STR_FMT.format('df.tail()', df.tail()))
# Calculate and plot the percentage daily returns
daily_changes = df.pct_change(periods=1)
print(STR_FMT.format('daily_changes.describe()', daily_changes.describe()))
daily_changes.plot()
plt.savefig(IMGDIR + 'pct_change.png', bbox_inches='tight')
# Calculate and plot the cumulative returns
# Equivalent to "df / df['Last'][0] - 1"
df_cumprod = (daily_changes + 1).cumprod() - 1
df_cumprod.plot()
plt.savefig(IMGDIR + 'cum_return.png', bbox_inches='tight')
# Calculate and plot a histogram
daily_changes.hist(bins=50, figsize=(8, 4))
plt.savefig(IMGDIR + 'hist.png', bbox_inches='tight')
# Calculate and plot standard deviation / volaility over one month
df_filled = df.asfreq('D', method='ffill') # Pad missing entries
df_returns = df_filled.pct_change()
df_std = df_returns.rolling(window=30, min_periods=30).std()
df_std.plot()
plt.savefig(IMGDIR + 'volatility.png', bbox_inches='tight')
def __init__(self, path, table, date):
if not date:
import datetime
now = datetime.datetime.now()
date = {
'start': '2000-01-01',
'end': now.year + '-' + now.month + '-' + now.day
}
quandl.read_key(filename=path)
self.data = quandl.get(table,
start_date=date['start'],
end_date=date['end'])
def plot():
from pandas_datareader import data
import datetime
import dateutil.relativedelta
import quandl
from bokeh.plotting import figure, show, output_file
from bokeh.embed import components
from bokeh.resources import CDN
start = datetime.datetime.now() - dateutil.relativedelta.relativedelta(days=30)
end = datetime.datetime.now()
quandl.read_key()
infy_stock =data.DataReader('BSE/BOM500209', 'quandl',api_key=quandl.ApiConfig.api_key,
start = start, end = end)
infy_stock["Status"] = ["Increase" if c > o else "Decrease" if c < o else "Equal"
for c,o in zip(infy_stock.Close,infy_stock.Open)]
infy_stock["Midpoint"] = (infy_stock.Open + infy_stock.Close)/2
infy_stock["Height"] = abs(infy_stock.Open - infy_stock.Close)
plot_i = figure(x_axis_type = 'datetime', width = 800, height = 300, sizing_mode = "scale_width")
plot_i.title.text = "Candlestick Chart of Infosys"
plot_i.grid.grid_line_alpha = 0.5
hours_12 = 12*60*60*1000
plot_i.segment(infy_stock.index, infy_stock.High, infy_stock.index, infy_stock.Low, color="magenta")
plot_i.rect(infy_stock.index[infy_stock.Status == "Increase"], infy_stock.Midpoint[infy_stock.Status == "Increase"],
hours_12, infy_stock.Height[infy_stock.Status == "Increase"], fill_color="green", line_color="black")
plot_i.rect(infy_stock.index[infy_stock.Status == "Decrease"], infy_stock.Midpoint[infy_stock.Status == "Decrease"],
hours_12, infy_stock.Height[infy_stock.Status == "Decrease"], fill_color="red", line_color="black")
plot_i.rect(infy_stock.index[infy_stock.Status == "Equal"], infy_stock.Midpoint[infy_stock.Status == "Equal"],
hours_12, infy_stock.Height[infy_stock.Status == "Equal"], fill_color="blue", line_color="black")
script1, div1 = components(plot_i)
cdn_js = CDN.js_files
cdn_css = CDN.css_files
return render_template("plot.html", script1 = script1, div1 = div1, cdn_js = cdn_js, cdn_css = cdn_css)
def correlation() -> None:
"""Downloading multiple time series data and display their correlation."""
quandl.read_key()
# Get data for ABN Amro, Banco Santander, and Kas Bank
df = quandl.get(['EURONEXT/ABN', 'EURONEXT/SANTA', 'EURONEXT/KA'],
column_index=4,
collapse='monthly',
start_date='2016-01-01',
end_date='2017-12-31')
print(STR_FMT.format('df.head()', df.head()))
print(STR_FMT.format('df.tail()', df.tail()))
print(STR_FMT.format('df.describe()', df.describe()))
# Plot
df.plot()
plt.savefig(IMGDIR + 'multiple_data.png', bbox_inches='tight')
# Compute the correlation for the daily changes
corr = df.pct_change().corr(method='pearson')
print(STR_FMT.format('corr', corr))
def sma() -> None:
"""Simple moving average of a time series."""
