return df_trades
if __name__ == '__main__':
symbols = ['JPM']
start_date = '2008-01-01'
end_date = '2009-12-31'
# #for out-sample time using following sd and ed
# start_date = '2010-01-01'
# #end_date = '2011-12-31'
prices = get_data(symbols, pd.date_range(start_date, end_date))
prices = prices[symbols]
# # print prices
df_trades = testPolicy(sd=start_date, ed=end_date)
# df_joined = df_trades.join(prices, lsuffix='_best', rsuffix='_benchmark')
portvals = marketsim(symbols[0], df_trades, commission = 9.95, impact = 0.005)
print_info(portvals)
# # portvals = marketsim(df_trades, prices, commission = 0, impact = 0)
# df_joined = df_joined.join(portvals, lsuffix='_best', rsuffix = 'whatever')
# prices_val = prices.values
# # to generate Benchmark values
# d = np.zeros(len(prices.index))
# d[0] = 1000
# df_trade_none = pd.DataFrame(data=d, index=prices.index, columns = ['val'])
# port_benchmark = marketsim(df_trade_none, prices)
# portvals = portvals / portvals.ix[0]
# port_benchmark = port_benchmark / port_benchmark.ix[0]
print 'cumulative return: ' + str(
float(portvals.values[-1] / portvals.values[0]) - 1)
print 'Stdev of daily returns: ' + str(float(daily_returns.std()))
print 'Mean of daily returns: ' + str(float(daily_returns.mean()))
if __name__ == '__main__':
names = ['JPM']
start_date = '2008-01-01'
end_date = '2009-12-31'
prices = get_data(names, pd.date_range(start_date, end_date))
prices = prices[names]
df_trades = testPolicy()
portvals = marketsim(df_trades, prices)
print 'Best Possible Policy'
print_info(portvals)
d = np.zeros(len(prices.index))
d[0] = 1000
df_trade_none = pd.DataFrame(data=d, index=prices.index, columns=['val'])
port_benchmark = marketsim(df_trade_none, prices)
print 'Benchmark'
print_info(port_benchmark)
portvals = portvals / portvals.ix[0]
port_benchmark = port_benchmark / port_benchmark.ix[0]
# df_joined = portvals.join(port_benchmark, lsuffix='_best', rsuffix='_benchmark')
# plot_data(df_joined)
symbol = 'JPM'
# sd = datetime.datetime(2010, 1, 1)
# ed = datetime.datetime(2011, 12, 31)
sd = datetime.datetime(2008, 1, 1)
ed = datetime.datetime(2009, 12, 31)
dates = pd.date_range(sd, ed)
prices_all = ut.get_data([symbol], dates) # automatically adds SPY
prices = ut.get_data([symbol], pd.date_range(sd, ed))
prices = prices[symbol]
benchmark_trades = prices_all[[symbol,]].copy(deep=True) # only portfolio symbols
benchmark_trades.values[:, :] = 0
benchmark_trades.values[0, :] = 1000
vals_bench = marketsim(benchmark_trades, prices)
learner = sl.StrategyLearner(verbose = False, impact = 0) # constructor
learner.addEvidence(symbol = symbol, sd=sd, ed=ed, sv = 100000) # training phase
df_trades = learner.testPolicy(symbol = symbol, sd=sd, ed=ed, sv = 100000) # testing phase
vals_ml = marketsim(df_trades, prices, impact = 0)
manual_trades = ms.testPolicy(symbol=symbol, sd=sd, ed=ed)
vals_manual = marketsim(manual_trades, prices, impact = 0, commission = 0)
# portvals = portvals / portvals.ix[0]
vals_bench = vals_bench / vals_bench.ix[0]
vals_ml = vals_ml / vals_ml.ix[0]
vals_manual = vals_manual / vals_manual.ix[0]
benchmark, = plt.plot(vals_bench, 'b', label = 'Benchmark')
prices = ut.get_data([symbol], pd.date_range(sd, ed))
prices = prices[symbol]
res = []
num_trades = []
impacts = [0, 0.025, 0.05, 0.075, 0.1, 0.125, 0.15, 0.175, 0.2, 0.3, 0.4]
