Stock Screening

backtrader

Use the link below to go the original post

Click here to see the full blog post

cwse

How do you stop the output printing:
"""- Datas:
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
- Data0:"""
for each data source added?

backtrader

You cannot with the standard WriterFile which is the one creating the output.

You can if you subclass it, check the sources, and create your own output.

cwse

Thank you. How do I subclass it, check the sources and create a different output?

backtrader

There is no other way.

In any case the expected (even if small) API of the Writer should have a small documentation page to ease up subclassing.

spyamine

Thank you for this example. can you please complete it by using this screener as input to a strategy. for instance to be long under and short the long list and close the trades if they go out of the list. thx a lot

backtrader

See this reddit post: Reddit - The Conservative Formula in Python: Quantitative Investing made Easy

The code

class St(bt.Strategy):
    params = dict(
        rperiod=1,  # period for the returns calculation, default 1 period
        vperiod=36,  # lookback period for volatility - default 36 periods
        mperiod=12,  # lookback period for momentum - default 12 periods
        torank=100,  # number of stocks to rank as top
        reserve=0.05  # 5% reserve capital
    )

    def __init__(self):
        # Return, volatility and momentum
        rets = [bt.ind.PctChange(d, period=self.p.rperiod) for d in self.datas]
        vols = [bt.ind.StdDev(ret, period=self.p.vperiod) for ret in rets]
        moms = [bt.ind.ROC(d, period=self.p.mperiod) for d in self.datas]

        # simple rank formula: (momentum * net payout) / volatility
        # The highest ranked: low vol, large momentum, large payout
        self.ranks = [d.npy * m / v for d, v, m in zip(self.datas, vols, moms)]

        # allocation perc pro stock
        # reserve kept to make sure orders are not rejected due to
        # margin. Prices are calculated when known (close), but orders can only
        # be executed next day (opening price). Price can gap upwards
        self.perctarget = (100.0 - self.p.reserve) % self.p.torank

    def next(self):
        # get the ranks sorted for each iteration in a list
        ranks = sorted(
            ((self.datas[i], rank[0]) for i, rank in enumerate(self.ranks)),
            key=lambda x: x[1],  # use rank (elem 1 in the tuple) to sort
            reverse=True,  # highest ranked 1st ... please
        )
        # Put top 100 in ordereddict with data as key to test for presence
        r100 = collections.OrderedDict(ranks[:self.p.torank])

        # Remove those no longer Top 100, prepare quick lookup dict
        positions = {pos: pos.data for pos in self.getpositions() if pos}
        for pos, data in positions.items():
            if data not in r100:  # open and no rank100 ... close
                self.order_target_percent(data, target=0.0)

        # Add the newcomers to the Top 100, prepare easy reverse lookup
        revpositions = {data: pos for pos, data in positions.items()}
        for data in (d for d in r100 if d not in revpositions):
            self.order_target_percent(data, target=self.perctarget)

spyamine

@backtrader You are awesome!! many thanks

spyamine

@backtrader Hi, I've tryed to test your code but unfortunatly it didn't work properly as the compiler do not reconignize "pos.data" from this line "{pos: pos.data for pos in self.getpositions() if pos}" and also i suspect this line too "self.order_target_percent(data, target=self.perctarget)" the self.order_target percent part. Can you please share the full code that works for you?

spyamine

from backtrader.feeds import PandasData

class PycaishenData(PandasData):

# Add a 'pe' line to the inherited ones from the base class
lines = ('turnover','mkt_cap')

# openinterest in GenericCSVData has index 7 ... add 1
# add the parameter to the parameters inherited from the base class
params = (('turnover', -1),('mkt_cap',-1))

class StAmine(bt.Strategy):
params = dict(
rperiod=1, # period for the returns calculation, default 1 period
vperiod=36, # lookback period for volatility - default 36 periods
mperiod=12, # lookback period for momentum - default 12 periods
torank=10, # number of stocks to rank as top
reserve=0.05 # 5% reserve capital
)

def __init__(self):


    # The highest ranked: mkt_cap

    self.ranks = [bt.ind.movav(d.mkt_cap,period = 1) for d in self.datas]

    # allocation perc pro stock
    # reserve kept to make sure orders are not rejected due to
    # margin. Prices are calculated when known (close), but orders can only
    # be executed next day (opening price). Price can gap upwards
    self.perctarget = (100.0 - self.p.reserve) % self.p.torank

def next(self):
    # get the ranks sorted for each iteration in a list
    ranks = sorted(
        ((self.datas[i], rank[0]) for i, rank in enumerate(self.ranks)),
        key=lambda x: x[1],  # use rank (elem 1 in the tuple) to sort
        reverse=True,  # highest ranked 1st ... please
    )
    # Put top 100 in ordereddict with data as key to test for presence
    r100 = collections.OrderedDict(ranks[:self.p.torank])

    # Remove those no longer Top 100, prepare quick lookup dict
    positions = {pos: pos.data for pos in self.getpositions().values() if pos}
    for pos, data in positions.items():
        if data not in r100:  # open and no rank100 ... close
            self.order_target_percent(data, target=0.0)

    # Add the newcomers to the Top 100, prepare easy reverse lookup
    revpositions = {data: pos for pos, data in positions.items()}
    for data in (d for d in r100 if d not in revpositions):
        self.order_target_percent(data, target=self.perctarget)

def run(tickers, datas,strategy = StAmine,plotdetails= True,writedetail = False):

cerebro = bt.Cerebro()

print ('Starting Portfolio Value: %.2f ' % cerebro.broker.getvalue())

cerebro.addstrategy(strategy)

for ticker, data in zip(tickers , datas):
    dt = PycaishenData(dataname=data)
    if plotdetails == False:
        dt.plotinfo.plot = False
    cerebro.adddata(dt,ticker)

cerebro.broker.set_fundmode(True)
cerebro.broker.set_fundstartval(10.0)  # the default is 100
cerebro.broker.setcommission(commission=0.0025)

# Add the Analyzers
cerebro.addanalyzer(btanalyser.SQN)
cerebro.addanalyzer(btanalyser.AnnualReturn)
cerebro.addanalyzer(btanalyser.SharpeRatio)

cerebro.broker.setcash(100000.0)

cerebro.addanalyzer(btanalyser.TradeAnalyzer)

if writedetail:
    cerebro.addwriter(bt.WriterFile, csv=True, rounding=4)

cerebro.run()

cerebro.plot()

print ('Final portfolio value: %.2f' % cerebro.broker.getvalue())

run(tickers, datas,strategy = StAmine,plotdetails= False,writedetail = False)

the code seems to do nothing!!

backtrader

@spyamine said in Stock Screening:

Hi, I've tryed to test your code but unfortunatly it didn't work properly as the compiler do not reconignize "pos.data" from this line "{pos: pos.data for pos in self.getpositions() if pos}" and also i suspect this line too "self.order_target_percent(data, target=self.perctarget)" the self.order_target percent part. Can you please share the full code that works for you?

It is a hand-crafted snippet, giving an indication and orientation as to how a proposed algorithm could be implemented.