Fiscal calendars

The fiscal calendar of a company specifies the accounting periods the company uses when publishing their financial reports. A fiscal calendar specifies when the fiscal years start and end, and how they are subdivided into shorter reporting periods.

Time series are often aligned to the fiscal periods of companies. Since companies can use different kinds of fiscal calendars, and it happens that a company changes its fiscal calendar, it can be challenging to work with such time series. The Exabel DSL therefore has extensive support for handling such time series.

This section gives some information about company calendars. If you are more interested in the actual usage of fiscal periods in the DSL, you can look at the following:

Calendar types

The most common choice is to follow the normal calendar, such that the fiscal year ends 31 December, subdivided into four three-month quarters or two six-month semi-annual periods. However, many companies choose to deviate from this.

Fiscal years come in two main varieties:

Calendar type


12 months

Each fiscal year consists of twelve months and ends on the last day of a
specific month.

52/53 weeks

Each fiscal year consists of 52 or 53 weeks, ending on the same weekday.
The last day of the fiscal year is either

  • the last day with a specific weekday in a specific month, or

  • the day with a specific weekday closest to the last day in a specific month.

In both cases, December is the most common month to end the year in, but any month is possible. This may also depend on the country a company is located in: in India and Japan most companies end their financial year 31 March. Likewise, for the 52/53-week calendar, fiscal years commonly end on Saturday or Sunday, but again, any weekday is possible.

These fiscal years are then subdivided into either quarters or semi-annual periods.

For the 12-month calendars, the most common choice is to subdivide the year by calendar months. This means that a fiscal year is either divided into four quarters of three calendar months apiece, or into two semi-annual periods of six calendar months apiece. However, some companies choose to end the intermediate fiscal periods (Q1, Q2 and Q3, or H1) on a specific weekday, rather than follow the calendar months.

For 52/53-week calendars that are subdivided into quarters, the most common is to let each quarter consist of thirteen weeks in a 52-week year and let one of them be extended to a 53-week year whenever necessary. Some companies choose instead to have one long quarter of sixteen weeks and three short ones of twelve weeks each in a 52-week year. In a 53-week year, one of these quarters must be lengthened, which means that the longest fiscal quarter in common usage consists of seventeen weeks.

Similarly, for 52/53-week calendars that are subdivided semi-annually, most companies choose to have 26 weeks in each semi-annual period, with one of them lengthened by a week in 53-week calendars.

Identifying the fiscal calendar of a company

There is no structured dataset which offers descriptions of various companies’ fiscal calendars. We instead infer the rules of the fiscal calendars from historical FactSet data. This is done by identifying the fiscal calendar which best describes the dates of previous fiscal periods. Unfortunately, this is complicated by the fact that companies in the course of their lifetimes sometimes switch between calendars. A company may switch between quarterly and semi-annual reporting, and they may switch between using a month-based calendar and a 52/53-week calendar.

Correctly identifying the fiscal calendar a company uses, allows us to predict when the company’s future fiscal periods end, and it makes it possible to identify dates when a time series may have missing values. In addition, it tells us whether a company primarily reports quarterly or semi-annually, which determines the frequency used when a signal is called with the FQ/FS frequency argument, or similar.

The calendar identification algorithm is not infallible. Here are some situations where it may fail to correctly identify the calendar:

  • The company has changed their fiscal calendar, but we do not have enough data points from the “new” calendar to be able to identify which it is.

  • A company is relatively new and there are multiple common fiscal calendars which could fit the observed dates. In such cases we choose the calendar in most common use among other companies, which will sometimes be incorrect.

  • The calendar uses an uncommon calendar which we do not recognize.

  • There are erroneous dates in the time series we use as input.

  • The company has previously changed their calendar, and while we are able to identify the calendars in use both before and after the transition, there are missing data points in the period when the company is transitioning, which means that the fiscal calendar is not complete during that period.

The consequences of such an error may be one of the following:

  • We are not able to extrapolate the calendar into the future.

  • We are able to extrapolate the calendar into the future, but some of the future dates are incorrect.

  • There are historical fiscal periods whose start and end points we are not able to pinpoint.