Picking up in the third world, the Antlion miniboss was a cinch. The Antlion's only dangerous move is Sonic Wave/Dischord to cut a character's level in half, and that was irrelevant for Delphi since Predict deals damage independent of level or character stats. For the first time ever, I had a solo character that didn't pick up either of the secret weapons in Mua. Delphi couldn't use the Brave Blade or the Chicken Knife as a staff-only class; the only other class in the same position is the White Mage, which had been played by T-Hawk. Therefore it was off into the Pyramid next:
The Gargoyles at the front entrance were surprisingly difficult to deal with, largely due to the fact that they absorbed holy element. That made Divine Judgment useless and forced Delphi to get a bit more creative. Their continued use of the Fusion ability was also super annoying, since it would restore both Gargoyles back to full health (and a zero digit for their HP totals) which dropped the damage out of Predict back down to virtually nothing. Eventually I found a solution by falling back on Cleansing + Blessing, since Cleansing deals non-elemental damage and has the highest base magic multiplier of any prediction. I repeated Cleaning/Blessing combos until one of the Cleansings did 6000+ damage to both Gargoyles and cleared the path forward.
Condemn continued to rule the monsters in the random encounters in the Pyramids. One thing that using Condemn so much teaches you is which enemies have the "Heavy" flag in the code and which ones do not. In the Pyramid, that was the ZephyrZone woman who summons creatures to fight for her. Everything else could be cursed with Doom and speedily killed, while the ZephyrZone encounters forced Delphi either to run away or use repeated predictions. As for the Machine Head robots roaming around in the middle of the dungeon, it's well known that they have a weakness to lightning element. What's no so well known is that they are weak to water element as well, the rarest element in Final Fantasy 5, and Delphi could actually take advantage of this. The Deluge prediction deals water element damage to both the enemies and the party at the same time, and it's probably the easiest of the self-harming predictions to use because the damage to the party is very low. Unfortunately Deluge is linked to the "1" digit in the Prediction cycle and that makes it a challenge to pair together with other desirable predictions. For the Machine Head robots, Delphi would smack them with his Morning Star to try and get them off their starting 7210 HP total and then follow that up with a Deluge prediction for double damage. Delphi's ability to heal with the Rejuvenation curse or the Healing Staff minimized the danger, and he slowly worked his way through the robots until they were all gone.
Pretty much everything else fell to Condemn in the Pyramid. While Delphi took a pounding and often had to alternate Doom/Rejuvenation, Doom/Rejuvenation with his curses, he cleared his way through the whole area with a minimum of difficulty. Against Merugene following the end of the dungeon, Delphi equipped a Flame Ring and opted out of trying to guess her elemental weakness by using the Cleaning/Blessing prediction combo. That wasn't fast either but it worked and there was very little danger from the boss fight.
Access to the airship opened up the whole of the third world to Delphi. He picked up the standard gear in Mirage village, and access to permanent Haste status via the Running Shoes/Hermes Sandles changed up the gameplay once again. Haste and Slow status do apply to the curses that come out of Condemn, and Delphi's self-targeted curses now had a ridiculously low timer of two seconds. That's right, two seconds. 
He could Condemn himself with Rejuvenation and get back 1000 health in the virtual blink of an eyeball. (Side note: the Condemn timer gets affected by Haste/Slow status on monsters as well, and I saw it freeze entirely against an enemy that was in Stop status.) Alternately, Delphi's Magic Power stat was so high by now that self-attacking with the Healing Staff would also restore almost as much at about 750 HP. For that matter, most nasty status ailments could also be cured in two seconds via the Recuperation curse. This combination of speed and endless self-healing made Delphi a very difficult customer to defeat in battle, a little bit reminiscent of the solo Time Mage that lacked direct damage capacity but who compensated with overwhelming utility. And of course Delphi still had all of the various predictions in his back pocket, plus the Doom curse to mow down random encounters. He was proving to be a decently strong solo character, albeit one of the strangest I had ever played.
One thing that I was diappointed to find was that the holy element damage from Divine Judgment was not boosted by equipping the Sage Staff, the only one of the legendary weapons that Delphi could wield. I had tested the same thing with the Rock Slide prediction and the Gaia Gear armor earlier, and it was sad to see that confirmed. No 50% damage boost for Delphi in the final Neo Exdeath fight then, as it seems Predict is just completely independent of what your character happens to be doing. As for the next dungeon in the Solitary Island Temple, Delphi used his newfound Haste status to tear apart everything that moved with Condemn. I captured one of the mosnters in mid-animation above with the blacked out colors; Delphi had a cursing party taking place with the Owazoral two seconds away from death and Delphi himself two seconds away from being healed. For Stalker at the end, Delphi gave up Haste status to equip a Flame Ring again, then went back to the good old Divine Judgment/Eruption pairing. Both predictions healed him with each use plus the Flame Ring defended against the Blaze attacks that Stalker likes to use. Easy victory on the first attempt.
