improve fallback logic for deriving reward addresses

improve fallback logic for deriving reward addresses #68