to close this question:

import backtrader as bt
import numpy as np
import datetime

class A(bt.Indicator):
    lines = ('avg',)
    def __init__(self):
        self.addminperiod(3)
    def next(self):
        self.lines.avg[0] = np.mean(self.datas[0].close.get(size=3).tolist())


class B(bt.Indicator):
    lines = ('avg',)
    def __init__(self):
        self.addminperiod(5)
    def next(self):
        self.lines.avg[0] = np.mean(self.datas[0].close.get(size=5).tolist())

class C(bt.Indicator):
    lines =('xxx',)
    def __init__(self):
        pass
    def next(self):
        print([a-b for a,b in zip(self.datas[0].avg.get(size=3).tolist(), self.datas[1].avg.get(size=3).tolist())])

class TestStrategy(bt.Strategy):
    def __init__(self):
        self.avg3 = A(self.datas[0])
        self.avg5 = B(self.datas[0])
        self.delta = C(self.avg3, self.avg5)

    def next(self):
        pass

cerebro = bt.Cerebro()
cerebro.addstrategy(TestStrategy)
data = bt.feeds.YahooFinanceCSVData(
    dataname='orcl-1995-2014.txt',
    fromdate=datetime.datetime(2000, 1, 1),
    todate=datetime.datetime(2000, 1, 31),
    reverse=False)
cerebro.adddata(data)
cerebro.run()

It seems the code below doesn't work

i = int(f)

Many thanks @tracy , @backtrader

Say if I have a pool of 300-500 stocks, I would like to sort out top 10 stocks by ranking algo, and compute the portfolio and then test win/loss. Do we have example of this?

In some cases I need to set different status in a line.
For example

s1 = indicator > 0.2
s2 = indicator < -0.2
signal = s1 - s2 

the value of signal will be

 [1, nan, nan, -1, -1, nan, -1, 1, nan, nan...]

I would like to know how to avoid nan in this case?

Today I use the code below

signal = bt.Max(0, 0)
singal = b - s

I wonder is there an efficient way to initialize signal with value 0?

BTW, how to initialize a line with 0?

thanks! so fast! :-)

Say if I try to check, trix.trix cross trix.signal and trix.trix < 0, How to write it in an efficient way.
I know I can use CrossOver(trix.trix, trix.signal), but how to add another condition, trix < 0

thanks, I will try

Does backtrader support multi-process? I suppose this shall be supported. I see backtrader use multi-process when do optimizing, but I don't find how to use multi-process examples. I am asking for this because I want to scan on stocks. The strategy is same for all, but need different data feeds.

I try to use Zigzag function to check W-bottom, M-top. A quick question, it seems you are setting indicator value by future data. For example, at day 11, you detected day 10 is valley-pivot and mark day 10 as valley.