for impact in impacts:
learner = sl.StrategyLearner(verbose=False,
impact=impact) # constructor
learner.addEvidence(symbol=symbol, sd=sd, ed=ed,
sv=100000) # training phase
df_trades = learner.testPolicy(symbol=symbol, sd=sd, ed=ed,
sv=100000) # testing phase
num_trades.append(np.count_nonzero(df_trades))
vals_ml = marketsim(df_trades, prices, impact=impact)
vals_ml = vals_ml / vals_ml.ix[0]
res.append(vals_ml)
plt.xlabel('impact')
plt.ylabel('number of trades')
plt.xticks(np.arange(11), impacts)
plt.plot(num_trades)
plt.show()
handles = []
for i in range(len(impacts)):
mlstrategy, = plt.plot(res[i], label=str(impacts[i]))
handles.append(mlstrategy)
if __name__ == '__main__':
syms = ['JPM']
start_date = '2008-01-01'
end_date = '2009-12-31'
#for out-sample time using following sd and ed
#start_date = '2010-01-01'
#end_date = '2011-12-31'
prices = get_data(syms, pd.date_range(start_date, end_date))
prices = prices[syms]
# print prices
df_trades = testPolicy(sd=start_date, ed=end_date)
df_joined = df_trades.join(prices, lsuffix='_best', rsuffix='_benchmark')
portvals = marketsim(df_trades, prices, commission=9.95, impact=0.005)
# portvals = marketsim(df_trades, prices, commission = 0, impact = 0)
df_joined = df_joined.join(portvals, lsuffix='_best', rsuffix='whatever')
prices_val = prices.values
# to generate Benchmark values
d = np.zeros(len(prices.index))
d[0] = 1000
df_trade_none = pd.DataFrame(data=d, index=prices.index, columns=['val'])
port_benchmark = marketsim(df_trade_none, prices)
portvals = portvals / portvals.ix[0]
port_benchmark = port_benchmark / port_benchmark.ix[0]
print "My strategy performance"
print_info(portvals)
sd = dt.datetime(2008, 1, 1)
ed = dt.datetime(2009, 12, 31)
dates = pd.date_range(sd, ed)
#import all prices with SPY
prices_all = ut.get_data([symbol], dates)
#only prices in symbol
prices = prices_all[symbol]
#benchmark portvals table
trades_benchmark = prices_all[[
symbol,
]].copy(deep=True)
#benchmark buy and hold
trades_benchmark.values[:, :] = 0
trades_benchmark.values[0, :] = 1000
vals_benchmark = marketsim(trades_benchmark,
prices) #default impact and commission are 0
#ML portvals table
rtl = sl.StrategyLearner(verbose=False, impact=0)
#train learner
rtl.addEvidence(symbol=symbol, sd=sd, ed=ed, sv=100000)
#testing
trades_rtl = rtl.testPolicy(symbol=symbol, sd=sd, ed=ed, sv=100000)
vals_rtl = marketsim(trades_rtl,
prices) #default impact and commission are 0
#ML_3indicators portvals table
# rtl3 = sl1.StrategyLearner(verbose = False, impact = 0)
#train learner
# rtl3.addEvidence(symbol = symbol, sd = sd, ed = ed, sv = 100000)
#testing
print 'cumulative return: ' + str(
float(portvals.values[-1] / portvals.values[0]) - 1)
print 'Stdev of daily returns: ' + str(float(daily_returns.std()))
print 'Mean of daily returns: ' + str(float(daily_returns.mean()))
if __name__ == '__main__':
names = ['JPM']
start_date = '2008-01-01'
end_date = '2009-12-31'
prices = get_data(names, pd.date_range(start_date, end_date))
prices = prices[names]
trades_df = testPolicy()
portvals = marketsim(trades_df, prices)
print 'Best Possible Policy performance'
print_info(portvals)
d = np.zeros(len(prices.index))
d[0] = 1000
trade_df_none = pd.DataFrame(data=d, index=prices.index, columns=['val'])
