There is not even a formula. They just put the rough number for each value in the sourcecode on the server. You see the pure values parsefloat(x