Much like the solo Necromancer, it was the Minotaur side of the tower that was more challenging for the solo Oracle. Not that Delphi was really in any danger though, it's just that this boss fight took a lot more time. Delphi stuck with the Running Shoes and attacked with the Sage Staff itself, which has slightly higher attack value than the Morning Star. Maybe the Morning Star would have done more damage with its armor piercing properties, but I figured that any such extra damage would be lost with its 10% miss rate. Delphi alternated between attacking and healing with the Rejuvenation curse; the Minotaur could never deal enough damage to get through his endless self-healing, and that meant this was a guaranteed victory. Omniscient was quite easy as well, with Predict not triggering the Reset spell from the boss. I had Delphi alternate between Divine Judgment and Hurricane here, taking advantage of Omnicient's weakness against wind element. That canceled out the Shell status on the boss and allowed Hurricane to deal 3000-4000 damage per pop when the numbers lined up correctly. It took about four total cycles of Divine Judgment/Hurricane to trigger the final Flare spell, which bounced harmlessly off of Delphi's Wall Ring. Oh, I should also mention that Condemn ignores wall status and the Rejuvenation curse works just fine with a Wall Ring on. Very nice!
The last two main dungeons in the third world tended to be more of the same. Condemn destroyed everything in the Great Trench without need for predictions. Against the Skeleton Unknowns, Delphi had a curse-off match with both sides flinging about Doom curses. However, their ticking death clock started at 15 seconds (30 seconds cut in half for Haste status) while Delphi's began at 5 seconds, and that was enough time to drop even three of them before time ran out for Delphi. Everything else was little more than a joke, and the lava floors were the only ones that did any damage of note. Against the piggy trio at the end, Delphi equipped a Flame Ring again to block their Fire 3 and Blizzard spells, then relied on the Cleansing/Blessing combination again to deal non-elemental damage to the group. I had Delphi attack with the Healing Staff (heh - these bosses are undead) to manipulate their HP totals accordingly before finishing off all three with a big Cleansing at the end of the fight. The Oracle is the most important class in the game to be tracking monster health in a spreadsheet to understand what's going on. Boss fights like this would be virtually impossible without doing so.
More Doom curses out of Condemn carried the day at Istory Falls. Even the mighty Tonberris lacked the "Heavy" flag in the code and died instantly to Condemn. Everything else was simply mowed down one after another, almost like Delphi had his own private Death Note and was writing the names down one at a time. For Leviathan, Delphi gave up Haste status again to equip a Coral Ring and defend against the Tidal Waves from the boss. It took some time to wear down all 40k HP on Leviathan with repeated uses of Divine Judgment + Hurricane, and I wouldn't say that this was an easy fight due to the physical damage that Leviathan was dishing out when he attacked twice in a row. Nevertheless Delphi made it work, slowly, and did win on his first attempt.
I decided to do the rest of the optional challenges in the third world with Delphi since I knew he was going to need more levels before facing the final bosses. Or that was the plan anyway. I started out with Odin, opened up the locked door in the basement of Bal Castle, and then realized that Predict is a really slow ability when it comes to dealing damage. Could Delphi deal 17k damage in one minute of real world time, give the slow nature of Predict and a lack of any lightning element damage? That seemed like a bad idea. Then I headed over to Phoenix Tower with the intention of climbing to the top, only to discover inside that the Bella Donna enemies will use Roulette frequently in their AI script. That spell bounces harmlessly off the Bone Mail for most solo characters, but it would instantly kill Delphi if it hit him, and there's no save points anywhere inside the tower. I actually made it up to the 20th floor before realizing this, then turned around to head out and died on the second floor to Roulette. Argh! Well there was no point in trying that optional dungeon either.
I took Delphi over to North Mountain to fight Bahamut and then ran into trouble there as well. He needed to equip a Wall Ring to bounce back the Mega Flare spells, only that left Delphi with no Haste status from the Running Shoes and damage accumulation from the boss piled up too quickly for me to heal it away. Now this fight was doable if I were willing to dip into Delphi's Elixir stash, but for an optional boss fight, no deal. Hmmm, we were already scraping the bottom of the barrel in terms of remaining optional side quests. I actually went down into the underwater Worus Tower for fun where I snapped the above picture. Unfortunately that wasn't very useful either, as the enemies down there were weak and gave only modest experience. I will remember for the future that Gogo provides 50 ABP for standing around doing nothing, but naturally that was pointless for Delphi by now. He had long since maxed out his class. So those side quests all proved to be a big bust for this character. Side quests? Who mentioned side quests? Obviously I meant that Delphi would be going straight into the final dungeon. 
