I worked through one sequence of play for the previous goldfish draft, giving a lower bound to the possible score. I'm not fully satisfied with it, but it has become too cumbersome to work through the possibilities.
The basic idea is to get the [mtgcard Glint-Eye Nephilim] down, give it double strike with the [mtgcard Silverblade Paladin], use [mtgcard Gisela, Blade of Goldnight] and [mtgcard Anthem of Rakdos] to increase the damage, use the card draw from the first attack to pump the [mtgcard Glint-Eye Nephilim|Nephilim] for the second attack and [mtgcard Niv-Mizzet, the Firemind] for damage. Throughout, [mtgcard Elbrus, the Binding Blade|Withengar] keeps growing, and most the extra cards get turned into land to fuel the card dump, and also to allow [mtgcard Ghave, Guru of Spores] to shuffle counters around, allowing the [mtgcard Goblin Sharpshooter] to do damage. Finish off with [mtgcard Soul's Fire] and [mtgcard Sword of the Ages] for further damage.
The resulting score was 4,139,209,534 (details are available in this spreadsheet); this looked like it would be the best of the group for some time until one of the others found an alternative line of play which bumped his score up to 27 billion, a target which seems sufficiently far ahead that I'm not motivated to find a better line of play. There are some obvious areas for potential improvement, however; starting a bit slower might allow [mtgcard Elbrus, the Binding Blade|Withengar] to grow earlier. It should also let [mtgcard Chandra Ablaze] use her ability in the last round to re-cast [mtgcard Soul's Fire] and [mtgcard Titanic Ultimatum], with the latter also opening up the possibility of cycling a lot of damage through the [mtgcard Crypt Rats]. Certainly options worth considering, and I suspect that they would lead to an overall increase of score... but I'm rather doubtful that it could reach the required 27 billion mark.
