Here is one possibility:
Data = [100,200;
300,100;
400, 100
300, 100];
Data = [Data; 100,200; 100,200; Data; 100,200; 100,200; 100,200;];
TF = diff(Data,[],2)<0;
state = [diff([0; TF])>0 diff([0; TF])<0 diff([0; TF])==0];
The ‘TF’ vector is ‘true’ =1 when the second column element in a particular row is less the first. The ‘state’ matrix is such that when ‘TF’ goes from 0 to 1 the corresponding element in the first column is 1, when ‘TF’ goes from 1 to 0 the second column element is 1, and when there is no change, the element in the third column is 1.
Combine these to define your ‘streak’ variable, since I am not certain exactly how you want to do that. Note that the series of 1 values in the third column is actually one element shorter than the actual streak of sequential values in ‘TF’.
The leading zero values in the diff arguments in my ‘state’ matrix are to make the indices correspond to those in ‘TF’.
If you want to see how it all works, combine ‘TF’ and ‘state’ into one matrix (I call it ‘L’ here):
L = [TF state];