Random encounters continued to be a breeze for Delphi as almost everything was susceptible to his Doom curse. Point and click and five seconds later the enemy was dead. The dreaded Land Crawls in the desert section were yet another big foe with no protection against impending Doom. I used the standard pattern of alternating Divine Judgment/Hurricane against Calofisteri, and was amused to find that Divine Judgment sucked away all of her magic points. She only has 1000 MP in total and Divine Judgment was taking away about 500 of them per casting (along with doing about 3000 damage on average). Running some bosses out of magic points could be a viable strategy with Divine Judgment, although I should caution that the MP drain doesn't function with every casting. It's one of those percent chance things like the instant death effect from Starfall, and I've been unable to find any information about what triggers the effect or what the odds are. Anyway, Calofisteri fell very quickly as always.
A quick word here about my prediction usage in the lategame. I was increasingly using the Divine Judgment/Hurricane combo and had determined that it would produce the highest damage output under normal circumstances. Cleansing does the most damage out of all the predictions, with a base spell attack of 250 and the highest multipliers, however Cleansing doesn't combo very well with anything else in the prediction cycle. It leads naturally to either Starfall, which has way too high a chance of instantly self-killing the Oracle, or to Blessing, which is a great recovery spell but does no damage. Instead, I could use Divine Judgment to do almost as much damage as Cleansing (spell attack = 220) while also healing Delphi to full in the process (it always restored 2500 HP), then follow it up with Hurricane (spell attack 200) to cycle back to Divine Judgment again. The danger here was that Hurricane would also injure Delphi on each use, up to about 1000 damage if he used the prediction with a "9" as the final digit of his health. Fortunately, the following Divine Judgment prediction would heal that away and Delphi increasingly had a larger HP pool to afford self-damaging predictions. The other natural counterpart to Divine Judgment was Eruption, which I had used throughout the second world, but without a Flame Ring equipped it was also self-damaging and it was weaker offensively than Hurricane at only a spell attack of 165. While this kind of theorizing might seem tedious, it was critical that Delphi could get the absolute maximum usage out of Predict possible. Since Predict does damage independent of level or stats, Delphi could never increase it further and it was starting to get long in the tooth for some of these endgame bosses. The elaborate setup requirements and slow pace of use was going to make the last few opponents very difficult indeed.
The hardest part of the Cleft of Dimension proved to be the Lonka Ruins section, which was very unusual since that's normally an easy area. The big issue here was the reappearance of Roulette from the very common Doom Dealer monsters in this area. Without the Bone Mail, Delphi could only hope that the bouncing cursor would miss him and he died multiple times trying to pass through the area. That was pretty annoying. By contrast, the Forest of Mua section was no sweat at all and I used it to build some levels once Delphi had secured the save point in the Caves section. At about 5000 experience per battle, the forest area made for very fast leveling. The Caves enemies are usually quite deadly, and yet once again they all were vulnerable to the Doom curse. I just had to make sure not to leave one of the Orcat monsters as the last one alive since they would then use a petrification attack. Delphi died one time to that and I had to walk all the way back in from the entrance again, argh.
Apanda was one place where Delphi fell back into his Divine Judgment/Eruption cycle of old, with the fire element on Eruption dealing double damage into the boss weakness. Here's one other reason why Eruption isn't too good in the lategame: it has a chance to inflict HP Leak status, and while that sounds great and all, it makes it impossible to know how much health the enemies have remaining. Due to the prediction mechanics, it's extremely important to know if a boss has a health total ending in a "0" digit or a "9" digit, which will result in very different damage outputs. Eruption makes that sort of planning impossible and therefore I kept it off limits aside from this one-off situaiton.
Next up was Apocalypse/Azulmagia, and Delphi found himself confronted with the first true roadblock boss of the whole run:
This guy was difficult for all sorts of reasons. For starters, Apocalypse has the highest magic resistance of any normal boss in the whole game at 70 points. That was enough to reduce the damage from Delphi's predictions noticeably. Making this worse was the fact that Delphi had no way to attack Apocalypse's weakness to poison element. The only prediction attuned to poison is Pestilence, which damages the party while doing nothing to monsters - it's the worst prediction by far and useless as an offensive weapon. Delphi couldn't break Venom Rods either since he was unable to equip rods. He couldn't even do what most other weak classes fall back upon by equipping the Chicken Knife and hacking away, since Oracles can't equip knives either. I doubted that attacking with the Morning Star or Sage Staff at under 1000 damage per swing was going to make this happen either. No, this battle was going to have to be done through very careful use of the Predict command.
