Back to Community
How can I get adjusted prices (minute resolution) in research notebooks?

While building a machine learning algorithm you need to have the same kind of data for training and testing. I want to train the model in a research notebook, save model parameters and apply them to an algo. Problem is that when I get to the algo itself, the data is different from research data (get_pricing) - that's because of Q2 split and dividends adjustments. So how do I get minute resolution split and dividends adjusted prices inside the research? Is that impossible?

3 responses

You mean like this?

get_pricing(symbols, start_date='2013-01-03', end_date='2014-01-03', frequency='minute')

Petter, in that case the data you get is different from the IDE data. So it does not help with this problem.

@Rodrigo Still having issues?

First off, both the IDE (algo) and the research environments query the same database for info so the data IS the same between the environments.

One needs to be aware of how prices and volumes are adjusted for splits and dividends. When using 'get_pricing' in the research environment, the data is adjusted as of the end date. So, in the example above, all data will be adjusted as of 2014-01-03. When using 'data.history or 'pipeline' in an algo all data is adjusted as of the current simulation date. So, the only time they will definitely match (in this case) is when running the simulation on 2014-01-03.

There are several posts here in the forums about how others have trained ML both in an algo and in research. I can find those posts and point you to them if it would help.