"apples and oranges"

With fruit, it's easy... it's either an apple or it's not. With cars, it's tough because an Accord is not an Impala which is not a Fusion which is not an Optima. Now take oranges (the color, not the fruit). If it's an orange XV Crosstrek it's good but if it's an orange RDX it's positively auction-proof. My point is, how can similar cars differ so much in value? Within one segment such as the Accord, Impala, Fusion, Optima sample... each of these sedans fit essentially the same niche. Yet, [a] group of 2012 Accord EX sedans sold at [a] auction on [one] day can vary $2,000. And the same can be said for the Impala, Fusion and Optima. Now consider a composite of vehicles; say a multi-franchise, multiple location new car dealers' entire pre-owned inventory that that dealers management calls its ACV at $11,450,000. That is to say, that if all 760 units in its inventory were cashed today, they would get a check back for the 11.4 million. What are the odds would you guess that they get back the 11.4? Or is it more reasonable to use a percentage of the total, like say, 95%? Or use 102%, 91%... what ever figure. But how best to derive at that number without auctioning each single unit off? What if there was a willing buyer for the total package and that buyer valued the composite based on what he or she thought they would have to pay for each one at auction? What percentage of the 11.4 million would that be? Less, right! Probably and that's because the sample principal that's been in effect for ever is at work here: the owner of [a] car always thinks its worth more! As for apples and oranges... take your pick... they're both good for you! W/