This code runs on Python version 3.8.2 in 32bit

Polls / Estimates Data

We need to gather our economic estimates data to calculate the surprise seen in the metric of interest (e.g.: US Inflation) straight after the release of a report (e.g.: US Inflation Report). Since Refinitiv's DataStream Web Services (DSWS) allows for access to the most accurate and wholesome economic database (DB), naturally it is more than appropriate. We can access DSWS via the Python library "DatastreamDSWS" that can be installed simply by using pip installpip install.

Make sure to entre correct DataStream credentials

We need to find out exactly when (date and time) the last report was released. In our example, we looked at CPI in the US. This is reported retroactively (i.e.: when a press release with CPI figures are published, they inform readers of what the CPI was in past dates). What we were interested in up to now were the datapoints themselves; but now we'll look at FX changes at time of press release. For the sake of this example, let's say it was last released on 11-May-2022 at 13:34. That's what the TREL1 is there for.

DSWS returns dataframes with MultiIndex columns.

MultiIndex([('USCONPRCE', 'X'),
    ('USCONPRCE', 'TREL1'),
    ('USCONPRCE', 'DREL1')],
    names=['Instrument', 'Field'])

While this is extremely useful for data analytics, it'll better for us to flatten it to a single dimention

Now let's tidy our data up in teh reportActual data-frame:

Let's create a function that does that for us

Let's plot our data to visualise it; we can use plotly with the pd.options.plotting.backend function:

Tick Data

In the cell bellow, we import the Python library os, then we use it to point to the path where our Refinitiv Credentials are; we need this file in order to authenticate ourselves to the Refinitiv data services and collect data. You can find out more on this here, and a copy of the Configuration file here.

The last report was on 2022-06-10T12:36:30 and we'll get FX data till 2022-06-10T12:51:30

2149 rows × 9 columns

2149 rows × 4 columns

Display Table

We don't necessarily want to show all the data calculates in our app. What about just past press release data?

For historical FX data, we will have to use the ek library and authenticate ourselves to it. This library will only work with Workspace/Eikon running in the background or in CodeBook.

We'll get data from 2022-04-12T12:36:30 till 2022-04-12T12:51:30

We'll get data from 2022-05-11T12:36:30 till 2022-05-11T12:51:30

We'll get data from 2022-06-10T12:36:30 till 2022-06-10T12:51:30

Interface

Now we will try and display our plot above the table of relevant data:

'5.8.2'

Widget

(Credit to ac24)

We can now look into incorporating our above graph in a widget that allows us to toggle between graph and data (in a table). This is a rudimentary example taken from CodeBook examples (in '__ Examples __/04. Advanced UseCases/04.01. Python Apps/EX_04_01_02__WFCH_Company_Income_Statement_Waterfall.ipynb'), but more complex ones can easily be constructed to fit your workflow.

1345 rows × 4 columns

All Together Now

pandas version 1.3.5
numpy version 1.21.2
ipywidgets version 7.6.3
plotly version 5.8.2

Kudos to Kimberly Fessel

References

You can find more detail regarding the Datastream API and related technologies for this notebook from the following resources:

• Refinitiv Data API page on the LSEG Refinitiv Developer Community web site.
• Datastream DSWS Python Tutorial Series/Article 'Estimating Monthly GDP via the Expenditure Approach and the Holt Winters Model'.

For any question related to this example or Datastream's API (DSWS), please use the Developers Community Q&A Forum.