port_benchmark = marketsim(trade_df_none, prices)
print 'Benchmark performance'
print_info(port_benchmark)
#normalization the portvals
portvals = portvals / portvals.ix[0]
port_benchmark = port_benchmark / port_benchmark.ix[0]
# plot the graph
prices = ut.get_data([symbol], pd.date_range(sd, ed))
prices = prices[symbol]
res = []
trades_num = []
impacts = [0, 0.025, 0.05, 0.1, 0.2, 0.4]
for impact in impacts:
learner = sl.StrategyLearner(verbose = False, impact = impact)
# constructor
#training the learner
learner.addEvidence(symbol = symbol, sd=sd, ed=ed, sv = 100000)
#testing Period
trades_df = learner.testPolicy(symbol = symbol, sd=sd, ed=ed, sv = 100000)
trades_num.append(np.count_nonzero(trades_df))
vals_rtl = marketsim(trades_df, prices, impact = impact)
vals_rtl = vals_rtl / vals_rtl.ix[0]
print_info(vals_rtl)
print trades_num
res.append(vals_rtl)
plt.xlabel('impact')
plt.ylim(ymax = 60)
plt.ylabel('number of trades')
plt.xticks(np.arange(6), impacts)
plt.plot(trades_num)
plt.title("Impact on number of trades")
plt.show()
if __name__ == '__main__':
sd = dt.datetime(2008, 1, 1)
ed = dt.datetime(2009, 12, 31)
dates = pd.date_range(sd, ed)
symbol = 'JPM'
prices_all = ut.get_data([symbol], dates)
prices = prices_all[symbol]
summary1 = []
trades_number = []
impacts = [0, 0.05, 0.10, 0.15, 0.20, 0.25, 0.30]
for impact in impacts:
rtl = sl.StrategyLearner(verbose=False, impact=impact)
rtl.addEvidence(symbol=symbol, sd=sd, ed=ed, sv=100000)
trades_rtl = rtl.testPolicy(symbol=symbol, sd=sd, ed=ed, sv=100000)
trades_number.append(np.count_nonzero(trades_rtl))
values_rtl = marketsim(symbol, trades_rtl, impact=impact)
values_rtl = values_rtl / values_rtl.iloc[0]
information(values_rtl)
print trades_number
summary1.append(values_rtl)
plt.xlabel('Impact')
plt.ylabel('Number of trades')
plt.xticks(np.arange(7), impacts)
plt.plot(trades_number)
plt.title("Impact VS. Number of trades")
plt.show()
summary2 = []
for i in range(len(impacts)):
rtlstrategy, = plt.plot(summary1[i], label=str(impacts[i]))
summary2.append(rtlstrategy)
plt.gcf().subplots_adjust(left=0.1,
float(daily_returns.std()))
if __name__ == "__main__":
sd = dt.datetime(2008, 1, 1)
ed = dt.datetime(2009, 12, 31)
dates = pd.date_range(sd, ed)
symbol = 'JPM'
prices_all = ut.get_data([symbol], dates)
prices = prices_all[symbol]
benchmark_record = prices_all[[
symbol,
]].copy()
benchmark_record.values[:, :] = 0
benchmark_record.values[0, :] = 1000
values_benchmark = marketsim(symbol, benchmark_record)
rtl = sl.StrategyLearner(verbose=False, impact=0)
rtl.addEvidence(symbol=symbol, sd=sd, ed=ed, sv=100000)
trades_rtl = rtl.testPolicy(symbol=symbol, sd=sd, ed=ed, sv=100000)
values_rtl = marketsim(symbol, trades_rtl)
trades_manual = manu.testPolicy(symbol=symbol, sd=sd, ed=ed)
values_manual = marketsim(symbol, trades_manual)
values_benchmark = values_benchmark / values_benchmark.iloc[0]
values_manual = values_manual / values_manual.iloc[0]
values_rtl = values_rtl / values_rtl.iloc[0]
benchmark_record, = plt.plot(values_benchmark, "b", label="benchmark")
rtl, = plt.plot(values_rtl, "g", label="RTLearner")
manual_strategy, = plt.plot(values_manual, "r", label="Manual Strategy")
plt.legend(handles=[benchmark_record, manual_strategy, rtl], loc=0)
plt.gcf().subplots_adjust(left=0.1,
bottom=0.05,