Error comparing stock returns to SPY returns in pipeline

Back to Community

edited Jul 3, 2019

Hi, I'm new to Quantopian and currently working on my first algorithm. I would like to calculate of the % price change in a stock over 20 days to the % price change in the SPY ETF over the same period of days. I currently have the following pipeline code:

def make_pipeline():    

    universe = QTradableStocksUS()

    stock_returns = Returns(window_length=20) * 100 + 100  
    spy_returns = stock_returns[symbols('SPY')]  
    stock_returns_to_spy_returns = stock_returns / spy_returns * 100

    return Pipeline(columns={'stock returns': stock_returns, 'stock returns to SPY': stock_returns_to_spy_returns, 'spy returns': spy_returns},  
                   screen=universe  
                   )

The problem is that this gives me an error for the spy returns column:

Cannot add column 'spy returns' with term NumExprFactor(...)[Equity(8554 [SPY])]. Adding slices or single-column-output terms as pipeline columns is not currently supported.

Does anyone know how to fix this? Ideally I would like to divide the values by each other indexed to 100. For example, if stock XYZ returns over a 20 day period are 10% and SPY returns over the same 20 day period are 5%, the stock returns to spy column should return 110/105*100=104.76~. Any different ways to achieve this would also be appreciated!

3 responses

Casper N

Jul 3, 2019

Update on the above, I managed to get the result that I was looking for by creating a custom factor, called it RelativeStrength:

class RelativeStrength(CustomFactor):  
    inputs = [USEquityPricing.close]  
    window_length = 20  
    def compute(self, today, assets, out, close):  
        out[:] = (close[-1] / close[0] *100) / (close[1][np.where(assets==8554)] / close[0][np.where(assets==8554)] *100) *100

def make_pipeline():    

    close = [USEquityPricing.close.latest]  
    rs = RelativeStrength()  
    return Pipeline()

Unfortunately I am now getting stuck on the next step, which is to look at the % increase of this custom factor over the past 10 days, which I will call RSMomentum:

class RSMomentum(CustomFactor):  
     window_length = 10  
    def compute(self, today, assets, out, rs):  
        out[:] = rs[-1] / rs[0] * 100

def make_pipeline():    

    close = [USEquityPricing.close.latest]  
    rs = RelativeStrength()  
    mom = RSMomentum(rs)

    return Pipeline()

This doesn't seem to work though, is there a way to use custom factors as inputs of other custom factors?

Cam Howe

Aug 1, 2019

I have it comparing today's position to its performance 1m ago. You can change that adjust the larger denominator to fit your need.

class RelativeStrength(CustomFactor):  
    inputs = [USEquityPricing.close]  
    window_length = 60  
    def compute(self, today, assets, out, close):  
        out[:] = ((close[-1] / close[-21]) / (close[-1][np.where(assets==8554)] / close[-21][np.where(assets==8554)])) /((close[-21] / close[-42]) / (close[-21][np.where(assets==8554)] / close[-41][np.where(assets==8554)]))-1

Cam Howe

Aug 1, 2019

I personally like this way better which shows rate-of-change of alpha.

class RelativeStrength20(CustomFactor):  
    inputs = [USEquityPricing.close]  
    window_length = 21  
    def compute(self, today, assets, out, close):  
        out[:] = (close[-1]/close[-1][np.where(assets==8554)]) / (close[-21] / close[-21][np.where(assets==8554)]) -1

You've successfully submitted a support ticket.

Our support team will be in touch soon.