def test_SP500_strat( ):
#Already pickled DJIA_AdjClose
dataObj = pickle.load(open('pickle_jar/SP500_AdjClose.pkl'))
dataObj.to_csv('Results/tmptmp.csv')
####THIS IS WEIRD THAT DB HAS ADJ CLOSE AS VALUE
###NEED TO ADD TO PARSING IN READ DATA
dataObj = dataObj.replace('Adj Close',np.nan).astype('float64')
#Read DJIA index
dbr = DBReader()
ref_series = dbr.readSeries(seriesName = 'SP500')
ref_series = ref_series['Adj_Close']
ref_series = pd.DataFrame(ref_series)#,columns=['DJIA'])
ref_series.columns = ['SP500']
series_list = dataObj.columns.values
#setup train/test ranges
period_length_1 = 90
period_length_2 = 180
start_date = dataObj.index[0]
end_date = dataObj.index[-1]
# start_date = datetime.datetime(2009,1,1)
# end_date = datetime.datetime(2010,1,1)
periods_dates = pd.date_range(start_date,end_date,freq = '3BMS')
#Do example
#Setup strat
#parameters
top_scores_num = 10
ema_par = 100
lag = 3
lookback = 40
#Initialise strat
strat = StrategyMSeriesMomentum(periods_dates,
ema_par,
lookback,
lag,
top_scores_num)
#Train - ie pick top X momentum and weights
#want generate weight matrix of constant weights between rebalance dates
strat.fit(dataObj)
sumweight =strat.weights_matrix.sum(axis=1,skipna=True)
# sumweight.plot()
strat.run_strategy(data=dataObj)
res = strat.result
ref_seriesindex = (ref_series/(ref_series.iloc[0]))
pltData = pd.merge(res,ref_seriesindex,how='inner',left_index=True,right_index=True)
pltData.plot()
plt.show()
from ResultsAnalyser import ResultsAnalyser
ra = ResultsAnalyser(pltData,valueIndex='Value',referenceIndex='SP500')
tstart = datetime.datetime(2010,6,1)
tend = datetime.datetime(2013,6,1)
ra.set_results_range(tstart, tend)
print "sa sharpe ratio", ra.sharpe_ratio()
print "to ref sharpe ratio", ra.sharpe_ratio(useMarketRef=True)
def test_strat():
#Already pickled DJIA_AdjClose
dataObj = pickle.load(open('pickle_jar/DJIA_AdjClose.pkl'))
#Read DJIA index
dbr = DBReader()
DJIA = dbr.readSeries(seriesName = 'DJIA')
#dataObj['DJIA'] = DJIA
DJIA = pd.DataFrame(DJIA)#,columns=['DJIA'])
DJIA.columns = ['DJIA']
series_list = dataObj.columns.values
#setup train/test ranges
period_length_1 = 90
period_length_2 = 180
start_date = dataObj.index[0]
end_date = dataObj.index[-1]
periods_dates = pd.date_range(start_date,end_date,freq = '3BMS')
periods = [ [periods_dates[i], periods_dates[i+1] ] for i in range(0, len(periods_dates)-1)]
train_periods = periods[0:-1]
test_periods = periods[1:]
#Do example
#setup data
print len(train_periods)
period_num = 32
train_data = dataObj.loc[train_periods[period_num][0]:train_periods[period_num][1]]
test_data = dataObj.loc[test_periods[period_num][0]:test_periods[period_num][1]]
#Setup strat
#parameters
top_scores_num = 10
ema_par = 50
lag = 1
lookback = 10
#Initialise strat
strat = StrategyMSeriesMomentum( ema_par, lookback, lag, top_scores_num)
#Train - ie pick top X momentum and weights
strat.fit(train_data)
#Setup portfolio
series_list = strat.top_series_weights.index.values
portfolio = strat.setup_portfolio( series_list )
portfolio_train = portfolio.copy()
#Check result on training
strat.run_strategy(market_data=train_data,
portfolio = portfolio_train)
train_result = strat.result
#Check results on test
portfolio_test = portfolio.copy()
strat.run_strategy(market_data=test_data,
portfolio = portfolio_test)
