Why do you think the arrow can't hit the bow when it goes out?

After the arrow shaft is released, it will bend under the push of the bowstring and its own inertia. When the bow and arrow are well matched, the flying state of the arrow shaft is a bit like a snake at first (of course, it is not twisted so much or bent so much), and this bending allows the arrow to bypass the bow handle. The estimation shows that when the bow rebounds, the effect of avoidance can be achieved as long as the handle of the bow is translated 1 cm and the bow is rotated less than 5 degrees. The large-scale rotation we see is actually the product of inertial motion, and it is not really an effective bow. What really works is that the bow rotates very little in the short time of rebound, accompanied by the translation of the handle. Because the motion range is very small and the time is very short, you can't see it unless you shoot it with a high-speed camera. However, the rotation we see can reflect whether the output is correct.