Import USEquityPricing from quantopian.pipeline.data.builtin and then add close = USEquityPricing.close.latest in the pipeline.

How would you get the close price of the day prior to this?

One can get the close price of the day before yesterday (ie 2 days ago) with a custom factor something like this:

class Factor_N_Days_Ago(CustomFactor):  
    def compute(self, today, assets, out, input_factor):  
        out[:] = input_factor[0]

Then instantiate this factor using inputs = [USEquityPricing.close] and window_length = 2.

close_day_before_yesterday = Factor_N_Days_Ago(inputs = [USEquityPricing.close], window_length = 2)

Notice this custom factor can be used to get the value of any factor and any number of days ago. Simply change the input and the window_length. Take a look at this post for more discussion on the topic. There are also some notebooks to look at and run. https://www.quantopian.com/posts/get-lagged-output-of-pipeline-custom-factor

Hope that helps.

What are we seeing here?
(Especially considering the potential of off-hours trading. I think this might be Eastern 8p on 4-15 and 4-16).

''' Output:
2019-04-17 07:30 trade:17 INFO ARNC  
2019-04-16 00:00:00+00:00    20.680  
2019-04-17 00:00:00+00:00    19.985  
'''
from quantopian.pipeline  import Pipeline  
from quantopian.algorithm import attach_pipeline, pipeline_output

def initialize(context):  
    schedule_function(trade, date_rules.every_day(), time_rules.market_open(minutes=60))  
    attach_pipeline(Pipeline(), 'pipeline')

def trade(context, data):  
    out = pipeline_output('pipeline')

    for s in out.index:

        prices = data.history(s, 'price', 2, '1d')

        log.info('{} \n{}'.format(s.symbol, prices))  
        assert 0 # deliberate halt for just 1 item and quit, see Logs tab

I thought I recalled discussions regarding the most recent record from data.history with 1 day resolution like this for more than one day being the latest price during the current market day underway, once upon a time.

By the way it seems timely to talk about time a bit. I'm lucky if I know what today's date is, and time is always confusing for me, and backtesting adds an extra dimension. I look forward to a time when all times everywhere are pegged to the particular market time , in this case NYC, letting user override timezone if they wish. We have a mix of local and UTC making things more complicated to talk about with a worldwide audience. For example, checking my watch, this might be a good time for me to take the time to explain that the 7:30 you see here is 1 hour after market open in pacific time because that's where I happen to be at the moment. If we are on the move flying around and saving logs [trying to compare later], they have the potential to not be consistent, or if we are collaborating with people in Sidney and London, we're all looking at different timestamps in logs. Making them all the same is easier said than done I think, considering the possibility of also picking up Shanghai and others in the future, but chances are good that there's an engineer who knows just what to do.

If that '1d' is changed to '1m', since this is running at the start of the current minute, how can it know the close price of this minute?
It is looking into the future is it not?

> get_datetime()  
    Timestamp: 2019-04-17 14:30:00+00:00  
> prices  
    prices: Series  
    Timestamp('2019-04-17 14:29:00+0000', tz='UTC'): 20.015  
    Timestamp('2019-04-17 14:30:00+0000', tz='UTC'): 19.985  

Sorry -- I'm looking for the same solution (except 1 yr back) and I'm not able to get your code to build -- I keep getting
Runtime exception: NameError: name 'CustomFactor' is not defined

I tried plugging the inputs directly into the class input, leaving it CustomFactor as shown, nothing is working.
(defined class before Initialize in my strategy), and called it within initialize like so:

prevYr = Factor_N_Days_Ago(inputs = [USEquityPricing.close], window_length = 255)

First tried:

prevYr = Factor_N_Days_Ago([USEquityPricing.close],window_length=255)

Still nothing. Please help!

I've tried putting these in every combination of locations within my strategy -- it will not build.
I don't understand this code AT all, should this have a return or something?

Please advise exactly how this should be entered into strategy so I can get PrevYr price.

Zach

The error NameError: name 'CustomFactor' is not defined seems to imply you haven't imported the class definition for CustomFactor. Ensure there is this import statement in the code

from quantopian.pipeline import  CustomFactor

# Define our custom factor  
class Factor_N_Days_Ago(CustomFactor):  
    """  
    Returns the factor value N days ago where window_length=N  
    This is the price adjusted as of the current simulation day.  
    """  
    inputs = [USEquityPricing.close]  
    def compute(self, today, assets, out, close_price):  
        out[:] = close_price[0]

# Create an instance of our close price 252 days (~1 year) ago  
close_1_yr_ago_factor = Factor_N_Days_Ago(inputs=[USEquityPricing.close], window_length=252)

Now, if one doesn't want to use a custom factor this can also be calculated from the returns like this

from quantopian.pipeline.factors import Returns  
close_price = USEquityPricing.close.latest  
close_1_yr_ago_calculated = close_price / (Returns(window_length=252) + 1.0)  

Both these approaches will return the close price 252 trading days ago adjusted as of the current simulation day. See the attached notebook. I'll also post an algo which uses this.

Notebook for above...

Algo using Factor_N_Days_Ago custom factor...

This is SO much simpler haha and I cannot believe I was missing an import — sorry about that.
I assume that’s what’s missing, let me give it a try !
Thanks so much :)