Resampling Elevation Profile

I have a general question about resampling. I'm aware that resampling with too small of an interval can create noise and lead to an overestimation of gain/loss. Does anyone have an interval that they use that seems to reflect the actual gain/loss in the field? I'm working on an 11.5 mile relatively flat segment in Glacier National Park (Kintla Lake Trailhead to Upper Kintla Lake). At 100' intervals, the net gain is around 1,200'. When I reduce to 25' intervals, the gain jumps to almost 1,900'. Any suggestions for how to determine how small an interval is too small?