Custom Indicator Help

SSbilis

I want to calculate a moving average and rolling standard deviations(Kinda like Bollinger Bands) but I want to calculate this on the spread that I have calculated using the returns of two stocks. I tried using indicators such as bt.SMA and such but I kept getting errors. I eventually just manually calculated the deviation bands and MA but was hoping someone could show me what is the right way to do it, because the manual results doesn't look 100% correct either.
Keep in mind this is part of an indicator class.


def next(self):
       l0_pct_change = (
                np.array(self.line0.get(ago=0, size=self.p.period_beta))
                / np.array(self.line0.get(ago=-1, size=self.p.period_beta))
                - 1
        )
       
        self.line1.idx = self.line0.idx 

        l1_pct_change = (
                np.array(self.line1.get(ago=0, size=self.p.period_beta))
                / np.array(self.line1.get(ago=-1, size=self.p.period_beta))
                - 1
        )
        model = sm.OLS(l0_pct_change, l1_pct_change)
        self.l.beta[0] = model.fit().params
        self.l.spread[0] = l0_pct_change[0] - l1_pct_change[0]*self.l.beta[0]

#These are the lines of code in question 
        datasum = math.fsum(self.l.spread.get(size=self.p.period_beta))
        self.l.spread_sma15[0] = datasum / self.p.period_beta
        self.l.spread_2sd[0] = ((((datasum-self.lines.spread_sma15)**2)/self.p.period_beta)**0.5)*2
        self.l.spread_m2sd[0] = ((((datasum - self.lines.spread_sma15) ** 2) / self.p.period_beta) ** 0.5) * -2

run-out

@ssbilis I would like to ask a clarifying question if I may. Are you trying to do the following:

Use two lines.
Calculate the one (could be more than one) period return for each line.
Create moving averages for each line and then subtract the lines from each other.
Calcualate standard deviation bands on the result from three.

Does this reflect what you are trying to do?

SSbilis

@run-out Kinda.
I want to create a pairs trading strategy.
Where I calculate the spread using the returns of the two assets. Then I want to calculate the moving average and standard deviation of the spread.
You get what I am saying?

davidavr

@ssbilis I think you could do a lot of the calculations in your __init__ method and then just access them in next() to simplify things. Something like this (if I'm understanding what you are trying to do). Note, I'm faking the beta calculation since I'm not familiar with the statsmodel library.

class St(bt.Strategy):
    def __init__(self):
        self.amdrets = (self.datas[0].close / self.datas[0].close(-1) - 1)
        self.intcrets = (self.datas[1].close / self.datas[1].close(-1) - 1)
        beta = calc_beta(self.amdrets, self.intcrets)
        self.spread = self.amdrets - self.intcrets * beta
        self.spread_sma = MovingAverageSimple(self.spread, period=15)
        self.spread_p2sd = StandardDeviation(self.spread_sma, period=15) * 2
        self.spread_m2sd = -self.spread_p2sd

    def next(self):
        print(f'amd: {self.amdrets[0]}, intc: {self.intcrets[0]}, spread: {self.spread[0]}, sma: {self.spread_sma[0]}, +2sd: {self.spread_p2sd[0]}, -2sd: {self.spread_m2sd[0]}')

SSbilis

@davidavr said in Custom Indicator Help:

    self.spread_m2sd = -self.spread_p2sd

This is actually very helpful thank you