The point is not that both balance and KOPS are related to gravity, though balance is intimately dependent on gravity and KOPS is not. KOPS points out the geometry and physics that the leg creates its maximum torque (driving) with the crank when the foot is farthest from the BB axis, and this happens when the crank arm and the knee to pedal path are perpendicular. At that point the maximum fraction of the force from the leg is converted into torque. Some is still lost to angular crank and BB flexing, but the jury may be out on whether that energy storage is loss or still available for propulsion, but that’s another topic. The fallacy of KOPS is that this happens once per revolution regardless of whether the bike is vertical or recumbent. So what I see from KOPS is that at all bicycle angles from upright to ‘bent there is a point of perpendicular drive, and that it can be close to a balance position in an upright bike. My balance position is usually with my knee a few cm behind KOPS, so it is not perfectly satisfied by KOPS. But my balance position is important not because of drive efficiency but to enable a position which supports my body in several positions which I like: riding on hoods, drops, and hooks, and with my hands or butt only lightly connected to the bicycle. This happens when my body CG is somewhere behind the BB plumb line. If I ever lose weight again, my riding position will probably move forward by a few cm.

An additional complication is the angle between the femur and the knee-to-pedal path near the maximum torque point (I say it this way because we don’t know what point on the knee is the actual pivot, and how to consider the offset of the foot). The optimum femur angle at the point of maximum chainset torque is a good question, and I think it’s significant for knee health.