I did realize right away that Delphi could get some extra damage via the Magic Lamp. With a Magic Power stat of 71 (!), he had a massive magic multiplier and the defense-piercing property of Bahamut would be good for at least one powerful shot of roughly 3500 damage. Leviathan was a bit weaker but still pretty nice at about 2500 damage. Aside from that, the remaining 27.9k HP on Apocalypse was going to have to be taken away using predictions. With some testing, I worked out a tactical plan: Delphi would start the battle with a Bahamut summons to knock the boss off his "0" digit starting HP total. Then I would start using predictions if boss health was on a high digit, or follow up Bahamut with Leviathan if it was on a low digit. Delphi was set up to begin the battle with Divine Judgment, his strongest prediction, and then cycle through to Hurricane before repeating the cycle. He could deal as much as 5000 damage if he could combine a Divine Judgment together with Apocalypse's health total ending in a "9". Given enough time, this was a winnable battle.
Unfortunately, Apocalypse was very good at wiping out Delphi in a hurry. Here's the boss AI routine when not casting Blue magic:
{DrkShock,GuardOff,MindBlst}
{Condemn,Roulette,L2 Old}
{L5 Doom,L4 Qrter,MghtyGrd}
{TinySong,Flash,????}
{Aero 2,Emission,Missile}
{Time Slip,DeathClaw,WhiteWind}
Oh boy, there's a *TON* of bad stuff there. 
The second line was particularly bad with Roulette and Condemn. The former was a 50/50 chance for instant death, while the latter really was instant death because Delphi couldn't win before the ticking death clock hit zero. (I'm still annoyed that the Oracle class can't override enemy Condemns with their own curses - that would have made this infinitely easier.) Mighty Guard in the third row was almost as bad, cutting in half all the damage that Delphi could do. That was also pretty much grounds for an instant reset, as was White Wind in the last row. If Apocalypse was healing himself for 9999 HP, Delphi wasn't going to win the fight. And then there was ???? in the fourth row, which was another instant kill any time that it appeared. Even some of the spells that seemed relatively benign like Dark Shock and Guard Off made Delphi much more vulnerable; Dark Shock would cut his level in half and while that didn't affect Predict, it made the Magic Lamp virtually useless. I needed Delphi to get lucky with the damage from his predictions while also dodging all of these incredibly deadly moves in the boss AI routine. What a nightmare.
One thing that actually came in handy here was the Oracle's "Read Ahead" ability. It greatly cuts down on the rate of random encounters and that was what I needed in walking up from the save point in the Caves over and over again. Read Ahead reduced this from several minutes worth of running away from random encounters between boss fights down to about 30 seconds before I could make another boss attempt. Thank you very much, otherwise useless ability! It was the very fact that I had to keep re-equipping Delphi after swapping Read Ahead for Predict that I started to hit on a solution to this battle. I accidentally equipped a Wall Ring instead of the Running Shoes once, and to my surprise I found that Delphi actually seemed to do better at the battle:
Wall status caused some of the Blue magic spells to bounce off of Delphi and hit Apocalypse, and that was more significant than I had been expecting. Delphi also didn't need the extra speed from the Running Shoes as much as I had thought, since his predictions took just as long with or without Haste status. He could simply keep using Predict and it would last about as long as a non-Hasted turn of action. When Delphi wasn't wearing a Wall Ring, he kept getting tagged with Dark Shock and having his level reduced, which then made him prey for Level 2 Old. (He went into the battle at Level 57 to avoid this, the same level as Apocalypse.) But with the Wall Ring on, none of that could happen, and in fact the bad stuff started happening to the boss instead. This was working well and I started getting closer to winning, taking Apocalypse down below 5k HP remaining on two different tries. On my best attempt yet, Apocalyse opened the battle with Dark Shock, which reflected off Delphi and hit him. Then he used Level 2 Old and managed to inflict Aging status on himself! That was critical because it slowed down the boss and limited how many of those deadly rounds of combat Delphi had to survive. Round three was a useless Level 4 Quarter, followed by Flash and then Emission and then Time Slip. No appearance of Mighty Guard, no ????, no Death Claw, and no White Wind. This was some serious luck.
