How this API is structured
The Allocations and Aggregate Allocations endpoints expose a sort of “giant table” view of timeseries data. The tables contain “Series”, representing some timeseries dataset the market has sent. When using the API, you request the subset of the data that you care about - for instance all Allocations pertaining to some Accounting Point, or all Aggregate Allocations pertaining to a particular area, balance responsible party or supplier. See List Allocations and List Aggregate Allocations for what filtering options are available to you. You can also list the series themselves, under List Allocation Series and List Aggregate Allocation Series.Multiple values per timestep
Within a series, there can be multiple values for each timestep! The reason for this is that markets send multiple versions of most timeseries of differing “quality” and “finality” levels. Initially you may receive coarse estimates or forecasts, moving to increasingly correct values and, in some cases, corrections arriving much later. Engrate stores all the values that arrive, and exposes them in this API such that you can see all of them. This serves two primary purposes:- It lets you train models on the data as it will be when the model runs. By only training on data that’s known by, for example, 10AM, you don’t produce misfit models.
- It allows you to make decisions on your level of trust in the data. For critical pipelines like controlling, you may choose to monitor how much data is of high quality or high finality levels, and delay running the pipelines until sufficient quality data has arrived.