Most forces in aerodynamics are proportional to area, speed squared, air density and a 'coefficient' that describes how efficient your object is. There's also usually a 1/2 for obscure theoretical reasons.
If you work in metric, the density of air at sea level is basically 1, which takes that out of the equation.
The maximum lift coefficient of an average wing is roughly 1. The drag coefficient of a non-aerodynamic object is also (very) roughly 1 (see link for examples), so for a first approximation, we'll use 1.
Now you've got to guess the maximum speed of your model, maybe by multiplying the motor Kv, battery voltage and prop pitch.
That should give you the maximum force on the control surface. We've no idea how it's distributed but we'll assume it's in the middle. Multiply the force by half the chord to get the torque.
Then allow for differences in the length of control horns. If they're the same, the servo needs the calculated torque. If the servo horn is half as long as the control surface horn, the servo will only need half the torque.
Finally, double or triple that value to give yourself a safety factor.
Alternately, just copy a similar model.
Then when you test fly, don't try any high-speed dives until after you'd tried a level full speed pass. If the controls feel sluggish at speed, you need larger servos.