On the offensive end, Delphi had opened up with 3743 damage from Bahamut, leaving Apocalyse on a very friendly "7" digit of health. I was then able to use the following predictions: Divine Judgment on a "7" digit, Hurricane on a "5" digit, Divine Judgment on a "9" digit, Hurricane on a "9" digit, and then Divine Judgment on a "7" digit again. Apocalypse had made it through his full AI routine by this time and he threw out Condemn on the second trip, but that still left Delphi with 30 seconds left and the boss was running out of life. With Divine Judgment's prediction ticking down, I had Delphi throw out a Leviathan summons from the Magic Lamp to spike out two big damage hits at once. The two landed almost simultaneously for 3500 damage and 2500 damage respectively, and that was the end of Apocalypse:
What a thrilling victory! 
That was one of the hardest fights I've ever done in FF5 and it took a complete understanding of the Prediction mechanics combined with a lot of good luck at the right time. I averaged 3800 damage from those five predictions that hit together, and that was just enough to sneak over the finish line with help from the Magic Lamp. Now it was obvious that if Apocalypse was this difficult, Delphi was going to have massive issues with Neo Exdeath's vastly higher HP total. I had some thoughts about what to do there as well, although only time would tell if my plans would work out.
Fortunately the next few bosses weren't nearly as difficult. Catastrophe was simply a matter of avoiding his petrification attack for a few rounds until the Divine Judgment/Hurricane combo could do him in. This took a few tries and was otherwise a routine boss fight. Halicarnassus was a matter of equipping a Wall Ring and waiting for Holy spells to bounce back and damage the boss. I actually had Hurricane trigger its secondary effect of Silence status and prevent Halicarnassus from casting Holy at one point, which amusingly served to prolong the battle a bit longer. Then there was Twin Tania:
I wanted to do this fight normally using the Rejuvenation curse to keep Delphi's health high enough to survive the Giga Flares from the boss. However, when the first Giga Flare hit it did about 3150 damage, and that was slightly more than Delphi's maximum HP of 3128. Oh well, probably not worth the effort anyway. I certainly wasn't going to waste Elixirs here when I had a ready solution in the form of Condemn. As I've established across many other solo game reports, the "Charging Up for Giga Flare!" form of Twin Tania is vulnerable to virtually every status ailment and instant death attack under the sun, which includes the Doom curse out of Condemn. The trick for Delphi was that he had to use Condemn before the text popup appeared, since otherwise Giga Flare would go off before the five second countdown. This isn't as hard as it seems because there's a brief pause on the action bar that indicates when the form shift takes place. I was early on the timing by a fraction of a second the first time that I tried this strategy, and then executed it correctly on the second try. Hello evil grinning skull, goodbye boss.
Necrophobia was the next boss on the slate, and I prepared for this encounter by leaving the final dungeon to recharge the Magic Lamp. Those summon spells were going to be part of my strategy for making it past the first half of the Necro fight. After picking up another level on the way back in again, Delphi returned back to the save point in the Dimensional Castle. I once again turned to the Read Ahead ability to avoid random monsters in the Void, since none of them would provide experience points for Delphi. This allowed me to make quick returns to fight Necrophobia repeatedly; I normally could make it all the way from the previous savepoint in the Dimensional Castle up to the boss without a single battle. That last Oracle class ability was coming in more handy than I'd thought.
When Delphi met up with the boss, it was time to break out the Magic Lamp again:
That's the damage from a Bahamut Mega Flare deployed against the Barriers. My original plan here was to open with a Bahamut use to scramble the HP totals of the Barriers and then follow it up with the Predict command with a Divine Judgment prepared. That didn't work out though, as the Barriers would simply kill Delphi with their spells before the prediction could go off. In fact, even if I opened the battle with Predict, it still wouldn't count down to zero before the Barriers had a chance to act and finished off Delphi. Hmmm. What I wouldn't have given for Shell status out of the Wonder Rod here. As it turned out, there was time for Delphi to act twice before the Barriers acted, just not enough time to get a prediction off during one of those two actions. I quickly realized that Delphi could use the Magic Lamp twice, cycling through Bahamut and Leviathan togeher and picking up enough damage in the process to destroy the Barriers. Several times the result was three of the four Barriers dying and a single one remaining alive with about 100 HP left over. That was the best possible result, as Delphi could run it out of magic points by repeatedly healing himself with the Rejuvenation curse and then use the Magic Lamp again for a useless Odin summon, preparing Syldra for the second half of the battle.
