Permutation t-test'ed Coherence NxN Fc graph

I'm not sure what kind of file you are referring to. If this is the example file you sent me: this is the result of a t-test, ie. a comparison between two conditions. The t-statistic is somewhat similar to a difference of averages, so it is normal that you obtain positive and negative values.
https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics

Still some "positive"-value (yellow-coloured) links appear in the Functional Connectivity map when we move the Intensity Treshold -slider inside the negative area of the parameter.

If you put this slider at -4 in a range of -10 to +10, it would remove anything below -4, therefore it is normal to see both positive and negative values.

I don't recommend you use this threshold slider for filtering the graph when display the result of statistical tests. Use only the "stat" tab and adjust the p-value threshold and correction of multiple comparisons instead. This is what really tells you what is significantly different in your difference.