Just a note for everybody here who just finished their CS education and is afraid of floating point numbers. If you want to implement floating point transaction values as in 2., you need to make custom implementations that handle it appropriately by using integers. You shouldn't use the inbuilt float or double types.