Sumifs in power query

Replaced with slightly faster version.

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("XcyxDcAgDAXRXVwj2XyMUycZw2L/NQiJpQDd6RXnTiclymBllFFVhFpyula24HthSHBMct58XMAwho6q0+XnMl8ePj7WbRJcZZ0E28utAw==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [ID = _t, #"week start date" = _t, #"Forecast value" = _t]),
    #"Changed Type" = Table.TransformColumnTypes(Source,{{"ID", type text}, {"week start date", type date}, {"Forecast value", Int64.Type}}),
    sumifs =
        (tbl as table) =>

            Table.AddColumn(
                tbl, 
                "12 week forecast", 
                each List.Sum(Table.SelectRows(
                    tbl,
                    (r) => r[week start date] > [week start date]
                    and r[week start date] <= Date.AddDays([week start date],84)
                )[Forecast value]
                )
            )
        ,
    group1 = Table.Group(#"Changed Type", {"ID", "week start date"}, {"Forecast value", each List.Sum([Forecast value])}),
    // first group by ID and dates then by just id.  you could skip this grouping if there you do not have multiple rows with the same ID and date.
    group2 = 
        let
            buffer = Table.Buffer(group1)
        in
            Table.Group( buffer, "ID", {"sumifs", each sumifs(_)}),
    #"Removed Other Columns" = Table.SelectColumns(group2,{"sumifs"}),
    #"Expanded sumifs" = Table.ExpandTableColumn(#"Removed Other Columns", "sumifs", {"ID", "week start date", "Forecast value", "12 week forecast"}, {"ID", "week start date", "Forecast value", "12 week forecast"})
in
    #"Expanded sumifs"

hi, Anonymous a little bit complicated but I tried to speed things up. Replace your_table with your table or last step name. 

    // this function calculates adds "weeks fc" column as running total for upcoming number of weeks
    wks_forecast = (tbl, weeks) =>
        [rows = List.Buffer(Table.ToRecords(tbl)),
        values = List.Buffer(tbl[Forecast value]),
        gena = List.Generate(
            () => 
                [i = 0, 
                fc = try List.Sum(List.Range(values, 1, weeks)) otherwise 0, 
                rec = rows{i} & [weeks fc = fc]],
            (x) => rows{x[i]}? <> null,
            (x) => 
                [i = x[i] + 1, 
                fc = x[fc] - values{i} + (values{i + weeks}? ?? 0), 
                rec = rows{i} & [weeks fc = fc]],
            (x) => x[rec]
        ),
        t = Table.FromRecords(gena)][t],
    // here we group by ID ans run our function
    g = Table.Group(your_table, "ID", {"fcast", each wks_forecast(Table.Sort(_, "week start date"), 12)}),
    // combine things together
    z = Table.Combine(g[fcast])