Python: Eikon is generating unrequested dates for some ID's, using get_data.

aquilesjlp300

I have the following code that requests local returns for a list of companies.

import pandas as pd
import eikon as ek #import Eikon module
import datetime
import time

start_time=time.time()
ek.set_app_id('someId') #setting AppID
#Getting a list of dates
start=datetime.date(2017,11,27)
end=datetime.date.today() - datetime.timedelta(days=1)
row_dates=[x.strftime('%m/%d/%Y') for x in pd.bdate_range(start,end).tolist()]
#getting identifiers to be used on Eikon
csv_data=pd.read_csv('identifiers_test.csv', header=None)
identifiers=csv_data[0].tolist()

This is a list with 2500 identifiers or so (all SEDOLS).

To request the returns, I do this:

df=ek.get_data(identifiers,["TR.TotalReturn.Date","TR.TotalReturn"], {'SDate': row_dates[0], 'EDate': row_dates[len(row_dates)-1], 'Frq':'D'})[0]

To pivot the table, and re-arrange it in a way that I can have dates as column headers and identifiers as indexes without being repeated, I do this:

df=pd.crosstab(df.Instrument, df.Date,values=df['Total Return'], aggfunc='mean')

But the outcome is really strange, it places columns with dates that I didn't even requested, filled with NaN's, and I would like to know how to filter that info. I already tried a couple of approaches with dropna() and other Pandas functions but I can't seem to get them to work , I'll attach 2 photos so you can see what I'm talking about).

And it properly shows dates I've requested, like this (shown 28-Nov, 29-Nov and 30-Nov. Not displaying in the picture but appeared properly in the dataframe, are 27-Nov and 01-Dec).

How can I get rid of those dates? Obviously these dates were produced by Eikon's API, somehow. I printed the original dataframe into a CSV file, I should have only 5 dates, I obviously found more. I can't upload the screencapture for this since this forum doesn't allow more than 2 image uploads.

Why Eikon does this? Is there a way to ellude it? A way to fix it? I'd like to know.

Thanks in advance!

Zhenya Kovalyov

Ok, I was able to get the list of the instruments, that are generating incorrect dates, you can use the following code:

instruments = {}


for ix, row in df.iterrows():
    instruments.setdefault(row[1], set()).add(row[0])

instrument

A quick manual check shows that the incorrect date is when the asset traded last, for example:

B2Q1MC6 (PGD.P) - 5th of September
B4TDMX4 (SBV.P) - 19th of September

Etc.

So, when you are requesting the latest Total return value, what it can give you is the last possible estimate.

Jirapongse

Can you share the list of 2500 identifiers? I have used you code with some instruments and it works fine. I am using Eikon 0.1.9.

Zhenya Kovalyov

I can not reproduce the issue by requesting the same data for S&P 500 constituents, so there might be an issue with one of the instruments in your list

aquilesjlp300

files.zipjirapongse.phuriphanvichai, here I uploaded two files. The first one has all the identifiers I use for this code. And the other file is where I saw which SEDOL's are giving me issues.

Is there a way to code an exception for these type of errors?

aquilesjlp300

Yes, I'm guessing that as well. But why? I mean, how does Eikon handle errors with identifiers? Is there a way to detect/handle that kind of error and delete it from the outcome?

aquilesjlp300

Do you know a way to code a 'filter' for these unrequested dates? Since they are very specific, I didn't find a 'conventional way' that works.

e.g: I cannot use dropna() pandas function because, even though these dates are mostly filled with NA's, a single return value is not a NaN, so I can't use that solution.

Zhenya Kovalyov

you can drop the rows that have the dates that are not equal to the requested range with the standard pandas function

aquilesjlp300

I tried another approach that worked:

added this after the crosstab line:

df=df[row_dates]

Before doing that, I had to convert the Date column to Datetime using:

df['Date']=pd.to_datetime(df['Date']) 
df['Date'] = df['Date'].dt.strftime('%m/%d/%Y')