Please find an example chart, we wanted to be available to select AUTOMATIC (without need to use the filters) only the top (10,5, etc.) lines to show in this view.
In this specific examples the criteria to rank, It should be the average of this lines over the time, or, MAX value, etc...

The RANK function currently only allows you to rank individual records by their value within a column (optionally a subset of the data). Whereas *you* need to apply the RANK after aggregating the data, so you are ranking each time series (a set of multiple records) against other time series.

There is a good workaround, however. See attached IOK file.

Here I have created a sample set of 4 time series identified by field "Cat" (A, B, C, D).
"Val" is my Y axis, and "Time" is my X axis. Each time series has 4 points. I hope this corresponds accurately to your data structure.

I have a field "One" containing 1 in every cell, purely so I can use the new RUNNINGTOTAL function to determine the index of each record within its category ("Per-cat index").

I separately find the max value in each time series ("Per-cat max").

Finally, I use RANK to find the position of the time series against all other time series. I pass in SUBSET([Per-cat index], 1) so each time series only counts as one record for ranking effectiveness.

In this file, I'm ranking to show the top 3 out of 4. You should be able to adapt all this for your file perfectly.

Ideally, in future, we would have additional functions that would facilitate this kind of problem. Ideas welcome - please post here. For example, instead of using the "One" field and a RUNNINGTOTAL field, we could add an optional "subset" argument to CURRENTROW, yielding the record number within a subset.

Apologies - there appears to be a problem with attachments in this page. I'll post the file later.

Meanwhile, here are the formulae:
"Per-cat index": RUNNINGTOTAL([One], SUBSET([Cat]))
"Per-cat max": SUBSET_MAX([Val], SUBSET([Cat]))
"Rank": RANK([Per-cat max], [Per-cat max], SUBSET([Per-cat index], 1))

Here is the file Steve is referring to.