AVERAGEX with FILTER is very slow - running/cumulative total

[BIdeveloperDK]

Dear PBI Oracles,

I have a table consisting of some items named cItem with four belonging measures. I'm trying to calculate a running average of mAdjustPickRatio in a measure named mCumMeanAdjustPickRatio - based on their rank of the measure mRankAvgPickSize.

As I'm calculating an average of an existing average measure (mAdjustPickRatio), I'm sadly forced to use AVERAGEX. As demonstrated with the red box, for row/rank 11, I want the running average of all items less or equal to rank 11. 

However, the problem is that using AVERAGEX or SUMX in combination with FILTER, is very slow (two minutes or so), although I only have a thousand rows. I've tried following two measures:

mCumMeanAdjustPickRatio = 
VAR MyData = SUMMARIZE(ALLSELECTED(ITEM_DIM[cItem]);ITEM_DIM[cItem];"AvgPick";[mAvgPickSize];"AdjustPick";[mAdjustPickRatio];"Rank";[mRankAvgPickSize])
VAR CurrentRank = [mRankAvgPickSize]

RETURN
AVERAGEX(
    FILTER(MyData;[Rank]<=CurrentRank)
    ;[AdjustPick]
)

And: 

mCumMeanAdjustPickRatio2 = 
VAR CurrentRank = [mRankAvgPickSize]
RETURN
AVERAGEX(
    CALCULATETABLE(
        FILTER(
            ALLSELECTED(ITEM_DIM[cItem]);[mRankAvgPickSize]<=CurrentRank))
            ;[mAdjustPickRatio]
)

Both really slow.
The issue with AVERAGEX and SUMX, in combination with FILTER, is that is loops over the same data, and thus calculate a new table for each row, although I tried to make a constant table with VAR in example 1.

How can I fix this insane slow formula?

[v-chuncz-msft]

@BIdeveloperDK,

You may take a look at https://www.sqlbi.com/topics/optimization/.