No, no, no, no, no.
First, unless I'm bat crazy, I already provided the same data that Verm is giving, simply through a different method. If I am, then just listen.
Because I got curious while testing (mad boring stuff, let me tell you) I decided to see what would happen if a Ganoncide occurred during Sudden Death. Victim victory is still possible; however, if that does not occur, then the victor is determined by percent lead, and if both players are equal then Ganondorf is proclaimed the victor. An important thing to note is that the victim does not gain a +4% UNLESS he gains a victim victory - meaning that if both of you die, he gets no increase in percentage.
So yea. Stamina mode isn't going to prove anything - the +4% victim victory can still occur, and during modes that determine victory by percentage, Ganondorf will win if percents are even or he has a percentage lead.
Second, while I'm glad that you guys have begun to test this move that you've been using forever, don't make claims about it unless you use a reliable test method. Here are some basic steps to follow if you want to achieve this.
1) Establish a constant testing method. This means that variables must remain constant within a testing condition. Let me put this in a simplified example: I want to see if there is a difference in result between a Ganoncide performed in Port 1 and Port 4. Let's say the only other variable was percentage - if that was the case, then I must use the same percentage when testing both ports. For instance, I couldn't have equal percentages for the Port 1 testing and Ganondorf with a 30% lead for the Port 4 testing. Your variables must remain constant.
2) Run enough tests to get a reliable result. Four or five times isn't going to cut it - ten times is a minimum if you want results worth making claims on.
3) Test on a variety of characters. This is important because the move may interact differently with different characters, as we have already seen that it does from the OP. If you want credible results, you must have results from many places; if you test on only one character, there's no way to tell if that character is the exception to the rule or ordinary.
These are basic guidelines, and I think I covered everything.