In previous experiments, I had felt a more regular pull on the low radius loop than for long figure-eight trajectories, leading to equal or almost equal power with less traction peaks.
Losses due to relatively large figure-eight trajectories can be explained by the fast variations of traction force within the flight window. The position of the kite is always changing in relation to the central line of the flight window. This leads to increasingly greater losses as the kite moves away from that central line.
In addition to these losses, there are losses due to the irregularities themselves. For example, if you drive a car for an hour, covering 75 km, you will consume much less energy if you drive constantly at 75 km/h (by analogy with a regular traction force of a static parachute kite) than if you accelerate and decelerate (by analogy with an irregular traction force of a crosswind kite) constantly for the same average speed of 75 km/h.
In addition to this, the angle of attack is optimized to favor the L/D ratio of a crosswind kite, while the thrust coefficient is favored for a static parachute kite.
This seems to explain why I get only 3 times the average traction force for the crosswind kite compared to that of the static parachute kite. In reeling mode, the power obtained would also be 3 times higher. Indeed: