Preface

This is the python code appendix (2/2) of the research paper "Social interactions in volatile markets - a GameStop story".

Gather GameStop price data

# pip install alpha_vantage

Connect to API and download monthly data

from alpha_vantage.timeseries import TimeSeries
import pandas as pd

apiKey = '4OKA4AHOIV34TCGW'

ts = TimeSeries(key = apiKey, output_format = 'csv')

# Standard API call frequency is 5 calls per minute and 500 calls per day
# we need data from Sept 2020-11-30 - 2021-09-30
x = 11
# for x in range(1,5):
print(x)
totalData = ts.get_intraday_extended(symbol = 'GME', interval = '60min', slice = 'year1month'+str(x))


df = pd.DataFrame(list(totalData[0]))

#setup of column and index
header_row=0
df.columns = df.iloc[header_row]
df = df.drop(header_row)
df.set_index('time', inplace=True)

#show output
print(df)

df.to_csv('stockdata'+str(x)+'.csv')

11
0                      open   high    low  close    volume
time                                                      
2021-01-22 20:00:00    63.6   63.6   61.0   61.0    279826
2021-01-22 19:00:00    64.0  64.88  62.35   63.7    152240
2021-01-22 18:00:00    67.5  67.89  63.01   63.8    942544
2021-01-22 17:00:00    65.0  68.46   62.4   67.6   2300243
2021-01-22 16:00:00  59.185   68.0   57.2  64.85  30044433
...                     ...    ...    ...    ...       ...
2020-12-28 09:00:00    22.1  22.57  21.26  21.44    158480
2020-12-28 08:00:00    22.5   22.5  21.59   22.3     65176
2020-12-28 07:00:00   22.48  22.53   22.2   22.4      5334
2020-12-28 06:00:00    22.6  22.88  22.06  22.52      8735
2020-12-28 05:00:00   20.25   23.5   20.2  23.29     17963

[278 rows x 5 columns]

Merge monthly data to single data frame

Alpha Vantage gives out one month, so combining the stock data results in stockdatafull.csv which can be imported and adapted afterwards

import os
import glob
import pandas as pd

extension = 'csv'
all_filenames = [i for i in glob.glob('*.{}'.format(extension))]

#combine all files in the list
combined_csv = pd.concat([pd.read_csv(f) for f in all_filenames ])
#export to csv
combined_csv.to_csv( "stockdatafull.csv", index=False, encoding='utf-8-sig')

df = pd.read_csv('stockdatafull.csv', delimiter=',')
df = df.drop(1)
df['time'] = pd.to_datetime(df['time'], format='%Y-%m-%d %H:%M:%S')
df.set_index('time', inplace=True)
df.to_csv('stockdatafull.csv', header=True, index=True,)
df

	open	high	low	close	volume
time
2021-11-19 20:00:00	229.0000	229.00	228.52	229.00	2269
2021-11-19 18:00:00	228.5000	230.00	228.50	230.00	9977
2021-11-19 17:00:00	228.8000	230.48	228.05	228.67	37861
2021-11-19 16:00:00	224.5200	228.80	223.62	228.80	622632
2021-11-19 15:00:00	222.4599	229.39	220.48	224.43	916181
...	...	...	...	...	...
2021-02-23 09:00:00	43.8900	45.55	43.80	44.37	120850
2021-02-23 08:00:00	44.5200	45.24	43.82	43.84	48033
2021-02-23 07:00:00	43.8000	44.40	42.87	44.40	27430
2021-02-23 06:00:00	46.1400	46.15	43.80	43.80	30476
2021-02-23 05:00:00	46.0000	47.70	45.80	46.15	10720

4111 rows × 5 columns

Fit stock data to UTC

# Add 5 hours from time (EST), since UTC timezone is needed for combining pricing data with Reddit data
from datetime import timedelta
import pandas as pd
import pytz


df = pd.read_csv('stockdatafull.csv', delimiter=',')
df

df = df.drop(1)
df['time'] = pd.to_datetime(df['time'], format='%Y-%m-%d %H:%M:%S')
df.set_index('time', inplace=True)

eastern = pytz.timezone('US/Eastern')
df.index = df.index.tz_localize('US/Eastern').tz_convert('UTC')

# df.index = df.index.tz_localize('EST')
# df.index = df.index.tz_convert('Europe/Vienna')

# df['time'] = pd.to_datetime(df['time']) + timedelta(hours=5)
df.to_csv('stockdata_UTC.csv')
df

	open	high	low	close	volume
time
2021-11-20 01:00:00+00:00	229.0000	229.00	228.5200	229.000	2269
2021-11-19 22:00:00+00:00	228.8000	230.48	228.0500	228.670	37861
2021-11-19 21:00:00+00:00	224.5200	228.80	223.6200	228.800	622632
2021-11-19 20:00:00+00:00	222.4599	229.39	220.4800	224.430	916181
2021-11-19 19:00:00+00:00	217.4500	222.99	217.1601	222.565	411816
...	...	...	...	...	...
2021-02-23 14:00:00+00:00	43.8900	45.55	43.8000	44.370	120850
2021-02-23 13:00:00+00:00	44.5200	45.24	43.8200	43.840	48033
2021-02-23 12:00:00+00:00	43.8000	44.40	42.8700	44.400	27430
2021-02-23 11:00:00+00:00	46.1400	46.15	43.8000	43.800	30476
2021-02-23 10:00:00+00:00	46.0000	47.70	45.8000	46.150	10720

4110 rows × 5 columns

---

Plot graphs

import matplotlib.pyplot as plt
import datetime as dt
import pandas as pd
from datetime import datetime
import matplotlib.dates as mdates

# df = pd.read_csv('stockdata_UTC.csv', delimiter=',' , parse_dates=['time'], index_col=['time'])
# # convert the time column as datetime64[ns] instead of "object" variable
# df['time'] = pd.to_datetime(df['time'], format='%Y-%m-%d %H:%M:%S')
# pd.DatetimeIndex(df['time'])

# load and format the data
df = pd.read_csv('stockdata_UTC.csv')

df['time'] = pd.to_datetime(df['time'], format = "%Y-%m-%d %H:%M:%S").dt.strftime('%Y-%m-%d %H:%M:%S')
df = df.drop_duplicates(subset="time")
df = df.sort_values(by = 'time')
df['returns'] = df['close'].pct_change()
df = df.dropna()
df.set_index('time', inplace=True)


# df = df[~df.index.duplicated(keep='first')]

# plot
fig = plt.figure(figsize=(15, 7))
top = plt.subplot2grid((4,4), (0, 0), rowspan=3, colspan=4)
top.plot(df['close'], label='close')
plt.title('GME closing price')
plt.xlabel('Date')
plt.ylabel('Price in USD')
plt.gca().xaxis.set_major_locator(mdates.DayLocator(interval=30)) 
plt.gca().xaxis.set_tick_params(rotation = 45)
plt.show()


fig = plt.figure()
ax1 = fig.add_axes([1,1,0.8,0.8])
ax1.plot(df.returns)
ax1.set_xlabel("Date")
ax1.set_ylabel("Percent")
ax1.set_title("GME price")
ax1.xaxis.set_tick_params(rotation = 45)
ax1.xaxis.set_major_locator(mdates.DayLocator(interval=30)) 
plt.show()

df.to_csv('stockdata_UTC_final.csv')