The first half of the Necrophobia duel was therefore quite easy, with the Magic Lamp (and the ridiculously high Magic Power stat on the Oracle class) doing all of the work. It was the second half of the battle after the Barriers were down that was the tough part:
Necrophobia is weak against all eight elements and Delphi's predictions had the potential to deal massive damage. I could also break out Syldra via the Magic Lamp for a one-time burst of about 7000 damage as pictured above. The tough part here was the slow nature of Predict, as even the fastest Third Rank prediction still took about two rounds of Hasted combat to appear. Delphi also had to worry about the heavy physical damage from Necrophobia, who would attack twice in succession and deal about 700 damage with each attack. That was more damage than I could heal with Rejuvenation curse (still restoring right about 1000 HP per use), and although Divine Judgment would heal Delphi when it appeared, its partner prediction of Hurricane was still self-damaging and dangerous to use. I was forced to resort to a few Elixirs here for pretty much the first time in the entire solo Oracle run's duration. And worst of all was Necrophobia's Maelstrom/Fight combination, where he would reduce HP to critical with the first move and then immediately land another attack for guaranteed instant death. That was responsible for wiping out more of these attempts than anything else. Nevertheless, when a good prediction could deal 9999 damage and Necrophobia only had 34k health to cut through, it was only a matter of time until I was able to line up the luck factors in my favor. Delphi did a lot of walking from the previous save point to make different tries at the boss, but eventually it all worked out and Gilgamesh showed up for the kill.
Now there was only the final struggle against Neo Exdeath, and that was going to be a doozy indeed. I was not confident that Delphi could overcome that challenge with Predict alone, and therefore I resolved to rely on his one last trump card:
Staff users have the ability to break Light Staffs for a one-time use of the Holy spell. This is their own miniature version of the rod-breaking that most of the other mages can do, and it was the strategy that I had used to get several of the weaker classes past the Exdeath battle at the end of the second world. Light Staffs are dropped by the Metamorpha enemies found in Guido/Gheedo's Cave, however they can also be found as a rare encounter in the Caves section of the final dungeon. I needed to go farm a bunch of them for use agains Neo Exdeath, and that meant a whole lot of random battles against these guys. There's a trick in the Gameboy Advance version where the player can Quicksave and then Quickload, after which the second random encounter in any area will always be the rare encounter. While I don't like abusing those kind of features, I also didn't want to spend untold hours looking for Metamorphas to show up, and then have only 1/16 odds of a Light Staff dropping when the Metamorphas did appear. Quicksave abuse was the only practical way of pulling this off. I ended up in a pattern where the first random encounter would always be against three Achelones, and then the second one would be a Metamorpha fight. I was shooting for 10 total Light Staffs to have in hand, and that meant seeing a *LOT* of these two encounters. The bestiary said that Delphi slew 104 Metamorphas, but that was an understatement because I only went back to save after a Light Staff would drop. There were many other times where Delphi fought long strings of random encounters, found nothing, and then eventually died from a back attack. This was quite tedious even with heavy use of Fast Forward on the GBA emulator and I was more than glad when that tenth Light Staff finally dropped.
I had also been theorizing about how to use the Predict command against the last boss for a long time, and I thought that I came up with the correct solution. How do you work around the need to cycle through different spells as the Oracle's magic point total changes, and work around the slow pace of the predictions? Eventually I found the answer: make it so that the Predict command never cycles at all. The trick was to pick a starting level where Delphi's maximum MP total was keyed to the prediction that I wanted. Then I could use the Third Rank (7 MP) prediction, the one that goes off the fastest, which lined up almost exactly with two turns of Hasted movement on Delphi's part. That essentially gave Delphi one "free" turn while waiting for the prediction to go off, and he could use either an Ether or an Elixir on that turn to take his MP total right back to maximum. Then he could immediately use the same Third Rank prediction again, use another Ether/Elixir, and keep the cycle going endlessly. There would never be a need to waste time with the weaker predictions, as Delphi could go right ahead chaining the most powerful ones over and over again.
I wanted to use Cleansing for this as the highest damaging prediction in the set. There was just one problem: Cleansing meant using a level with a "0" digit at the end of Delphi's magic point total, and the only options were Level 60... or Level 83. There was nothing in between with a zero digit available. I tried to go for the final boss at Level 60 and that didn't work out as planned. The magic multiplier for Delphi was exactly M = 20 at Level 60 on all non-Predict damage, and that was bad because every damage roll when multiplied by 20 had a zero digit at the end, making it impossible to cycle between different ending numbers on boss health. For example, the Exdeath Tree form starts with 49,001 HP and I obviously wanted to get off that "1" digit at the end. But every non-Predict form of magic damage would always end with a zero, thus keeping Delphi stuck with that "1" digit for the boss HP total! This was no good so I took Delphi up to Level 64, where his MP total became 555 and therefore became keyed to Divine Judgment, while his magic multiplier went up to M = 21. Divine Judgment was probably a better choice than Cleansing anyway since it healed Delphi while dealing almost as much damage. This was where I would make my stand against the last enemy.
