def grid_tester(run_strat, do_plots): ##################Initialise########################## #Read in data and setup object # dbpath = "/home/phcostello/Documents/Data/FinanceData.sqlite" # dbreader = dbr.DBReader(dbpath) #SP500 = dbreader.readSeries("SP500") #BA = dbreader.readSeries("BA") #dim = 'Adj_Close' #SP500AdCl = SP500[dim] #BAAdCl = BA[dim] #dataObj = pd.merge(pd.DataFrame(BAAdCl), pd.DataFrame(SP500AdCl), how='inner',left_index = True, right_index = True) #dataObj.columns = ['y','x'] #Read data for all pairs #Note have checked top 10 in and constructed data in #CheckingSeriesCoint.py dataObj = pickle.load(open('pickle_jar/DJIA_AdjClose.pkl')) cointAnalysis = pickle.load(open('pickle_jar/results_DJIA_coint.pkl','rb')) cointAnalysis = pd.DataFrame(cointAnalysis, columns = ['SeriesX_name','SeriesY_name','Adf_pvalue', 'reg_scaling', 'reg_intercept']) cointAnalysis = cointAnalysis.sort('Adf_pvalue', ascending=True) #cointAnalysis.info() seriesNames = cointAnalysis[['SeriesX_name','SeriesY_name']] #dataObj.info() top10 = seriesNames.iloc[0:5] top10['rank']= range(1, len(top10)+1) top10.head() top10 =top10.values.tolist() #Filter to what we want #setup logging logfile = 'logs/GridTester_ma.txt' logging.basicConfig(filename= logfile, filemode='w', level = logging.ERROR) ##################Make par Table########################## #Parameter table rows are #[ pairs_labes, strategy_params, train_ranges, test_ranges ] #series series_names = top10 #strat pars entry_scale = np.arange(1,3.5,0.5) exit_scale = np.arange(0.0,3.0,0.5) strat_parameters = [ [sw, lw] for sw in entry_scale for lw in exit_scale if sw >= lw] #Train - test date pars dmin = datetime.datetime(2006,1,1) num_periods = 15 period_delta = datetime.timedelta(182) #Create range start/end offsets from min data train_test_ranges = [ [dmin + i*period_delta, #train start dmin + (i+1)* period_delta, #train end dmin + (i+1)* period_delta, #test start dmin + (i+2)* period_delta] #test end for i in range(0,num_periods)] #Make table of parameters + dates parameter_table = [ series + pars + date_ranges for series in series_names for pars in strat_parameters for date_ranges in train_test_ranges] parameter_table = [ parameter_table[i] + [i] for i in range(0,len(parameter_table))] colnames =['runtype',\ 'series_X',\ 'series_Y',\ 'rank',\ 'entry_scale',\ 'exit_scale',\ 'train_start',\ 'train_end',\ 'test_start',\ 'test_end',\ 'par_rownumber'] ##################Make Res Table########################## #We just define result_cols = ['returns',\ 'volatility',\ 'sharpe',\ 'sortino'] #Use x will be analyser object which has result functions #result cols should have same order as function below result_func = lambda x : [x.get_result().iat[-1,0]/x.get_result().iat[0,0] - 1 , x.get_volatility(), x.sharpe_ratio(), x.sortino_ratio()] result_table_cols = colnames + result_cols ##################Test strat on pars in Table########################## if run_strat: #open file for output outfile = open('Results/dummy.csv','wb') #run simulations tic = time.clock() runStrat(parameter_table,dataObj, outfile, result_table_cols, result_func) toc = time.clock() print "All strategies took {} seconds to run".format(toc - tic) outfile.close() #################$ Plot for one set of parameters ######################## #Setup Data if do_plots: this_pars = parameter_table[496] #Setup Data #Setup initial portfolio trade_equity_spread = TradeEquity('spread', notional=0, price_series_label='spread') portfolio_0 = Portfolio("portfolio", cashAmt=100) portfolio_0.add_trade(trade_equity_spread) #No more trade types portfolio_0.fixed_toggle() print this_pars strategy, train_data, test_data = setup_strategy(this_pars,dataObj) data = test_data strategy.run_strategy(data,portfolio_0) analyser = ResultsAnalyser(strategy.result,referenceIndex=None) fig =plt.figure() ax1 = fig.add_subplot(2,1,1) ax2 = fig.add_subplot(2,1,2) portfolioVal = analyser.get_result() print portfolioVal['Value'] start = 0 end = len(portfolioVal) portfolioVal['Value'].iloc[start:end].plot(ax=ax1) # trades_at = portfolioVal[portfolioVal['Signal_bs'].isin(['buy','sell'])] # # #Putting trade bars on # for dt in trades_at.index: # plt.axvline( dt, ymin =0, ymax =1 ) plt.subplot(2,1,1) pairs_md(data.loc[portfolioVal.index].iloc[start:end],0,0).plot_spreadAndSignals(ax2) #plt.title(seriesName) plt.show()
def test_ewma_strat(): #prepare data import DataHandler.DBReader as dbr from strategy_tester.trade import TradeEquity from strategy_tester.Portfolio import Portfolio dbpath = "/home/phcostello/Documents/Data/FinanceData.sqlite" dbreader = dbr.DBReader(dbpath) SP500 = dbreader.readSeries("SP500") dim = 'Adj_Close' data = SP500 startDate_train = datetime.date(2012,12,30) endDate_train = datetime.date(2013,6,30) dataReduced = data.loc[startDate_train:endDate_train] #Initialise strategy parameters strat = EWMA_Trend( MA_win_len=50.0, std_win_len=50.0, entryScale= 0, exitScale= 1) #Initialize Portfolio #Generate Signals on Market data #Run strategy #Setup portfolio trade_equity_spread = TradeEquity('Equity', notional=0, price_series_label='Adj_Close') port = Portfolio("portfolio", cashAmt=100) port.add_trade(trade_equity_spread) #No more trade types port.fixed_toggle() strat.plot_flag=True strat.run_strategy(market_data = data, portfolio = port) # plt.show() plt.figure() strat.result['Value'].plot() # plt.show() # strat.result.to_csv('Results/pairsmd.csv') from strategy_tester.ResultsAnalyser import ResultsAnalyser ra = ResultsAnalyser(data= strat.result) #Sharpe for strat print ra.sharpe_ratio(), ra.get_cumulative_return().iloc[-1,0] ra2 = ResultsAnalyser(data = SP500,valueIndex='Adj_Close') #Sharpe for SP500 print ra2.sharpe_ratio(), ra2.get_cumulative_return().iloc[-1,0] combresdata = pd.merge(strat.result,SP500, how='inner', left_index=True, right_index=True) ra3 = ResultsAnalyser(data = combresdata,valueIndex='Value',referenceIndex='Adj_Close') print ra3.sharpe_ratio(useMarketRef=True), ra3.get_cumulative_return(useReference=True).iloc[-1,0]
dbreader = DBReader() AAPL = dbreader.readSeries('AAPL') SP500 = dbreader.readSeries('SP500') print AAPL.info() print SP500.info() data = pd.merge(pd.DataFrame(AAPL['Adj_Close']),pd.DataFrame(SP500['Adj_Close']), how='inner', left_index=True, right_index=True) data_dict = {'AAPL':AAPL , 'SP500':SP500} panel_series = pd.Panel.from_dict(data_dict, intersect=True, orient = 'minor') print 'panel items', panel_series.items dim = 'Adj_Close' df_allseries_dim = panel_series[dim] dataRed = df_allseries_dim.loc['2011-1-1':'2012-1-1'] rets = dataRed['SP500'].pct_change() print 'rets std', rets.std() print rets.head() #dataRed['SP500'].plot() plt.show() ra = ResultsAnalyser(dataRed,valueIndex = 'SP500')#, referenceIndex='SP500') print "vol", ra.get_volatility(annualising_scalar=1, returns=True) sharpe = ra.sharpe_ratio(useMarketRef=False) print sharpe