Identifying turning points in data

I assume the highlighted cells mark your choices for the turnaround point. Can you explain how you made these choices? Part of programming formulas is to understand the logic you want to use and "translate" that into a sequence of Excel formulas.

My thought, though it comes to different conclusions is:

1) Some kind of "slope" or "change" calculation. Perhaps in B8 put =B1-A1 (note relative references) then copy across and down.
2) Analyze each row for the "smallest" (most negative) change. =MIN(B8:J8) -- maybe in L8
3) A MATCH() function to locate that smallest change in the row. M8 -- =MATCH(L8,B8:J8,0)

This algorithm comes to different conclusions for some rows, but it also illustrates the overall algorithm I would use. Change step 1 to whatever calculation makes sense for determining "turn around", then adapt the other steps to make sense of that calculation.

Correct, the logic is different because I am not sure I understand the logic you want to use.

row 1 and row 6 could be done using a simple MIN()/MAX() logic L1=MATCH(MIIN(A1:J1),A1:J1,0) L6=MATCH(MAX(A6:J6),A6:J6,0), but that simple logic does not work for the other rows.

Ultimately, the answer to the "how do I return a column number" is usually a MATCH() function of some kind. The difficult part of this is figuring out the logic for the "lookup_value", then figuring out how to work that logic into the spreadsheet.

To help illustrate my confusion:
why did you pick column F (2.75) in row 2 instead of column B (3.17)?
in row 3, why column F instead of column D?
In row 5, why the minimum at H-I instead of the minimum at E-F and why not the maximum at D?

Sorry, I've been away for a bit.

One common thing I see in your latest -- you appear to be taking the "most recent" (rightmost) peak/trough. Excel's lookup functions prefer to search from left to right rather than from right to left, so the first thing I would probably do is to rearrange the data so that the most recent is on the left and the oldest to the right. Then, a slope or change calculation like I suggested in post #2, followed by something to detect the first time that the change or slope changes from positive to negative (for a peak) or from negative to positive (for a trough) should capture the most recent region where the data turnaround. After creating a copy of the data in inverse order (using an INDEX() function), I calculated a change, then used something like =SIGN(PRODUCT(B17:C17)) to detect where the changes in sign were. Then a MATCH() function [=MATCH(-1,B17:J17,0)] would return the position of the peak trough.

Row 4 is still a bit of a challenge because the data on that row do not change very often. Because the change on that row is 0 most of the time, That row could probably be detected by searching for the first non-zero change.

Is that getting closer?