Should dealers be stocking vehicles based on past ROI?

Last week I was with a colleague for whom I have a great deal of respect. He and I were discussing our respective approaches for stocking used vehicle inventory. His approach used an algorithm for stocking that is based, in part, on the ROI of past vehicle sales. In other words, if a particular type of vehicle produced a favorable ROI in the recent past, his algorithm would likely recommend it for replacement. After reflecting upon this approach, I find it to be troubling.

The basis for my concern is that I don’t believe there is an inherent ROI in any particular type of vehicle. This is because both the wholesale acquisition and retail sales prices are constantly fluctuating. There is just no way to guarantee a future ROI based on past performance under such market conditions. It would be like me believing that I should buy more AT&T stock this month because I made a good return on the AT&T stock that I bought and sold last month. Unfortunately, the world just doesn’t work this way.

If I’ve got this wrong, please help me understand how this approach to inventory optimization is supposed to work.