The High Yo Yo is initiated from a position on the bandit's turn circle in his rear hemisphere. All you are doing is altering the shape of your turn circle and moving it out of plane from the bandit turn in order to reduce closure. As soon as you have solved the closure problem you put the lift vector in lead on the bandit (roll to point the lift vector in front of the bandit velocity vector) and pull. This brings your velocity vector back to lag pursuit on the bandit turn circle in the control position. If the bandit is low energy at this point, you can probably pull lead for a shot.
Dawger is spot on (once again) with an excellent description of a High Yo-Yo, it's purpose and execution, particularly the mention of lift-vector placement (which is so often overlooked for a relatively simple concept).
The important point being, if the attacker is not immediately re-engaging (either for a shot or to force another E-depleting defensive maneuver) then it is not really a High Yo-Yo. So, Junky, your point about the High Yo-Yo not being performed right may be true, because they probably aren't really performing a High Yo-Yo.
When I am the low defender, recognizing an aggressive High Yo-Yo is one of my primary means of assessing the skill-level of an opponent:
If an attacker simply climbs away after my defensive turn, the "one-pass wonder", then I know I am likely facing a less-skilled opponent. In this case I may bait them into attacking again and generally be more aggressive.
If an attacker performs a High Yo-Yo after the first pass, then I am more concerned, especially if they maintain E and avoid my reversal shots. Then I know the opponent is more skilled, so I am more likely to use extensions (if possible), trying to equalize E for a scissors fight or looking for a way out, depending on relative aircraft performance.
This also answers the OP's question (what do you do?) with the answer I usually give: It depends --
mostly upon what the opponent does but also on a number of other factors.
