A forecast horizon is the length of time for which forecasts are generated. Forecaster calculates the forecast horizon based on your historical data and your selected algorithm. 

Historical data prompts Forecaster to set the time intervals that apply to the forecast horizon. Forecaster calculates if forecast predictions are weekly, monthly, or yearly.   

Some algorithms predict for the same length as the historical data timeline. Others only predict for a third of the timeline. For the former, even if the algorithm can predict a forecast horizon of 12 months, the forecast is capped based on historical data.  

Horizon calculations 

Based on historical data and the selected algorithm, Forecaster can generate predictions weekly for a total period of six months. The total period is the forecast horizon.  

Forecaster tracks the number of data points you provide. For all algorithms (except Prophet and MVLR), no matter what the algorithm limits are, or the amount of historical data, only 500 data points are forecast.  

Horizon limits

If the frequency is daily, Forecaster doesn't forecast more than 500 days.  

If the frequency is weekly or monthly, then the limit of 500 may not be an issue. Keep this limitation in mind.

Shorter forecast horizon 

Even though Forecaster supports long-term forecasts, a best practice is set your horizon for the shortest horizon that meets your business needs. A shorter horizon improves accuracy and data quality. We recommend you set your forecast horizon to cover half the time frame included in your historical data.  

Example: For 24 months of history, we recommend you forecast 12 months ahead.