Good catch the downtube bottle data makes me suspect this data even more. I know it has been shown that the deltas between different frames can change significantly when there’s a dummy on top due to frame+person interactions, and cervelo has stated that they’ve optimized their bikes specifically with a person on top. .
That’s what we found when we did the Aero Shootout. The P5x was the 2nd or 3rd fastest bike only yet with a rider on was clearly above the rest of the pack.