Backtest with custom columns problem.
-
Hello,
I have a predicted file with action columns (it is a ticker for buy/sell/hold) . I would like to load this csv and use feeds data via PandasData Class and I found an error at a cerebro.run(). Could you please suggest me.Thank you.
my dataframe
dataframe.head() Out[58]: Datetime Open High Low Close Signal 0 2011-06-13 11:00:00 705.56 705.76 703.67 704.41 0.0 1 2011-06-13 12:00:00 704.25 704.52 701.51 701.69 1.0 2 2011-06-13 14:00:00 704.69 709.42 704.39 706.12 0.0 3 2011-06-13 15:00:00 705.86 708.25 705.68 707.52 2.0 4 2011-06-13 16:00:00 707.37 709.80 706.28 708.76 0.0import pandas as pd import backtrader as bt from backtrader.feeds import PandasData # load dataframe dataframe = pd.read_csv("model/SET50_H1_2011_2017_Model1_Predicted.csv",names = ["Datetime","Open","High","Low","Close","action"]) dataframe['Datetime'] = pd.to_datetime(dataframe['Datetime'],format='%Y-%m-%d %H:%M:%S') class PandasData_Signal(PandasData): # Add a 'action' line to the inherited ones from the base class lines = ('action',) # add the parameter to the parameters inherited from the base class params = (('action', 7),) data = PandasData_Signal(dataname=dataframe, #dtformat=('%Y-%m-%d %H:%M:%S'), #timeframe=bt.TimeFrame.Minutes, #tmformat=('%H:%M:%S'), datetime=0, open=1, high=2, low=3, close=4, #volume=-1, action=5, #openinterest=-1 #fromdate=date(2017,1,1), #todate=date(2017,1,10) ) class MLSignal(bt.SignalStrategy): def log(self, txt, dt=None): pass def __init__(self): # Keep a reference to the "close" line in the data[0] dataseries self.dataclose = self.datas[0].close self.action = self.datas[0].action def next(self): self.log(' Close, %.2f' % self.dataclose[0]) if self.order: return if self.action[0] == 1.0: self.log('BUY CREATE, %.2f' % self.dataclose[0]) self.order = self.buy() if self.action[0] == 2.0: self.log('SELL CREATE, %.2f' % self.dataclose[0]) self.order = self.sell() cerebro = bt.Cerebro() # Set our desired cash start cerebro.broker.setcash(1000000.0) cerebro.adddata(data) cerebro.addstrategy(MLSignal) cerebro.run() print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue()) print('Final value is %.2f times the initial investment'%(cerebro.broker.getvalue()/1000000.0)) cerebro.plot()Error code
File "/Users/tatum/anaconda3/lib/python3.6/site-packages/spyder/utils/site/sitecustomize.py", line 710, in runfile execfile(filename, namespace) File "/Users/tatum/anaconda3/lib/python3.6/site-packages/spyder/utils/site/sitecustomize.py", line 101, in execfile exec(compile(f.read(), filename, 'exec'), namespace) File "/Volumes/SANDISK/Algo_Trader_and_Robot_ProjectTraining/backtrader_code/test5.py", line 96, in <module> cerebro.run() File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/cerebro.py", line 1127, in run runstrat = self.runstrategies(iterstrat) File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/cerebro.py", line 1290, in runstrategies self._runonce(runstrats) File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/cerebro.py", line 1691, in _runonce strat._oncepost(dt0) File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/strategy.py", line 289, in _oncepost self.nextstart() # only called for the 1st value File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/lineiterator.py", line 342, in nextstart self.next() File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/strategy.py", line 1579, in _next_catch self._next_custom() File "/Volumes/SANDISK/Algo_Trader_and_Robot_ProjectTraining/backtrader_code/test5.py", line 78, in next if self.order: File "/Users/tatum/anaconda3/lib/python3.6/site-packages/backtrader/lineseries.py", line 461, in __getattr__ return getattr(self.lines, name) AttributeError: 'Lines_LineSeries_LineIterator_DataAccessor_Strateg' object has no attribute 'order'
-
@Ta-Tum looks like self.order need to be initialized in the init():
self.order = None
-
@ab_trader, it's work. but I found new problem that a new column to feed is nan (all value). Please suggest me.
Thank you
class PandasData_Signal(PandasData): # Add a 'action' line to the inherited ones from the base class lines = ('action',) # add the parameter to the parameters inherited from the base class params = (('action', 8),) data = PandasData_Signal(dataname=dataframe, datetime=0, open=1, high=2, low=3, close=4, #volume=-1, action=5 #openinterest=-1 #fromdate=date(2017,1,1), #todate=date(2017,1,10) )error
2017-09-25T16:00:00, Close , 1067.41 2017-09-25T16:00:00, action , nan 2017-09-25T17:00:00, Close , 1070.88 2017-09-25T17:00:00, action , nan
-
@Ta-Tum Hey mate. Please could you provide your file you read from, then I'll run through it and can probably figure it out quite quickly.
I noticed your line:
dataframe = pd.read_csv("model/SET50_H1_2011_2017_Model1_Predicted.csv",names = ["Datetime","Open","High","Low","Close","action"])Uses 'action' . and yet your dataframe has 'Signal'
dataframe.head()Out[58]: Datetime Open High Low Close Signal 0 2011-06-13 11:00:00 705.56 705.76 703.67 704.41 0.0 1 2011-06-13 12:00:00 704.25 704.52 701.51 701.69 1.0This may not be relevant, my familiarity with pandas loading csv with backtrader is not high enough that I can diagnose just from source, I'd need to run, look at errors in realtime & play around with it to figure it out.