The Exdeath Tree form was the hardest opponent in the game for Delphi. This was entirely due to White Hole, the petrification instant death attack that the Exdeath Tree has a chance of using in almost every single round of combat. Delphi had absolutely no protection at all against this attack and it wiped him out about 80% of the time when it appeared, which it did all too often. I needed Delphi to win this battle quickly to minimize the chance of White Hole appearing. I thought he had pretty good odds against Neo Exdeath if only he could make it there, but the darned Tree kept blocking Delphi's path. My plan was to open the battle with a Bahamut summons from the Magic Lamp, both to get a cool 5000 damage and also to scramble the HP total to a more friendly last digit. If the monster health ended with a "7", "8", or "9" digit, I would have Delphi use Predict to get another big blast of damage from Divine Judgment. If I was lucky enough to get a "9" ending digit, Divine Judgment would do more than 6000 damage, excellent stuff. On the more common occasions where that wasn't the case, I would follow up the Bahamut summons with a Leviathan summons and then follow the same reassessment process depending on where the boss HP total wound up. After that, it was time to start breaking Light Staffs for roughly 5000 damage apiece. Rinse and repeat until the Tree form collapsed.
That was a lot harder than it sounded. White Hole appeared again and again, with the appearance almost always proving fatal. The numbers often failed to line up correctly and Predict frequently wasn't very useful in these attempts. Nor was the Exdeath Tree simply sitting on its hands; it would launch regular attacks as well, and I would have to plan around the Flare/Holy round that always took place when the boss dropped under 30k health for the first time. Flare would do about 3000 damage and I had to ensure that Delphi could survive that or it would be yet another game over. I rarely made it far enough to reach the part of the AI script where Meteor can appear, although that wiped out Delphi once with extreme prejudice. Still, it was White Hole that was the great killer in this fight, just as it had been for so many other solo characters. There's a reason why I hype the Aegis Shield so much for my solo characters, as one of the few things that can stop that dreaded attack.
This was the whole reason why I had farmed those Light Staffs, specifically to make it past this roadblock boss. There was one neat thing that the Predict command could do though, and that involved "spiking" the Exdeath Tree for double damage at once. I could set up a Divine Judgment prediction on a juicy boss HP total, and then have it land at the same time that Delphi broke a Light Staff for additional damage. This was like getting off two rounds of damage together, and I used this tactic to get past the Meteor section of the AI routine by hitting for about 11k damage almost simultaneously. After a great deal of trying and retrying this fight, I finally did make it to Neo Exdeath for the first time:
The final form of the boss still remained though, including the dreaded Grand Cross. I had no answer for that move aside from crossing my fingers and praying for luck. However, I thought that I did have a solution for everything else that the boss could do, and a method for cutting through the huge health totals of the void monster in rapid fashion. With the threat of instant death removed from everything other than Grand Cross, I could finally use repeated Divine Judgment predictions to do the bulk of the damage. I opened the battle by using Odin from the Magic Lamp to remove Neo Exdeath Part #4, then broke one of my three remaining Light Staffs to randomize the HP total of the Grand Crossing Part #1. It did 5166 damage to leave that part with 49,834 HP left, and I figured that was good enough with a "4" digit to warrant a use of Divine Judgment. The prediction dealt 3345 damage to Part #1 (and virtually nothing to the other two parts because they were on "0" health totals), leaving the Grand Crossing part with 46,489 HP left. Bingo! There was the "9" digit that I wanted. We were in business now and Divine Judgment was about to be unleashed in full. This meant the maximum possible magic multiplier of M = 30 for Divine Judgment, and that further meant so long as I kept using nothing but Divine Judgment for damage output, every prediction would ALWAYS deal damage ending in a zero. In other words, the damage output against Part #1 would be 6480 damage, 6300 damage, 6510 damage, and so on, always keeping the boss health total on a "9" digit to continue doing the maximum damage possible. Delphi just had to keep using an Ether or Elixir on every other round of combat and chain the Divine Judgments together endlessly. 6000 damage, 6000 damage, 6000 damage and so on every single time out of Predict. This was what I had theorized a long time ago and it was incredibly gratifying to see it working in practice.
It was almost impossible for Neo Exdeath to damage Delphi while I had this combo taking place. My solo Oracle was either using an Elixir for full health restoration or Divine Judgment for 2000 HP back on every round of combat. Even Almagest and Vaccuum Wave attacks were mostly toothless in the face of that incredible healing power. Over time, the other two Neo Exdeath parts also randomly ended up on "9" digits and then also became perma-locked into the maximum damage output from every Divine Judgment prediction. Part #3 had done so first and ended up about 10k health lower than Part #2, but that was fine as I still had the two leftover Light Staffs to break to even them out after Part #1 was gone. That meant the one and only danger in the battle came from Grand Cross. I hoped and prayed that Delphi could survive the big honking dice roll since nothing else was going to kill him in this battle. The first one rolled Toad status, the second one did nothing, and here comes the last one... Aging status! That was curable with the Recuperation curse's Heal/Esuna function, and one more Divine Judgment took out the part. Success!