quandl.read_key()
# Get data of ABN Amro
df = quandl.get('EURONEXT/ABN', column_index=4)
print(STR_FMT.format('df.head()', df.head()))
print(STR_FMT.format('df.tail()', df.tail()))
# Fill in missing values on a daily basis
df_filled = df.asfreq('D', method='ffill')
df_last = df_filled['Last']
# Calculate the SMA for a 5-day and 30-day window
series_short = df_last.rolling(window=5, min_periods=5).mean()
series_long = df_last.rolling(window=30, min_periods=30).mean()
# Plot the long and short window SMAs
df_sma = pd.DataFrame(columns=['short', 'long'])
df_sma['short'] = series_short
df_sma['long'] = series_long
df_sma.plot(figsize=(12, 8))
plt.savefig(IMGDIR + 'sma.png', bbox_inches='tight')
def get_data(tickers, start_date='2010-07-01', end_date='2019-03-01', lags=5):
quandl.read_key('../data/quandl.key') # personal key to quandl API
df = quandl.get_table('WIKI/PRICES',
ticker='AAPL',
qopts={'columns': ['date', 'adj_close']},
date={
'gte': start_date,
'lte': end_date
},
paginate=True)
df.date = pd.to_datetime(df.date)
df['today'] = df['adj_close']
df['lag_00'] = df['today'].pct_change() * 100
df.set_index(pd.DatetimeIndex(df['date']), inplace=True)
df.drop('date', axis=1, inplace=True)
for i in range(lags):
df['lag_{:02}'.format(i +
1)] = df['today'].shift(i + 1).pct_change() * 100
df['direction'] = np.where(df['lag_00'] >= 0, 1, 0)
df = df.dropna()
return df
get_stock_data(None)
except AssertionError:
test1 = True
try:
get_stock_data(55)
except AssertionError:
test2 = True
self.assertTrue(test1)
self.assertTrue(test2)
def test_getting_data(self):
"""Test that data is returned"""
gme = get_stock_data("GME")
self.assertFalse(gme.empty)
multiple = get_stock_data(["NOK", "GME", "AMC"])
self.assertFalse(multiple.empty)
erroneous = get_stock_data(";laksdj;flkasjd;a")
self.assertTrue(erroneous.empty)
if __name__ == '__main__':
# Set quandl api key
quandl.read_key(filename=API_FILE)
unittest.main()
import quandl
import os
import pandas as pd
import sys
#use this line the first time you use quandl. For more info see: https://github.com/quandl/quandl-python
# quandl.save_key("supersecret")
#this line afterwards
quandl.read_key()
cot_period = 26
dir = 'data\\'
continue_contract_month_base = 2
quandl_cot_futures_map = {
"ZN": "043602",
"ZS": "005602",
"ZM": "026603",
"ZL": "007601",
"ZC": "002602",
"ZW": "001602",
"HE": "054642",
"LE": "057642",
"GC": "088691",
"SI": "084691",
"HG": "085692",
"CL": "067651",
"HO": "022651",
"RB": "111659",
"NG": "023651",
Yahoo has significantly altered its data api, and pandas_datareader has deprecated many
of its readers, so I have elected to rewrite the data acquisition portion of this project
using the Quandl api instead.
'''
import datetime
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import quandl
from sklearn.preprocessing import Normalizer
from sklearn.pipeline import make_pipeline
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
quandl.read_key('../data/quandl.key') # personal key to quandl API
company_dict = {
'Amazon': 'AMZN',
'Apple': 'AAPL',
'Walgreen': 'WBA',
'Northrop Grumman': 'NOC',
'Boeing': 'BA',
'Lockheed Martin': 'LMT',
'McDonalds': 'MCD',
'Intel': 'INTC',
'Navistar': 'NAV',
'IBM': 'IBM',
'Texas Instruments': 'TXN',
'MasterCard': 'MA',
'Microsoft': 'MSFT',
def __init__(self, filename):
quandl.read_key(filename=filename)
self.data_stream = "EOD/"
