We did use actual data from existing lines to measure the gap between our models and reality. We were +/- 5% accurate.