Do the zscore screening first and then apply the weights, .6 and -.4 using the len of those that made it through.

Would that have to be done through pipeline as well? Or could I just write another function that handles the z-score screening after I assign longs and shorts based on sentiment and then finally order my securities?

Both ok. I pictured a route like this, guard against margin though, at a minimum, record cash.

def rebalance(context, data):  
    #long_weight, short_weight = compute_weights(context, data)  
    context.long_secs  = context.output[context.output['longs']].index  
    context.short_secs = context.output[context.output['shorts']].index  
    longs = {}  
    shrts = {}

    # zscore screen, both at same time  
    for stock in context.long_secs.union(context.short_secs):  
        zscore = calculate_zscore(stock ,data)  
        if   zscore < -1.65:  
            # The value of zscore here so far is unused, just a way of storing these in dict instead of list  
            longs[stock] = zscore  
            log.info(' long {}: {}'.format(stock.symbol, '%.4f' % zscore)) # to 4 decimal places  
        elif zscore > 1.65:  
            shrts[stock] = zscore  
            log.info('short {}: {}'.format(stock.symbol, '%.4f' % zscore)) # to 4 decimal places

    # Place orders for longs  
    for stock in longs:  
        order_target_percent(stock, .6 / len(longs))  
        log.info(' long ' + stock.symbol)

    # Place orders for shorts  
    for stock in shrts:  
        order_target_percent(stock, -.4 / len(shrts))  
        log.info('short ' + stock.symbol)