What date to use as transition for a company fundamental value

I have the same problem as discussed already in this thread https://community.developers.refinitiv.com/questions/28994/sample-period-from-eikon.html

Current date is 24th of June 2018. For Company A whose fiscal year end is on 30th of June, the FY0 will show the data for 2017 as 2017 is the last fiscal year. For company B whose fiscal year end is 31st of March, the FY0 will show the data for 2018 as 2018 is the last fiscal year. So FY0 is not showing the same year for different companies in this situation.

I want to fetch and analyse historically fundamental values of a company for last 3 years . Each company has its own release date for the fundamental values, relative period FY0, FY-1 etc or FY2019, FY2018 are not working for me. For some companies Eikon return FY0 for others it returns FY-1 data

Below is how I fetch the data

for yr in [0,-1,-2]:
    _df, err = ek.get_data(LIST_OF_ASSETS,
                          ['TR.EPSACTVALUE(Period=FY%s)'%yr,
                          'TR.EPSMeanEstimate(Period=FY%s)'%yr,
                          'TR.Ebit(Period=FY%s)'%yr, 
                          'TR.CompanyMarketCap(SDate=%s)'%(datetime.datetime.now()-pd.Timedelta(days=365*(1-yr))).date().strftime('%Y%m%d'),
                          'TR.NetDebt(Period=FY%s)'%yr, 
                          'TR.TotalLongTermDebt(Period=FY%s)'%yr, 
                          'TR.CommShareholdersEqty(Period=FY%s)'%yr, 
                          'TR.NumberofSharesOutstandingMean(Period=FY%s)'%yr], {'Curn':'USD'})
    _df = _df.set_index('Instrument')

I tried various dates for an asset

Query used for above df

df, e = ek.get_data(LIST_OF_ASSETS, ['TR.NetDebt.date',
                             'TR.NetDebt.fperiod',
                             'TR.NetDebt.value',
                             'TR.AnnEPSDate',
                             'TR.ISOriginalAnnouncementDate',
                             'TR.ISPeriodEndDate',
                             'TR.ISPeriodEndDate.fpa',
                             'tr.revenue',
                             'tr.revenue.date',
                             'tr.revenue.fperiod',
                             'TR.ISOriginalAnnouncementDate', 'TR.ISStatementLastUpdatedDate','TR.ISSOURCEDATE'],
                    {'Period': 'FY0', 'SDate': 0, 'EDate': -10, 'Frq': 'FY', 'Curn': 'USD'}, field_name=True)

If you see in the dataframe above, ISSOURCE DATE and essentially many others dont even change for FY-1, FY-2.. Its just random..

My List of Assets compromise of companies around the world.

Which date is the most appropriate to use which marks the transition from FY(X) to FY(X-1) for a company? Because i see for each company this date is different and I cannot backtest the results I have obtained in the past.

UPDATE

@chavalit thanks for your reply but u missed my main point here.

I use data i pull from Eikon to make reports, so I need to be able to fetch data historical data from Eikon consistently. My reports are based on a collection of assets from across globe that do not follow the same financial cycle start and end dates.. Its different in USA, its different in India, its different in Europe.

Example

Current date is 19/03/2020. Company A whose fiscal year end is on 31/03/2020. Company B whose fiscal year end is 31/12/2019

If I queried FY0, FY-1,FY-2 on 4 different dates 30/12/2019, 01/01/2020, 30/03/2020, 01/04/2020

I will get different values for FY0, FY-1, FY-2

I hope you are understanding my problem here

If on 04/04/2020 I want to pull the data I obtained on 30/12/2019, which I used to prepare report in the past (on 30/12/2019) to backtest. How do I do that ??

That is why i need to know the date on which each company transition its fiscal year. Infact Eikon starts reflecting the change in fundamental already as soon as a company publishes its annual reports. So i am looking for that date..

Regarding your reply on ISSOURCEDATE, the ISSOURCEDATE for the first 3 years is the same.. How can the income statement source date for 2018 or 2019 be 2020?