Hello Darth.
This might work.

topByMomentum = results.groupby('sector')['momentum'].nlargest(2)  

When placed inside 'before_trading_start', this will create a dataframe from your pipeline output containing the symbols that have the largest 2 momentums for each sector. nsmallest(2) would get you the 2 smallest momentums.
Hope that helps.

Hello Bryan,
Thank you for your input. I tried you suggestion but get an error "Expected assets argument to be of type or iterable of type Asset". I also tried this:

ranktop = results.sort_values(by='combo_rank', ascending=True).iloc[:1000]
topbymomentum = ranktop.sort_values(by='momentum', ascending=True)
topbysector = topbymomentum.groupby('sector')
topofall = topbysector.iloc[:context.num_long]
context.longs = topofall

This brings up the error "Cannot access callable attribute 'iloc' of 'DataFrameGroupBy' objects, try using the 'apply' method". Ultimately I'd like to take the 1000 stocks after ranking them on value, then be able to sort by momentum, whereby if my final long count goal was 20 stocks for example I'd have at most 3 stocks from any sector.

Do you have any other ideas?

Darth,
Sorry, but it's getting late and I'm leaving in the morning to take my boys to the gulf coast for a couple of weeks.
This will give you the top three momentums by sector. Maybe ditch the iloc 1000 idea?
Best of luck.

Bryan,
Had a look at your suggestion and this is what I came up with. I'm getting a new error. Any suggestions?

Hey Darth,

That error is being thrown from your for loop code block beginning at line 178. If you 'print(security)' from within that block you'll notice that 'security' contains (101, Equity(1595 [CLF])). I'm guessing 101 is the sector.
By inserting 'security = security[1]' as a new line at 179 the symbol will be stripped out and work as expected.
Like this:

for security in context.longs.index:  
        security = security[1]  
        if get_open_orders(security):  
            continue  
        #print(security, weight)  
        if data.can_trade(security):  
            if (security in context.portfolio.positions) or (can_buy):  
                order_target_percent(security, weight)  
                equity_weight += weight  

Your other ordering sections don't seem to be affected by this, but you might want to check to make sure.
When looping through lists of lists remember that they are zero based indices. security[0] is the first element, security[1] is the 2nd and so on.
You might already know that, but thought I'd throw it in just in case.
Hope that helps.
Bryan

Bryan,
I want to thank you for your help, the suggestions you have made have fixed the problem. Everything is working well thanks to you!