Forcite geometry optimization

Hi,

I have been doing simulations on CNT systems recently, and I use Forcite for optimization and dynamics. This query may sound naive, but I couldn't find the answer to it myself.

While performing geometry optimization, the energy of CNT system goes down from a a higher positive value to a lower one. Shouldn't the energy be negative and with optimization lead to a value further negative? I would be glad to know if my understanding isn't correct or of there is something wrong in my simulations.

Thank you.
Prathamesh