My understanding is...
If you called epsilon % more than your mixed strategy optimal calling percent, the nemesis will automatically adjust to never bluffing as now EV Bluff < EV Not Bluff and you will only be paying off value bets
If you called epsilon % less than your mixed strategy optimal calling percent your opponent will always bluff and although you will always pick him off, the villain will more than compensate by picking up the pot enough when your hand is worse than your (now smaller) calling frequency
Now of course you would observe this and adjust yourself leaving the game with oscillating pure strategies. It's been shown that this actually isn't optimal
Bill chen's book (MOP) offers god insight into this
Good luck at the tables
Hutch