import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.dates as mdates

ts1 = pd.read_csv(r"C:\Users\Joseph\Desktop\TS_1.txt", parse_dates=["Date"])
ts2 = pd.read_csv(r"C:\Users\Joseph\Desktop\TS_2.txt", parse_dates=["Date"])

ts_merged = ts1.merge(ts2, on='Date', how='inner')

# Evitiamo valori inferiori a zero, altrimenti np.log() non funziona
ts_merged['Close_x'] = ts_merged['Close_x'] + np.abs(ts_merged['Close_x'].min()) + 1
ts_merged['Close_y'] = ts_merged['Close_y'] + np.abs(ts_merged['Close_y'].min()) + 1

# Normaliziamo i daily returns
ts_merged['ts1_returns'] = np.log(ts_merged['Close_x']).diff()
ts_merged['ts2_returns'] = np.log(ts_merged['Close_y']).diff()

print(ts_merged[['ts1_returns', 'ts2_returns']].describe())

ts_merged['60d rolling correlation'] = ts_merged['ts1_returns'].rolling(60).corr(ts_merged['ts2_returns'])

fig, ax = plt.subplots()
fig.autofmt_xdate()
plt.plot(ts_merged['Date'], ts_merged['60d rolling correlation'])
plt.grid(True)
plt.legend()
xfmt = mdates.DateFormatter('%d-%m-%y')
ax.xaxis.set_major_formatter(xfmt)
plt.show()