Trying to gain insight into the problem by reading might in this case be more effective than a manual MC search of parameter space...

You are right that reading is the first thing to do to solve a problem. However, my problem is not how to run a Monte Carlo simulation. I'm trying to understand why there is a discrepancy between the total energy computed by the MC module and the output of the standard ENERGY routine (reported in the Delta-E column).

But I like your suggestion to increase the temperature to improve sampling. Thanks!