Main Content


Convert to semiannual

tosemi is not recommended. Use convert2semiannual instead.


newfts = tosemi(oldfts)
newfts = tosemi(oldfts,'ParameterName',ParameterValue, ...)



Financial time series object.


newfts = tosemi(oldfts) converts a financial time series of any frequency to a semiannual frequency. The default semiannual days are the last business day of June and December. tosemi uses holidays.m to determine valid trading days.


If oldfts contains time-of-day information, newfts displays the time-of-day as 00:00 for those days that did not previously exist in oldfts.

Empty ([ ]) passed as inputs for parameter pair values for tosemi triggers the use of the defaults.

newfts = tosemi(oldfts,'ParameterName',ParameterValue, ...) accepts parameter name/parameter value pairs as input, as specified in the following table.

Parameter Name

Parameter Value




Returns the cumulative sum of the values within each semiannual period. Data for missing dates are given the value 0.



Returns the exact value at the end-of-period date. No data manipulation occurs.



(Default) Returns the values located at the end-of-period date. If there is missing data, Nearest returns the nearest data point preceding the end-of-period date.



Returns an averaged semiannual value that only takes into account dates with data (non-NaN) within each semiannual period.



This mode is compatible with previous versions of this function (Version 2.1.x and earlier). It returns an averaged end-of-period value using a previous tosemi algorithm. This algorithm takes into account all dates and data. For dates that do not contain any data, the data is assumed to be 0.


If you set CalcMethod to v21x, settings for all the following parameter name/parameter value pairs are not supported.



Generates a financial time series that ranges from (or between) the first date to the last date in oldfts (includes NYSE nonbusiness days and holidays).



(Default) Generates a financial time series that ranges from the first date to the last date in oldfts (excludes NYSE nonbusiness days and holidays and weekends based on AltHolidays and Weekend). If an end-of-quarter date falls on a nonbusiness day or NYSE holiday, returns the last business day of the quarter.

NYSE market closures, holidays, and weekends are observed if AltHolidays and Weekend are not supplied or empty ([]).



(Default) Returns all semiannual dates between the start and end dates of oldfts. Some dates may be disregarded if BusDays = 1.


The default is to create a time series with every date at the specified periodicity, which is with DateFilter = Absolute. If you use DateFilter = Relative, the endpoint effects do not apply since only your data defines which dates appear in the output time series object.



Returns only semiannual dates that exist in oldfts. Some dates may be disregarded if BusDays = 1.



(Default) The end-of-period date is the last day (or last business day) of the semiannual period.


1 - 31

Specifies a particular end-of-period day. Months that do not contain the specified end-of-period day return the last day of the semiannual period instead (for example, ED = 31 does not exist for February).


1 - 12

End month of the first semiannual period. All subsequent period dates are based on this month. The default end-of-period months are June (6) and December (12).


[Begin, End]

Denotes the minimum number of days that constitute an odd semiannual period at the endpoints of the time series (before the first whole period and after the last whole period).

Begin and End must be -1 or any positive integer greater than or equal to 0.

A single value input for EndPtTol is the same as specifying that single value for Begin and End.

-1   Do not include odd period dates and data in calculations.

0    (Default) Include all odd period dates and data in calculations.

n   Number of days (any positive integer) that constitute an odd period. If there are insufficient days for a complete semiannual period, the odd period dates and data are ignored.

The following diagram is a general depiction of the factors involved in the determination of endpoints for this function.



Returns only the observation that occurs at the first (earliest) time for a specific date.



(Default) Returns only the observation that occurs at the last (latest) time for a specific date.



Vector of dates specifying an alternate set of market closure dates.



Excludes all holidays.



Vector of length 7 containing 0's and 1's. The value 1 indicates a weekend day. The first element of this vector corresponds to Sunday. For example, when Saturday and Sunday are weekend days (default) then Weekend = [1 0 0 0 0 0 1].


collapse all

This example shows how to transform a time series object from weekly to semiannual values.

