Need help connecting screener to trading algorithm

I have created a simple stock screener using pipeline in my notebook. Now, I want to somehow integrate (if that's the right word?) into a simple moving average algorithm. My goal is to run each stock that passes through the screener through the algorithm and eventually end up with a trade.

So far I have:
context.pipe = pipeline_output('my_pipe')

def open_positions(context,data):
ma17 = data.history(context.pipe, 'price', 17, '1d').mean()
ma40 = data.history(context.pipe, 'price', 40, '1d').mean()
current_positions = context.portfolio.positions.amount
if (ma17 > ma40) and current_positions == 0:
order_target_percent(stock, 1.0)
This doesn't work whatsoever and I don't know how to find the moving average for each individual stock that passes through the screener.

Any help or sources that can help would be greatly appreciated
Also, I am very new to Quantopian so please, tread lightly.

First, welcome to Quantopian!

You have a great start. It's a good practice to test out your pipeline definition in a notebook first to ensure the data is what you expected. Moreover, define the pipeline as a standalone function so it can then be easily imported into an algo. You did both. While, in theory this works, there are a couple of things to watch out for. One is the symbols method. In notebooks one can get either a list of symbols or a single symbol by using the one symbols method. However, in an algo, one must use either the symbols method (to get a list of assets) or the symbol method (to get a single asset). So, to migrate the pipeline definition to an algo in this case, change symbols to symbol. Then it will work as expected.

I have attached an algo based on a template which I frequently use for a long only strategies. It uses python sets to easily code which groups of assets to open, hold, and close. It also uses the order_optimal_portfolio method instead of order_target_percent to place orders. This eliminates looping through each asset one wants to open or close. Additionally, one doesn't need to check if an asset can trade or not. This is really the preferred way to place market orders.

Also, it's fastest to get all the data using pipeline and not with data.history. Unless one wants minute data (which is only available using data.history) use pipeline. I made that change, along with some cosmetic changes (consistent use of 'Fundamentals' instead of also 'Morningstar' in pipeline definition).

Check out the attached algo. Reply if you have any questions.

Good luck.