Any plane will snap out of a loop if the wing is forced deep enough into stall by the elevators, even with perfect lateral balance, and exactly equal incidence and airfoils on each wing. At stall, any wing becomes dynamically unstable in roll, because an increase in angle of attack produces less lift. Once a tiny amount of roll begins, the downgoing wing has a higher induced angle of attack than the upgoing wing, and thus develops less lift and more drag than its partner wing, so the process rapidly accelerates. Use of wing washout will help, but with enough excess elevator travel, even the washout-equipped wing will stall deeply enough to make things get pretty murky.

The best way that I know of to beat this problem is to set maximum elevator travel so that it is just adequate to barely stall the wing. I like to test for correct elevator travel by taking the model safely high, and looping it at full throttle with full "up" elevator. If it flies only a partial loop before careening wildly in yaw and roll, it has too much elevator travel. If it continues to fly smooth consecutive loops, it probably has too little elevator travel. The ideal that I have found is to keep fine-tuning elevator travel until you get at between about three-quarters and one full loop before the model yaws and rolls drunkenly. That way, if you pull full "up" elevator to recover from a vertical dive at low altitude, the model will turn at close to its minimum possible radius, without snapping over on its back and re-kitting itself. I have seen a lot of models that were set up with roughly double the elevator travel that is needed - ideal for 3D flying, but producing a model that has to be flown very carefully to avoid low altitude disasters.