We flipped a 500 mL bottle 10 times with 125 mL in the bottle. We flipped it the same way each time. We then recorded the data and repeated it with 250, 375, and 500 mL in the bottle.
My claim is that the closer to ¾ of the way full, the more it will land successfully.
My evidence is (on average) with 125 mL it landed 1 time, with 250 mL, 2 times, with 375 mL, 2 times, and with 500 mL, 0 times.
I think this happened because there was enough water to weigh it down, but not having to much water and being too bouncy.