You are not very specific in your details but I suspect you are describing Hugh Piggott's 4ft alternator.
I assume this only has one magnet rotor is you say 8 magnets, but you could use a spinning disc to close the flux path and that would change the cut in speed very drastically.
If what you describe is in fact Hugh's alternator then yes you can calculate back to get cut in speed if you can make an inspired guess at the average flux density seen by the coils. This is very dependent on the distance from magnet rotor and stator.
With typical clearance you would have a cut in speed of about 300 rpm. With metric magnets 46 x 30 x 10 mm I got a typical average flux density of 250mT. The imperial magnets would not be much different.
As for torque input then take your output watts, assume 60% efficiency to get input watts and then use your speed to get torque.
How the speed will relate to output load will depend on the flux density and the winding resistance. You can calculate all of this once you have measured cut in speed and deduced your actual flux density.
If it is any help then for your 20W out you would be looking at close on 650 rpm for a typical rotor/stator clearance.
Does that help if not clarify the question some more.
Flux