It's hilariously improbable to chess players, but some of my non-playing friends thought it was clever.
I actually don't think this is true. As I understand it, a given routine will have a number of high-risk tricks that may work, say, only 1-in-3 times. If they do work, they're highly effective, precisely because they're so implausible, and if they don't work you can just pretend it's a bluff and move onto plan B. It's all about effective showmanship and managing expectations.
(Disclaimer: not a magician!)
One of my friends had a card trick that involved me selecting a card. Before starting the trick, he jokingly picked a card at random from the deck, and asked me if that was my card. He had 1/52 chances of being right, but it would have been an insane trick if it had worked.
He guessed wrong (naturally), shrugged and went on with the actual trick as though the first attempt was just a joke, to be forgotten because an actual trick was about to take place.
And we don't want symmetry on one board, but between two boards.
Try it yourself vs a few instances of your favourite chess software or just here http://www.chess.com/play/computer (requires Java). Open a few instances and see how can you chuck away a move.
There is an interesting trick that was common for a time and still gets referenced by magicians where even the worst chess player can guarantee that they will win at least half of n chess games where n is any even number. They will also be able to appear to need very little time when they get to each table.
The trick of course is just to play each player against another. The magician acts as a proxy and makes the moves of one player against another. Of course it takes a little practice at memorisation but magicians are often practised at mnemonics anyway.