The confusing point seems to be the "seasonal" calculation (peak data for the season are not as high as the input data).  Also, I'm probably using HOMER way outside it's range (I'm actually looking at State wide data, not just facility data).  But I do know the loads for FPL, and entered them for each month.  But on the seasonal (monthly data) it shows peaks that are say 30% beyond what the real values are...

In HOMER’s Primary Load Inputs window, when you specify a 24-hour load profile for January, you are specifying the average daily load profile for January.  HOMER puts 31 such days together to create a full month of January data, and then it adds in the amount of random variability that you specify.

If you specify zero variability, then every day in January will be identical to the load profile you entered.  Here’s the load data for January in your file when I set the random variability to zero:

HOMER has taken the load profile you entered for January and simply replicated it 31 times in a row.  So the January peak load is just the highest of the 24 values you specified:  13,419 kW.

But with the time-step-to-time-step variability set to 20%, as you have it set, the January data looks like this:

For each hour in January, HOMER has taken the value you entered and multiplied it by a value it drew randomly from a normal distribution with a mean of 1 and a standard deviation of 20%.  As a result, every day is unique, the peak value is higher, and the minimum value is lower, although the mean has stayed the same.  The peak load in January is now a little more than 20,000 kW.

If you want HOMER to assume that the January load data will consist of 31 days identical to the 24-hour load profile you entered for January, then just set the time-step-to-time-step random variability value to zero.