Well for 1 the reverse / reaction / anti torque of the motor plays a huge role in the issue you are facing here.
You seem to be using a single motor. Usually in fixed wing UAVs where only 1 electric motor is used at the nose a slight down torque angle and slight left torque angle (both between 1° and 5°) are introduced to counter act this. You can check it up online - there are numerous RC enthusiast blogs explaining this in detail.
I haven't done adequate testing for coanda effect craft to tell you an exact remidy yet. But the above method like in fixed wing is a quick and easy way to reach a solution. I'm fairly certain that playing around with the torque angles should give you an immediate fix.