Point In Time Data Sets

Pete Millington

Pete Millington·– November 23rd, 2021

Advantage

Point in time financial data sets have been built so that users can access historical financial statements including a date of when the financial statements were known. Non-point in time data sets do not include a date of when the information became available. Researchers using non-point in time data sets use a lag number of periods, e.g., 15 business days, after a quarter end to adjust their available financial statement information for an assumed amount of time for the information to become available. In Quotient, Scientific Financial labels this ‘known’ date as the vantage date. The vantage date can be either the original report date or the date that the financial statement was recast or restated. Companies adjust their prior period financial statement for a wide variety of reasons including complying with new accounting standards.

An example

In the example table below from Quotient, Reuters fundamental point in time data set shows that Northrop Grumman originally reported net income for their 12/31/2018 quarter of $657M as of a vantage date of 1/31/2019. On the 1/31/2020 vantage date, Reuters fundamental point in time data set, shows that Northrop Grumman recast their net income for their 12/31/2018 quarter to $356M.

Table from Quotient, Reuters fundamental point in time data set

Ease of access

Quotient, the modern quantitative modeling tool available from Scientific Financial Systems, makes switching from standard non-point in time data sets to point in time data sets seamless for factor construction. Quotient has been built on a Python platform. Factor construction uses this python platform.

In the example below, we simply change the data source one time in the factor from the non-point in time financial statements data set, in this example Returns fundamentals, to the point in time financial statements' data sets, in this example Returns fundamentals point in time. After changing the data source, all data items retrieved from Reuters will be from the Reuters fundamental point in time data set. In the Python code example below, the first line is commented out. 

Many other quantitative modeling tools require the factor builder user to switch to point in time data at the individual data item level. With complex factors that have numerous financial statement items, changing every raw data item can be tedious and subject to errors.

 

Quotient factor definition for quarterly net income:

# reuters_fnd = root_lib['Reuters Fnd (CFT)'] #non point in time financial statements

reuters_fnd = root_lib['Reuters Fnd (PIT CFT)'] # point in time financial statements

ni_lq = reuters_fnd['NI_FQ']