Send the .csv file (or a shortened version) and should be easy to fix
-
@Richard-O-Regan : For lasted version. I changed a column name to "Signal". I get same error.
Thank you
csv file pic and data
Datetime Open High Low Close Signal0 2011-06-13 11:00:00 705.56 705.76 703.67 704.41 0
1 2011-06-13 12:00:00 704.25 704.52 701.51 701.69 0
2 2011-06-13 14:00:00 704.69 709.42 704.39 706.12 0
3 2011-06-13 15:00:00 705.86 708.25 705.68 707.52 0
4 2011-06-13 16:00:00 707.37 709.8 706.28 708.76 0
5 2011-06-14 09:00:00 712.97 718.68 712.87 715.43 0
6 2011-06-14 10:00:00 715.56 715.97 712.11 712.49 0
7 2011-06-14 11:00:00 712.65 716.43 710.17 715.85 0
8 2011-06-14 12:00:00 715.96 716.51 711.77 713.37 0
9 2011-06-14 14:00:00 710.2 713.07 710.17 712.67 0
10 2011-06-14 15:00:00 712.84 714.79 710.87 714.64 0
11 2011-06-14 16:00:00 714.79 724.12 712.5 724.12 1
12 2011-06-15 09:00:00 724.29 724.33 718.51 719.03 0
13 2011-06-15 10:00:00 719.03 721.46 718.51 720.83 0
14 2011-06-15 11:00:00 720.83 723.87 718.51 723.35 0
15 2011-06-15 12:00:00 723.51 723.77 720.77 720.95 0
16 2011-06-15 14:00:00 718.66 722.06 718.51 720.82 0
17 2011-06-15 15:00:00 720.82 722.82 718.81 722.5 0and coding
import pandas as pd from backtrader.feeds import PandasData from backtrader.feeds import GenericCSVData class PandasData_Signal(PandasData): # Add a 'pe' line to the inherited ones from the base class lines = ('Signal',) # add the parameter to the parameters inherited from the base class params = (('Signal', 8),) import backtrader as bt class MLSignal(bt.Strategy): def log(self, txt, dt=None): dt = dt or self.datas[0].datetime.datetime(0) print('%s,%s' % (dt.isoformat(),txt)) pass def __init__(self): # Keep a reference to the "close" line in the data[0] dataseries self.dataclose = self.datas[0].close self.datasignal = self.datas[0].Signal self.order = None def next(self): self.log(' Close, %.2f' % self.dataclose[0]) self.log(' Signal, %.2f' % self.datasignal[0]) if self.order: return if self.datasignal[0] == 1.0: self.log('BUY CREATE, %.2f' % self.dataclose[0]) self.order = self.buy() if self.datasignal[0] == 2.0: self.log('SELL CREATE, %.2f' % self.dataclose[0]) self.order = self.sell() cerebro = bt.Cerebro() dataframe.to_csv("/Volumes/SANDISK/Algo_Trader_and_Robot_ProjectTraining/Class 2/Ex6/model/SET50_H1_2011_2017_Model1_Predicted_reformat.csv") dataframe['Datetime'] = pd.to_datetime(dataframe['Datetime'],format='%Y-%m-%d %H:%M:%S') print(dataframe.head()) data = PandasData_Signal(dataname=dataframe, #dtformat = ('%Y-%m-%d %H:%M:%S'), datetime=0, open=1, high=2, low=3, close=4, #volume=5, #openinterest=-1 Signal=5 ) # Set our desired cash start cerebro.broker.setcash(1000000.0) cerebro.adddata(data) # Set our strategy cerebro.addstrategy(MLSignal) cerebro.run() print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue()) print('Final value is %.2f times the initial investment'%(cerebro.broker.getvalue()/1000000.0)) cerebro.plot()
-
Seeing that
open,closeand the others are not written in the code with capital letters, you may trysignalinPandasData_Signal
-
hi @Ta-Tum, the parameters of PandasData are programmed as below:
('datetime', 0), ('open', 1), ('high', 2), ('low', 3), ('close', 4), ('volume', 5), ('openinterest', 6),May be you fit ur dataframe as follow:
df = pd.DataFrame(index = dataframe .index)
df['Open'] = dataframe ['Open'].values
df['High'] = dataframe ['High'].values
df['Low'] = dataframe ['Low'].values
df['Close'] = dataframe ['Close'].values
df['Volume'] = dataframe ['Volume'].values
df['openinterest'] = dataframe ['Signal'].valuesThen at class MLSignal:
class MLSignal(bt.SignalStrategy):
def log(self, txt, dt=None): pass def __init__(self): # Keep a reference to the "close" line in the data[0] dataseries self.dataclose = self.datas[0].close self.datasignal = self.datas[0].openinterest def next(self): self.log(' Close, %.2f' % self.dataclose[0]) self.log(' Signal, %.2f' % self.datasignal[0] if self.order: return if self.action[0] == 1.0: self.log('BUY CREATE, %.2f' % self.dataclose[0]) self.order = self.buy() if self.action[0] == 2.0: self.log('SELL CREATE, %.2f' % self.dataclose[0]) self.order = self.sell()I think this should pass dataframe data signal into the strategy part.
Hope this helps. I've been looking for this solution for quite long too. ;) Thanks
