forgive me if i appear to write "down"
native page is used for charge/size separation. gel (and crosslinker) percentage is adjusted to make the gel more or less restrictive.
4-20% should allow most proteins to migrate, at least part way, through the gel. only the largest proteins may not enter the gel.
charge on the protein is influenced by the pH of the buffers used in the gel and the electrode (most important).
the electrode buffer for the biorad system is pH 8.3. that is very close to your proteins of interest, their charges will small so they won't migrate well through the gel (as you have seen). just reversing electrodes will not help, in fact, it will make matters worse. the proteins will still have a net negative charge so will still migrate to the positive electrode (out of the gel).
three things you can do:
1) run for a longer time and/or higher current (careful not to let the system get too hot), until the proteins migrate far enough to suit you (not the best idea but will work, if you don't mind other proteins running out of the gel, although with 20% at the bottom many proteins will just bunch up).
2) switch to an acid pH gel. then you can switch electrodes and get proper migration (you will, however, have to cast your own gel). i think i posted the buffer formulations somewhere in these fora, you can find them with a search.
3) raise the pH of the electrode buffer. you can use the ornstein-davis buffer formulation for the electrode buffer (this should also be available within these fora, i think i posted it also). this will give you a sort of hybrid protocol. the o-d electrode buffer runs at pH 9.5. this should solve your problem and resolve your proteins.