We do something similar to look for memory leaks, but the idea is to use 'getRRDValue' to retrieve the value of the datapoint averaged over some period of time and then using that value as the theshold.
For example, given a datapoint "OldGen_used", we set a Maximum threshold of:
here.getRRDValue('OldGen_used', 'end-7d', None, 'AVERAGE') * 1.25
so this means the threshold is breached if the datapoint is 125% greater than it has been over the last week.
We also do this to compare between datapoints. For example, given a datpoint "Heap_Memory_Used" and "Heap_Memory_Max", we define the max threshold as :
here.getRRDValue('Heap_Memory_max') * .9
so that the threshold is breached when the "used" is greater than 90% of Max.
Hope this helps,
Joseph