Hi Sabir, bt.DivByZero is what you're looking for. Here's an example where I use it

class WeighInverseVol(bt.Strategy):
    params = (
        ('period', 30),
    )

    def __init__(self):
        rs = [bt.ind.PctChange(d, period=1) for d in self.datas]
        vs = [bt.ind.StdDev(ret, period=self.params.period) for ret in rs]
        inv = [bt.DivByZero(bt.LineNum(1), v) for v in vs]
        invsum = bt.LineNum(0)
        for inv_ in inv:
            invsum += inv_
        # self.weights is an array of asset allocation weights
        # one weight per asset in self.datas
        self.weights = [inv_ / invsum for inv_ in inv]

Here's the source code for bt.DivByZero: https://github.com/mementum/backtrader/blob/master/backtrader/functions.py#L43

@backtrader since you're in the thread, is it possible to modify the official backtrader repo to give some more people write access (probably just the admins of backtrader2 for now)?

(I'm assuming that continuing the official backtrader repo with a new team of maintainers is preferable to forking it but I could be wrong about that)

I have two strategies that I'd like to run sequentially. The first strategy allocates a portion of the portfolio between its assets (using self.order_target_percent) and the second strategy uses what remains of the portfolio.

What is the best way in backtrader to achieve the above?

Hi Sabir, bt.DivByZero is what you're looking for. Here's an example where I use it

class WeighInverseVol(bt.Strategy):
    params = (
        ('period', 30),
    )

    def __init__(self):
        rs = [bt.ind.PctChange(d, period=1) for d in self.datas]
        vs = [bt.ind.StdDev(ret, period=self.params.period) for ret in rs]
        inv = [bt.DivByZero(bt.LineNum(1), v) for v in vs]
        invsum = bt.LineNum(0)
        for inv_ in inv:
            invsum += inv_
        # self.weights is an array of asset allocation weights
        # one weight per asset in self.datas
        self.weights = [inv_ / invsum for inv_ in inv]

Here's the source code for bt.DivByZero: https://github.com/mementum/backtrader/blob/master/backtrader/functions.py#L43

Has anyone here implemented an indicator that returns weights for assets in a portfolio? Here is the Strategy version:

class WeighInverseVol(bt.Strategy):

    def __init__(self):
        rs = [bt.ind.PctChange(d.close, period=1) for d in self.datas]
        vs = [bt.ind.StdDev(ret, period=30) for ret in rs]
        inv = [bt.DivByZero(bt.LineNum(1), v) for v in vs]
        invsum = bt.LineNum(0)
        for inv_ in inv:
            invsum += inv_
        # self.weights is an array of asset allocation weights
        # one weight per asset in self.datas
        self.weights = [inv_ / invsum for inv_ in inv]

I'd like WeighInverseVol to be an Indicator so I can use it within other strategies but I don't know how to return a dynamic number of lines within an indicator. Any advice?

Thanks!

I remember seeing that backtrader is used in several banks and quant firms... would any of these companies be willing to sponsor backtrader's continued maintenance? Might incentivize Daniel R to keep working on it or incentivize the new admin team.

(I'm assuming that continuing the official backtrader repo with a new team of maintainers is preferable to forking it but I could be wrong about that)

@backtrader since you're in the thread, is it possible to modify the official backtrader repo to give some more people write access (probably just the admins of backtrader2 for now)?

This PR (https://github.com/mementum/backtrader/pull/401) seems like a nice bugfix to get the ball rolling with backtrader2. It's not mine but I'll steal it if the original author takes too long :P

@backtrader The PR addressing this issue has been approved and just needs to be merged: https://github.com/mementum/backtrader/pull/401