This behavior is likely due to how Bloomberg handles future dates and the interpretation of the 'nil_value' parameter in the history function from MATLAB's ‘Datafeed’ Toolbox. The period argument is expected to control how Bloomberg treats non-trading days or missing data, but the 'nil_value' option doesn't seem to be correctly interpreted in this context.
When requesting data for future dates (e.g., '31-Dec-2024'), Bloomberg may return the last available value, especially if 'non_trading_weekdays' is used without a clearly supported 'nil_value' equivalent. This can lead to unexpected repeats of values instead of NaN (as expected when no data is available).
Instead use 'non_trading_weekdays','nil' instead of 'nil_value', or filter the future dates manually after retrieving data.
Please refer to this sample MATLAB implementation:
c = blp;
c.DataReturnFormat = 'table';
myAsset = 'WBA UW Equity';
currency = 'EUR';
% Define time range
T = (datetime('31/12/1999', 'InputFormat', 'dd/MM/yyyy'):calyears:datetime('31/12/2024', 'InputFormat', 'dd/MM/yyyy'))';
% Use correct period settings
period = {'yearly', 'non_trading_weekdays', 'nil'}; % Try 'nil' instead of 'nil_value'
% Request data
Data = history(c, myAsset, 'SUSTAIN_GROWTH_RT', T(1), T(end), period, currency);
Data.DATE = datetime(Data.DATE, "ConvertFrom", "datenum");
% Post-process: Replace future values with NaN
today = datetime('today');
Data.Value(Data.DATE > today) = NaN;
save('Data', 'Data')
You can also refer to the following MathWorks documentation for more details:
- MATLAB documentation for Bloomberg historical data requests - https://www.mathworks.com/help/datafeed/bloomberg.history.html
- Datafeed Toolbox Documentation - https://www.mathworks.com/help/datafeed/index.html
Hope it helps!