Now I simply had to pull off the remainder of the battle without flubbing the execution. I kept using Divine Judgment for 6k damage per use until Part #3 was down to only a tiny amount of health left. I actually cut this closer than I wanted, with the part having just over 100 HP left. Yikes. Divine Judgment rolled really high damage on that last prediction. Meanwhile Part #2 had about 11k health left, so this was a simply matter of smashing those last two Light Staffs for 5k apiece and then finishing things off with another Divine Judgment. Except that when Delphi broke a Light Staff, it suddenly did only 1400 damage - what the heck?! Oh, it was Aging status at work, which had sucked away a lot of his Magic Power stat in the interval before I'd been able to cure it. Well that wasn't great. Delphi broke the other Light Staff but that still left Part #3 with about 8000 health remaining, far too much to remove in a single prediction. And I couldn't use the Magic Lamp or anything else because it would finish off Part #3, preventing me from eliminating both parts at the same time. What was left then?
Ummm, time to whack away with the Sage's Staff at 110 damage per attack. Heh. This was high comedy here, but there literally was nothing else to be done. Delphi simply had to reduce this last part down to under 6k health remaining, and then use Predict with a "9" digit on the health total to wipe it out with a final Divine Judgment. If that meant plinking the boss 20 times in a row, well, so be it. I was not going to do this whole fight again. Now this is a place where my obsessive planning and packrat instincts came to the fore. I had stockpiled 50 Elixirs because Delphi never needed to use them all game, and his inability to equip much of anything meant that virtually every treasure throughout the game could be sold. I never thought I would need all those Elixirs, but now they came in handy indeed, letting me fight this battle almost indefinitely without running out of resources. The Almagest and Vaccuum Wave Neo Exdeath parts were never going to kill Delphi under any circumstances, not if I used my healing items correctly, and I was too old a hand at this game to let that happen. I worked down Part #2 under 6k health remaining, at which time the game refused to give me a "9" digit on boss health. It was comical how I kept attacking again and again without getting a Predict-friendly HP total. Finally that part was down to 4800 health left and it ended up on a "7" digit. I calculated that even on a minimum damage roll from Divine Judgment, that would be just enough to get the kill (220 base damage - 19 magic resistance) * (M = 24) = 4824 damage. I did end up getting a low damage roll but it didn't matter. All of the number crunching proved true and the boss fell exactly as planned:
Victory on the very first try against Neo Exdeath, with a total of three Grand Crosses dodged. Behold the power of math and be amazed! 
Whew. The Neo Exdeath showdown was an intense affair and I'm still pumped writing this a few hours later. It was a fantastic way to close out the run of the solo Oracle, unquestionably one of the game's strangest and most unique classes. There's nothing else that plays in the same fashion as the Oracle, not even close. I'd have to say that this was one of the weaker classes in the game, if not quite so bad as some of the Insanity Challenges on my website. While Condemn might crush everything fought in random encounters, Predict requires a lot of planning and understanding of the math involved to use properly. I'm happy that I was able to use it to the maximum extent in the final Neo Exdeath battle, where a long succession of 6000 damage printouts made the Predict command quite a sight to behold. If I'm being honest here, the biggest weakness of the class is not the Condemn and Predict commands but the equipment restrictions. If the Oracle could equip either Rods or Shields, this class would be well above average. It's the crippling lack of anything but Staves that knocks the Oracle down the tier list. (Someone was also having a joke by giving the class with the highest Magic Power stat in the game two abilities that never use Magic Power in any way.)
The Oracle was the final job in the game that I hadn't played out as a solo class. Technically I didn't do the White Mage either, but that one was covered perfectly by T-Hawk and I have no desire to repeat the same steps through many hours of tedium. This is therefore the end of the road for a challenge I set myself a full decade ago. The solo Monk was the first class that I did in the fall of 2008, and here I am completing the Oracle in the early part of 2018. It's been a long journey and I've had a lot of fun along the way. I'm going to put together a tier list for solo classes at some point, and that will likely be the end of Final Fantasy 5 content on this website. Perhaps some more Four Job Fiestas in the summer, but otherwise this is it. I've done everything I wanted to do with the game and in more detail than is probably healthy.
Thanks to all of you for reading over the years. Without the support from the fans, I never would have finished this solo class project. Best wishes to all of you.
