def granger_casualties(banks, full_period=True, date_start="", date_end=""):
banks_name = []
#Creo la matrice delle correlazioni
for x in banks:
banks_name.append(x.ticker)
corr_matrix = pd.DataFrame(index=banks_name, columns=banks_name)
#Senza thread
for x in corr_matrix.index:
for y in corr_matrix.columns:
bank_a = bank.find_ticker_in_list(x,banks)
bank_b = bank.find_ticker_in_list(y,banks)
corr_matrix.loc[x,y] = correlation_a_granger_caused_by_b(bank_a,bank_b,full_period=full_period, date_start=date_start, date_end=date_end)
'''
#Thread version
def correlation_for_thread(x, y, banks, full_period, date_start, date_end):
bank_a = bank.find_ticker_in_list(x, banks)
bank_b = bank.find_ticker_in_list(y, banks)
corr_matrix.loc[x, y] = correlation_a_granger_caused_by_b(bank_a, bank_b, full_period=full_period,
date_start=date_start, date_end=date_end)
thread_list = []
for x in corr_matrix.index:
for y in corr_matrix.columns:
args = [x, y, banks, full_period, date_start, date_end]
t = Thread(target=correlation_for_thread, args=args)
thread_list.append(t)
t.start()
for thread in thread_list:
thread.join()
'''
return corr_matrix
def granger_casualties(banks, full_period=True, date_start="", date_end=""):
banks_name = []
#Creo la matrice delle correlazioni
for x in banks:
banks_name.append(x.ticker)
corr_matrix = pd.DataFrame(index=banks_name, columns=banks_name)
for x in corr_matrix.index:
for y in corr_matrix.columns:
bank_a = bank.find_ticker_in_list(x, banks)
bank_b = bank.find_ticker_in_list(y, banks)
corr_matrix.loc[x, y] = correlation_a_granger_caused_by_b(
bank_a,
bank_b,
full_period=full_period,
date_start=date_start,
date_end=date_end)
return corr_matrix
def portfolio_covar(covar_matrix, banks, year, quarter):
covar = covar_matrix.copy()
covar['Weighted'] = 0
mva_sum = 0
for ticker in covar['Ticker']:
bank = find_ticker_in_list(ticker,banks)
mask = (bank.mva['Year'] == year ) & ( bank.mva['Quarter'] == quarter)
mva = float(bank.mva[mask]['MVA'])
mva_sum += mva
covar.loc[covar['Ticker'] == ticker, 'Weighted'] = mva * covar.loc[covar['Ticker'] == ticker, 'COVAR']
portfolio_covar = covar['Weighted'].sum() / mva_sum
return portfolio_covar
def portfolio_ses(datamatrix, banks, year, quarter):
port_ses = 0
dm = datamatrix.copy()
dm['Weighted'] = 0
mva_sum = 0
for ticker in dm['Ticker']:
bank = find_ticker_in_list(ticker, banks)
mask = (bank.mva['Year'] == year) & (bank.mva['Quarter'] == quarter)
mva = float(bank.mva[mask]['MVA'])
mva_sum += mva
dm.loc[dm['Ticker'] == ticker,
'Weighted'] = mva * dm.loc[dm['Ticker'] == ticker, 'SES']
port_ses = dm['Weighted'].sum() / mva_sum
return port_ses
import pandas as pd
import bank
from getdata import get_banks_data
if __name__ == '__main__':
banks = get_banks_data()
yealds1 = bank.find_ticker_in_list('ACA:FP', banks).yealds
yealds2 = bank.find_ticker_in_list('AKTAV:FH', banks).yealds
yeald_inner_join = pd.merge(yealds1, yealds2, on='Date', how='inner')
print(yeald_inner_join[['Yeald_x', 'Yeald_y']])
print(yeald_inner_join.head())
print(yeald_inner_join.tail())
date_start = "2010-1-10"
date_end = "2011-1-1"
mask = (yeald_inner_join['Date'] > date_start) & (yeald_inner_join['Date']
<= date_end)
yeald_inner_join = yeald_inner_join.loc[mask]
print(yeald_inner_join.head())
predictions = results.predict(X)
df = x.yealds
plt.plot(df['Date'], df['Yeald'], label="Sample")
plt.plot(df['Date'], predictions, label="Prediction")
plt.title(x.name)
plt.legend()
plt.xlabel('Date')
plt.ylabel('Yeald')
plt.show()
#Testo il metodo di ricerca in lista
if False:
x = bank.find_ticker_in_list('ACA:FP', banks)
print(x.name)
yealds = x.yealds
print(yealds['Date'])
#Testo il metodo get_states_variable()
if False:
df = get_states_variable()
print(df.head())
#Plotto il grafico andamentale del portfolio covar
if False:
dates = [[2005, 3, 2007, 2], [2005, 4, 2007, 3], [2006, 1, 2007, 4],