I don't think we need to worry over-much about averages, even when "one person's variant isn't variant enough to appeal to somebody else." But the 'lowest price' number can be misleading, and we might want to consider (say) eliminating the low two auctions whenever we have at least ten completed auctions, or maybe a few lowest BIN sales.
Sometimes the low end is skewed by early Buy-It-Now listings, where the seller was not yet aware of the "real" value of what he had. The most famous case is with the original POA set, where one well-known seller put up an early BIN for Pear Tree Leaves and another well-known buyer happened to notice it immediately; it's still by far the Low ending price. For a while after that, I snuck in some checks for new BINs while I was at work using a "company" computer.
Then, occasionally a card will be listed with misspellings in the title, or in a category that didn't give it visibility, or unavailable to USA or UK buyers, etc.
On the other hand, the rare weird results are part of the legends and lore of the hobby. Just like the auction listings of "Buy-It-Now for 0.01 ... Or Best Offer".
As a guide to my own bidding, I watch the last-five-average and the low result, but quite often the low result is so much lower than the next lowest that it doesn't help much.