Load the data from the file predict_ret_data.mat and then use the fints function to create a time series object with a weekly frequency.

load predict_ret_data.mat
x0 = fints(expdates, expdata, {'Metric'}, 'w', 'Index')
Warning: FINTS is not recommended. Use TIMETABLE instead. For more information, see <a href="matlab:web(fullfile(docroot, 'finance/convert-from-fints-to-timetables.html'))">Convert Financial Time Series Objects (fints) to Timetables</a>.
x0 = 
    desc:  Index
    freq:  Weekly (2)

    {'dates:  (53)'}    {'Metric:  (53)'}
    {'01-Jan-1999' }    {[      97.8872]}
    {'08-Jan-1999' }    {[      97.0847]}
    {'15-Jan-1999' }    {[     109.6312]}
    {'22-Jan-1999' }    {[     105.5743]}
    {'29-Jan-1999' }    {[     108.4028]}
    {'05-Feb-1999' }    {[     134.4882]}
    {'12-Feb-1999' }    {[     117.5581]}
    {'19-Feb-1999' }    {[     106.6683]}
    {'26-Feb-1999' }    {[     118.2912]}
    {'05-Mar-1999' }    {[     105.6835]}
    {'12-Mar-1999' }    {[     128.5836]}
    {'19-Mar-1999' }    {[     115.1746]}
    {'26-Mar-1999' }    {[     131.2854]}
    {'02-Apr-1999' }    {[     130.7116]}
    {'09-Apr-1999' }    {[     123.1684]}
    {'16-Apr-1999' }    {[     107.2975]}
    {'23-Apr-1999' }    {[      91.5625]}
    {'30-Apr-1999' }    {[      78.5738]}
    {'07-May-1999' }    {[      65.2904]}
    {'14-May-1999' }    {[      70.8581]}
    {'21-May-1999' }    {[      72.4807]}
    {'28-May-1999' }    {[      72.9190]}
    {'04-Jun-1999' }    {[      64.3460]}
    {'11-Jun-1999' }    {[      59.8743]}
    {'18-Jun-1999' }    {[      55.0026]}
    {'25-Jun-1999' }    {[      49.4032]}
    {'02-Jul-1999' }    {[      49.9485]}
    {'09-Jul-1999' }    {[      47.8061]}
    {'16-Jul-1999' }    {[      61.0517]}
    {'23-Jul-1999' }    {[      58.9313]}
    {'30-Jul-1999' }    {[      53.9584]}
    {'06-Aug-1999' }    {[      44.8472]}
    {'13-Aug-1999' }    {[      45.0463]}
    {'20-Aug-1999' }    {[      45.1088]}
    {'27-Aug-1999' }    {[      56.4897]}
    {'03-Sep-1999' }    {[      61.2449]}
    {'10-Sep-1999' }    {[      58.1012]}
    {'17-Sep-1999' }    {[      50.8974]}
    {'24-Sep-1999' }    {[      46.5143]}
    {'01-Oct-1999' }    {[      38.0806]}
    {'08-Oct-1999' }    {[      33.6664]}
    {'15-Oct-1999' }    {[      34.2992]}
    {'22-Oct-1999' }    {[      33.4202]}
    {'29-Oct-1999' }    {[      36.9287]}
    {'05-Nov-1999' }    {[      35.1278]}
    {'12-Nov-1999' }    {[      41.8128]}
    {'19-Nov-1999' }    {[      35.8199]}
    {'26-Nov-1999' }    {[      36.9495]}
    {'03-Dec-1999' }    {[      36.2880]}
    {'10-Dec-1999' }    {[      33.8457]}
    {'17-Dec-1999' }    {[      33.3868]}
    {'24-Dec-1999' }    {[      32.7737]}
    {'31-Dec-1999' }    {[      28.5665]}

Use tosemi to obtain the semiannual aggregate for the x0 times series.

x1 = tosemi(x0)
Warning: FINTS is not recommended. Use convert2semiannual instead.
x1 = 
    desc:  TOSEMI: Index
    freq:  Semiannual (5)

    {'dates:  (2)'}    {'Metric:  (2)'}
    {'30-Jun-1999'}    {[     49.4032]}
    {'31-Dec-1999'}    {[     28.5665]}

Version History

Introduced before R2006a