def simulate(start, end, symbols, weights, data_path) :
import pandas as pd
# check data available, refresh if necessary
get_history(symbols, start, end, data_path)
pdata = pd.Panel(dict((symbols[i], pd.read_csv(data_path + symbols[i] + '.csv',\
index_col='Date', parse_dates=True).sort(ascending=True)) for i in range(len(symbols))) )
starting_cash = 1000000
# create empty dataframe with index = dates start-end
df = pd.DataFrame(index=pdata.major_axis).ix[start:end]
for i in range(len(symbols)):
ticker = symbols[i]
allocation = weights[i]
df[ticker] = pdata[ticker]['Adj Close']
df[ticker + ' C'] = df[ticker] / df[ticker][0]
df[ticker + ' C'][0] = 1.
df[ticker + ' I'] = 0.
df[ticker + ' I'][0] = starting_cash * allocation
df[ticker + ' I'] = [df[ticker + ' C'][j] * df[ticker + ' I'][0] for j in range(len(df))]
df['Total'] =pd.DataFrame([df[ticker + ' I'] for ticker in symbols]).sum()
df['Daily Rets'] = df.Total / df.Total.shift(1) - 1
return df[['Daily Rets', 'Total']]
def analyze(values, data_path, index_ticker='SPY', p_print='No'):
import pandas as pd
from finlib.get_history import get_history
from math import sqrt
idx_ticker = index_ticker
values = values
start = values.index.min()
end = values.index.max()
market_index = get_history([idx_ticker], start, end, data_path).ix[:,:,'Adj Close']
market_index['d_rets'] = market_index[idx_ticker] / market_index[idx_ticker].shift(1) - 1
values['d_rets'] = values['Total'] / values['Total'].shift(1) - 1
metrics = [''] * 8
metrics[0] = values['d_rets'].mean()
metrics[4] = market_index['d_rets'].mean()
metrics[1] = values['d_rets'].std() * sqrt(252)
metrics[5] = market_index['d_rets'].std() * sqrt(252)
metrics[2] = sqrt(len(values['d_rets'])) * values['d_rets'].mean() / values['d_rets'].std()
metrics[6] = sqrt(len(market_index['d_rets'])) * market_index['d_rets'].mean() / market_index['d_rets'].std()
metrics[3] = values['Total'][-1] / values['Total'][0]
metrics[7] = market_index['SPY'][-1] / market_index['SPY'][0]
if p_print == 'Yes':
print '\nPortfolio Metrics'
print '#################\n'
print 'Date Range: ', start , 'to', end
print
print 'Final Value of Portfolio: ', "{0:,.2f}".format(values['Total'][-1])
print
print 'Sharpe Ratio: ', metrics[2]
print 'Index Sharpe Ratio: ', metrics[6]
print
print 'Period Return: ', "{0:.3f}".format(metrics[3] * 100), '%'
print 'Index Period Return: ', "{0:.3f}".format(metrics[7] * 100), '%'
print
print 'Period Volatilty: ', "{0:.2f}".format(metrics[1]), '%'
print 'Period Index Volatility: ', "{0:.2f}".format(metrics[5]), '%'
print
print 'Average Daily Return :', "{0:.3f}".format(metrics[0] * 100), '%'
print 'Index Average Daily Return :', "{0:.3f}".format(metrics[4] * 100), '%'
return metrics
def marketsim(starting_cash, orders, data_path):
import datetime as dt
import pandas as pd
import pandas.io.data as web
from finlib.get_history import get_history
symbols = list(set(orders.symbol))
startday = min(orders.index)
endday= max(orders.index)
raw_data = get_history(symbols, startday, endday, data_path)
prices = raw_data.ix[:,:,'Adj Close']
# empty dfs for holdings and port_value
holdings = pd.DataFrame(0, columns=symbols, index=prices.index)
port_value = pd.DataFrame(0, columns=['Cash', 'Value', 'Total'], index=prices.index)
for i in range(len(orders)):
if orders.ix[i]['action'] == 'Buy':
holdings.ix[orders.index[i]][orders.symbol[i]] += orders['qty'][i]
port_value.ix[orders.index[i]]['Cash'] -= orders['qty'][i] * prices.ix[orders.index[i]][orders.symbol[i]]
elif orders.ix[i]['action'] == 'Sell':
holdings.ix[orders.index[i]][orders.symbol[i]] -= orders['qty'][i]
port_value.ix[orders.index[i]]['Cash'] += orders['qty'][i] * prices.ix[orders.index[i]][orders.symbol[i]]
else:
print 'Bad order'
raise
port_value['Cash'][0] += starting_cash
port_value = port_value.cumsum()
holdings = holdings.cumsum()
port_value['Value'] = (prices * holdings).sum(axis=1)
port_value['Total'] = port_value.sum(axis=1)
return port_value
