@Grant Glad you highlighted this! The Morningstar stock_type field is definitely underused. It's a great way to group and filter as companies with similar types may tend to behave similarly. What makes this field more interesting than morningstar_sector_code is that it isn't as correlated with the 'sector' exposure factor. Anyway, attached is a notebook which uses stock_type to group the results of Alphalens.

Hope that helps.

Thanks Dan -

So how would I do a stock filter/mask/screen thingy in pipeline?

For example, would this work:

stock_type = Fundamentals.stock_type.latest

stock_type_picks = (stock_type.eq('Distressed') | stock_type.eq('Cyclicals'))  

This would get me Distressed and Cyclicals, for example, correct?

Yes, that should work (sort of... use the stock_type integer codes rather than the names). So, like this

# Let's first get the  stock_type classifier  
stock_type = Fundamentals.stock_type.latest

# Make some factor to analyze later. Here it's what I call classic momentum  
momentum = Returns(window_length=152).log1p() - Returns(window_length=10).log1p()

# Filter for Distressed and Cyclicals  
stock_type_picks = (stock_type.eq(3) | stock_type.eq(4))  

my_pipe = Pipeline(  
    columns={'stock_type': stock_type,  
             'momentum': momentum,  
             'sector': Fundamentals.morningstar_sector_code.latest,  
             'score': score,  
            },  
    screen= QTradableStocksUS() & stock_type_picks  
)

Thanks Dan -

In my effort to develop algos for the recent $10K Third-Party Challenge: Design a Factor for a Large US Corporate Pension, I've become interested in ways to identify companies that are kinda sketchy from a financial viability standpoint. I've used financial_health_grade and the Factset zscoreso far, and figured that one or more of the stock_type classifications might be useful, too (e.g. Distressed).