Align a vector such that all of its derivatives are positive or negative

1. I have a unidimensional time series vector as such:

originalVector: http://imageshack.us/f/651/originaltimeseries.jpg/

2. I want to decompose the series into two vectors, one of the points with positive slopes and the other of points with negative slopes.

I can do this by using the commands:

positiveDer = originalVector(diff(originalVector)>0) http://imageshack.us/f/18/positiveder.jpg/

negativeDer = originalVector(diff(originalVector)<0) http://imageshack.us/f/407/negativeder.jpg/

3. As you can see both new vectors are characterized by "runs" of the majority derivative occasionally broken by the minority derivative. I would like to align the runs in such a way that they feed into one another such that the series is composed entirely of one derivative valence, the positive or negative 'majority" derivative.

I think this could be done first by splitting the positiveDer into subvectors into positive "runs" whenever a negative derivative is encountered, producing runs of different lengths: run_t, run_tPlus1, run_tPlus2,...

To align the the runs, I could subtract the first point of a run from the end point of the previous run: ans = run_tPlus1(1) - run_t(end) and then add the difference: run_tPlus1 + ans

This would need to be looped to align all the runs, and then all the subvectors must be concatenated back into one vector.

Please, how could I do this?

Thank you, Alex