from an implementation POV, the easiest would be to let you pick who you want to be the winner - and then at the end of the vote, your "winner" and the real winner get exchanged. That way, all the votes work out, very little special code is needed, and you have the maximum amount of plausible deniability, while it is still possible to work out that the election was rigged (e.g. if those voting for the real winner talk amongst each other and figure out that he should've gotten more votes than he did).
If the person you pick as a winner would have arrived second place anyway, this works out perfectly. However, if the person you pick was going to receive zero or very little vote, then your plausible deniability is gone.
That's not necessarily a bad thing however, as it would force people to pick and choose their fights.
And if it's simpler than adding x votes, I'm all for it.
