screener query in python with a large number of identifiers

asingh

Hi, I have a similar query. My screener expression in excel to get the tickers is TR("SCREEN(U(IN(Equity(active,public,primary))/UNV:Public/), IN(TR.HQCountryCode,""AR,BR,CL,CN,HU,IN,ID,MY,MX,PL,RU,ZA,KR,TH,TR""), NOT_IN(TR.GICSIndustryCode,""401010"",""401020"",""402010"",""402020"",""402030"",""402040"",""403010""), CURN=USD)","TR.CommonName;TR.HeadquartersCountry;TR"&".GICSSector","curn=USD RH=In CH=Fd")

From the above formula, I get a list of tickers in Column A from A2:A14186. Now I get the Cash data for these tickers like this: TR($A$2:$A$14186,"TR.CashandEquivalents(CURN=USD Scale=6) ","Period=FQ0 Frq=FQ SDate=0d EDate=2006-12-31 sortd=periodenddate CH=periodenddate NULL=blank RH=IN")

I want to understand how to replicate this query in python

Also running the above query in excel together, I get a restriction of identifiers> 7500 ,

so I have to run it in small batches in excel.

pierre.faurel

The replicated request in Python:

import eikon as ek
ek.set_app_key(<yout app_key>)
rics = ('SCREEN(U(IN(Equity(active,public,primary))/*UNV:Public*/),'
    'IN(TR.HQCountryCode,"AR,BR,CL,CN,HU,IN,ID,MY,MX,PL,RU,ZA,KR,TH,TR"),'
    'NOT_IN(TR.GICSIndustryCode,"401010","401020","402010",'
    '"402020","402030","402040","403010"))')
fields = ["TR.CommonName", "TR.HeadquartersCountry", "TR.GICSSector"]
data, error = ek.get_data(rics, fields) list2 = list(data['Instrument'])
field2 = ["TR.CashandEquivalents.periodenddate","TR.CashandEquivalents"] 

As list2 contains more than 14,000 items, single request will fail with Payload Too Large error. You need to iterate through the list 10 items at a time.

for i in range(1, len(list2), 10):
    data2, error = ek.get_data(list2[(i-1)*10:i*10], field2,
        parameters={"SDate": "0", "EDate": "-9", "Scale": "6",
        "Period": "FY0", "Frq": "FQ", "CURN": "USD"})
    print(data2) 

asingh

@pierre.faurel..Thanks for sending this.

I tried the above code, but i did not get any entry in data. I tried with one country but the result is same. It gives me no entries in the dataframe.

It gives me a KeyError: 'Instrument' as there is no Instrument column.

Also I believe tmp is the len of the dataframe

Alex Putkov.1

You're right, there were a few errors in the example script including a missing parenthesis in the screener expression, which resulted in no instruments being retrieved. And instead of the tmp variable it should indeed have been list2. I just corrected the errors in the answer. Try it and see if it works for you.

asingh

@Alex Putkov. @pierre.faurel

Hi Alex,

Thanks for the reply this works. Just had to make a small change in this line:

list2[(i-1)*10:i*10]

where it should not be multiplied by 10 and you have to make some indexing modification here

I have another doubt, here I am querying data for "TR.CashandEquivalents". Instead of this, I want to query for

"TR.CashandEquivalents/TR.InterestExpense". I make the changes in the field2 list as:

field2 = ["TR.CashandEquivalents/TR.InterestExpense.periodenddate","TR.CashandEquivalents/TR.InterestExpense"]

Now for testing, I run for small number (around 10) of rics, I find that I do not get any NaNs or blanks for the date i.e TR.CASHANDEQUIVALENTS/TR.INTERESTEXPENSE.periodenddate.

Is the above field2 setting correct.

asingh

@Alex Putkov. @pierre.faurel

Hi Alex,

Thanks for the reply this works. Just had to make a small change in this line:

list2[(i-1)*10:i*10]

where it should not be multiplied by 10 and you have to make some indexing modification here

I have another doubt, here I am querying data for "TR.CashandEquivalents". Instead of this, I want to query for

"TR.CashandEquivalents/TR.InterestExpense". I make the changes in the field2 list as:

field2 = ["TR.CashandEquivalents/TR.InterestExpense.periodenddate","TR.CashandEquivalents/TR.InterestExpense"]

Now for testing, I run for small number (around 10) of rics, I find that I do not get any NaNs or blanks for the date i.e TR.CASHANDEQUIVALENTS/TR.INTERESTEXPENSE.periodenddate.

Is the above field2 setting correct.

asingh

@Alex Putkov. @pierre.faurel

Thanks for the reply this works. Just had to make a small change in this line:

list2[(i-1)*10:i*10]

I have another doubt, I want to query for

"TR.CashandEquivalents/TR.InterestExpense". I make the changes in the field2 list as:

field2 = ["TR.CashandEquivalents/TR.InterestExpense.periodenddate","TR.CashandEquivalents/TR.InterestExpense"]

Now for testing, I run for small number (around 10) of rics, I find that I do not get any NaNs or blanks for the date i.e TR.CASHANDEQUIVALENTS/TR.INTERESTEXPENSE.periodenddate.

Is the above field2 setting correct.

asingh

@Alex Putkov.

Hi, I would like to get a similar result for the field "TR.CashandEquivalents/TR.InterestExpense" but I am getting NaN values.As a result I cannot pivot the final resulting dataframe.

Alex Putkov.1

I'm not sure I understand the question. Below is an example of the call that returns data for a small list of stocks. For your list you should expect to get a lot of NaNs because for this ratio to be returned both TR.CashandEquivalents and TR.InterestExpense must be available on quarterly report, which will not be true in many cases.

ek.get_data(["000416.SZ","ERW.BK","AFC.BK"], 
    ["TR.CashandEquivalents.periodenddate",
    "TR.CashandEquivalents/TR.InterestExpense"], 
    parameters={"SDate": "0", "EDate": "-9", "Scale": "6",
    "Period": "FY0", "Frq": "FQ", "CURN": "USD"})

asingh

@Alex Putkov. Thanks, so we will give the periodenddate for only one metric i.e "TR.CashandEquivalents.periodenddate" in the formula. I was using earlier "TR.CashandEquivalents/TR.InterestExpense.periodenddate"

Alex Putkov.1

You can use either TR.CashandEquivalents.periodenddate or TR.InterestExpense.periodenddate, but the period end date needs to be listed as a separate field from the ratio. In the expression "TR.CashandEquivalents/TR.InterestExpense.periodenddate" you're asking the system to divide the value of TR.CashandEquivalents by the period end date, which doesn't make sense and returns an error.