test_result =strat.result
#Analyse results
from ResultsAnalyser import ResultsAnalyser
train_result = pd.merge(train_result,DJIA,how='inner',left_index=True, right_index=True)
ra_train = ResultsAnalyser(data = train_result,
valueIndex = 'Value',
referenceIndex = 'DJIA' )
rets = ra_train.get_returns(annualing_factor=1, useReference=True)
rets += 1
cumrets = rets.cumprod(axis=0)
# print cumrets.tail()
cumrets.plot()
plt.show()
# print train_result.head()
print "Sharpe sa train", ra_train.sharpe_ratio()
print "Sharpe reference train", ra_train.sharpe_ratio(useMarketRef=True)
test_result = pd.merge(test_result,DJIA,how='inner',left_index=True, right_index=True)
ra_test = ResultsAnalyser(data = test_result,
valueIndex = 'Value',
referenceIndex = 'DJIA' )
rets = ra_test.get_returns(annualing_factor=1, useReference=True)
rets += 1
cumrets = rets.cumprod(axis=0)
# print cumrets.tail()
cumrets.plot()
plt.show()
# print train_result.head()
print "Sharpe sa test", ra_test.sharpe_ratio()
print "Sharpe reference test", ra_test.sharpe_ratio(useMarketRef=True)
def test_DJIA_strat(plots = False):
#Already pickled DJIA_AdjClose
dataObj = pickle.load(open('pickle_jar/DJIA_AdjClose.pkl'))
#Read DJIA index
dbr = DBReader()
DJIA = dbr.readSeries(seriesName = 'DJIA')
#dataObj['DJIA'] = DJIA
DJIA = pd.DataFrame(DJIA)#,columns=['DJIA'])
DJIA.columns = ['DJIA']
series_list = dataObj.columns.values
#setup train/test ranges
period_length_1 = 90
period_length_2 = 180
start_date = dataObj.index[0]
end_date = dataObj.index[-1]
periods_dates = pd.date_range(start_date,end_date,freq = '3BMS')
#Do example
#Setup strat
#parameters
top_scores_num = 5
ema_par = 50
lag = 1
lookback = 10
#Initialise strat
strat = StrategyMSeriesMomentum(periods_dates,
ema_par,
lookback,
lag,
top_scores_num)
#Train - ie pick top X momentum and weights
#want generate weight matrix of constant weights between rebalance dates
strat.fit(dataObj)
print strat.rebalance_dates
rebalance_flag = pd.DataFrame( [True]*len(strat.rebalance_dates) , index = strat.rebalance_dates)
rebalance_flag = rebalance_flag.resample('B')
rebalance_flag[pd.isnull(rebalance_flag)] = False
rebalance_flag.index.name = 'date'
rebalance_flag.columns = ['Rebalance_Flag']
rebalance_flag.to_csv('Results/StrategyMSeries_Momentum/rebalance_flag.csv')
# strat.weights_matrix.to_csv('Results/StrategyMSeries_Momentum/DJIA_weight_matrix.csv')
sumweight =strat.weights_matrix.sum(axis=1,skipna=True)
# sumweight.plot()
strat.run_strategy(data=dataObj)
res = strat.result
DJIAindex = (DJIA/(DJIA.iloc[0]))
tstart = datetime.datetime(2012,7,2)
tend = datetime.datetime(2013,7,4)
pltData = pd.merge(res,DJIAindex,how='inner',left_index=True,right_index=True)
if plots:
pltData.loc[tstart:tend].plot()
plt.show()
from ResultsAnalyser import ResultsAnalyser
ra = ResultsAnalyser(pltData,valueIndex='Value',referenceIndex='DJIA')
ra.set_results_range(tstart, tend)
print "sa sharpe ratio", ra.sharpe_ratio()
print "sa returns", ra.get_cumulative_return().iloc[-1]
print "sa vol", ra.get_volatility(annualising_scalar=1,returns = True)
print "to ref sharpe ratio", ra.sharpe_ratio(useMarketRef=True)
balance_matrix = strat.weights_matrix.loc[strat.rebalance_dates]#.iloc[-5:-1]
balance_matrix.to_csv('Results/StrategyMSeries_Momentum/balance_matrix.csv')