The Oracle was the final remaining job class for me to play out as a solo game, and I had been saving it as potentially the oddest choice of the whole bunch. The Oracle is another magic casting class, but unlike every other spellcaster in the game, they make "predictions" and initiate a ticking clock in the background. When it reaches zero, some kind of effect takes place that damages the enemies... or your party. The Oracle is generally seen as a nearly useless class, and for those unlocking the job in the normal fashion at the end of the third world, it's hard to see much value in the class. Much like the Dragoon, the Oracle is not a party-friendly character and functions better when he or she is off on their own conducting one of these solo missions. I hope to demonstrate in this playthrough that the Oracle is a better class than most people believe, even if using an Oracle optimally requires a lot of arithmetic to be effective.
The innate ability of the Oracle is Condemn, a skill that functions very similarly to the enemy ability of the same name. Condemn costs no magic points to use (something that's highly significant for this class - more on this later) and can be targeted against a single enemy or player character. When used, Condemn starts a ticking death clock that begins at 20 seconds, and when it hits zero the effect of that particular curse takes place. There are eight of these curses in all, with effects that include a strong healing ability, status recovery, damage attuned to fire/ice/lightning elements, and even an instant death spell. Condemn's biggest weakness is the slow wait for the ticking clock to count down against each opponent; the 20 second waiting time is just not very fast, although it can be sped up later in the game under certain conditions. Furthermore, using Condemn a second time against the same target does absolutely nothing. Once the curse is set in motion, it's impossible to stop even if the Oracle dies. This is a fairly weak ability on face value and has contributed to the reputation of the Oracle as an underpowered class.
The signature ability of the Oracle is Predict. The Oracle chooses one of three different predictions of different strengths, costing 1/3/7 magic points each, and a countdown begins in the background. When it hits zero, one of ten different big effects takes place. Unlike the single target nature of Condemn, Predict is always mass targeted at everything on the battlefield, and many of the predictions that come out of this ability deal damage to everyone. It's quite possible to hit your own party for thousands of points of damage, which is the biggest reason why this skill is typically seen as useless and counterproductive. Predict is not actually random, only appearing to be so, and with some practice it can be a powerful tool indeed. I'm going to need several paragraphs to explain how this skill works later on since it is not straightforward at all.
When it came to picking a name, there wasn't a lot to draw upon. Oracles aren't exactly commonplace as a character trope, and the only one that I could think of was the doomed seeress Cassandra from the Iliad. That had too many letters to fit within the six character limit and was the wrong gender for the Bartz character that I'd be using. Instead, I stuck with a Greek theme and used "Delphi" as my name. Delphi was a place, not a person, the seat of the ancient Oracle that has been memorialized in the Civilization series. I think it fit well enough for this character though. Here was the familiar screen after heading through the Wind Shrine and changing classes at Level 6:
The Oracle as a class tilts heavily in one direction for its stats. Strength was extremely low at only 19 points, tied with the Black Mage for one of the worst in the entire game. This is one job that doesn't want to engage in melee combat under any circumstances. Agility was average at 24, but the Vitality/Stamina stat was brutally low at only 23 points. That was worse than the Black Mage, worse than the Summoner, and only a few points higher than the Dancer. Delphi was going to struggle with a low HP pool for the whole game, and that would be a source of anxiety throughout this journey. The exception to this doom and gloom was the Magic Power stat, which was the highest of any class in the entire game at 61 points. That's three points higher than the starting Magic Power on the Summoner, the class that was the highest in the non-GBA version and the class with the most similar stat distribution. Unlike the Summoner, the Oracle cannot equip rods or knives however, and lands in the same boat as the White Mage as a staff-only class. Ugh. 
Well, at least Delphi wouldn't need to whack enemy bosses with a Flail to get past them. He had Condemn and Predict to dish out damage for him.
Delphi started out with Condemn as the only usable skill, and he would need 75 ABP before unlocking Predict. Condemn was the skill that put little ticking clocks above the target's head when used, like so:
I included a picture of the little stars-and-planets graphic that appears when using Condemn, after which the numbers appear above (or below) the target's head. There are eight different selectable curses that can be used via Condemn, which I'll list here:
Rejuvenation: Restores target's HP.
Recuperation: Cures target of status aliments.
Brimstone: Deals fire damage to target.
Black Frost: Deals ice damage to target.
Judgment: Deals lighting damage to target.
Salentia: Inflicts Toad status on target.
Doom: Kills target.
Still Wing: Inflicts Stop status target.
Some of these were very useful in the early stages of the game. Rejuvenation was a very strong healing ability, worth right around 1000 HP in curative power when used. That was as good as an Elixir at this stage of the game, albeit on a 20 second time delay, and I found myself using it freqently with Delphi. He would often curse all but one enemy, then use Rejuvenation on himself, and then curse the final surviving monster, getting a heal restore right before the last opponent dropped dead. Recuperation wasn't useful at the moment, although I made a note to remember it for the future and test to see if it would remove statuses that items couldn't fix. (Was this Heal/Esuna, Dispel, or something else?) Brimstone, Black Frost, and Judgment used the same graphical effects as Fire/Ice/Bolt 2, and did similar damage attuned to each element. They were not the same as these spells, however, since they dealt higher base damage and used a different formula for the magic damage multiplier. In fact, they appeared to have a static multiplier because the damage never went up as Delphi gained more levels. That was disappointing; these curses are essentially locked in at roughly 500 damage and can never be improved from there. Delphi couldn't even equip rods to get the Magic Damage Up property off of a Flame Rod.
Still Wing was essentially useless for a solo character, as the Stop effect would wear off before Delphi could do anything with it. The countdown on Condemn was too long for the Stop effect to stack. Salentia was amusing with its Toad effect, and would come in handy in a few places. Doom was the same as the Doom spell from the Black magic spell or the Assassin dagger, right down to using the same evil-looking skull graphical effect. Overall then, this was an interesting lot of different curses to play around with. I figured it would be enough to get Delphi through the first world even if I couldn't get much use out of the Predict ability.
The plan was to use Condemn against Karlabos, as either Doom or Judgment's lightning damage would be enough for the one-shot kill. The Karlabos encounter hit Delphi with an unexpected roadblock: Condemn doesn't work against enemies that have the "Heavy" flag in the code, including nearly all bosses. Uh, that wasn't good. Delphi certainly wasn't going to try killing the big red lobster with his staff, which meant backing up to farm monsters until he unlocked Predict with enough ability points. I will normally do this against the Mold Wind enemies in the Wind Shrine since they have a chance to drop Elixirs, only to find that Delphi kept dying periodically against them. Condemn was not a fast killer. After he picked up two Elixirs, I cleared out and did the rest of the leveling up in the Torna Canal, where the enemies yielded more ability points and didn't fight back. Finally Predict was unlocked, and it was time to put it to use against Karlabos:
The first prediction that Delphi used was named Cleansing, which produced about 750 damage and was more than enough to defeat Karlabos in one hit. While that was good enough for the moment, it wouldn't be enough damage to scale very far into the later stages of the game. Fortunately, Predict can do a lot more damage than this and I was only seeing the tip of the iceberg here. Since Predict is a confusing skill to use and would be integral to the rest of this run, I'm going to go into the details of how it works before continuing with the narrative. First, the details on which of the ten predictions get chosen:
As I've done with other characters, it was easier to screenshot the text file than convert everything into an HTML table. The specific prediction that comes out of this ability is not in fact random, but is based instead on the last digit of the caster's magic points. Note that Final Fantasy 5 Advance never explains this in any way, which is why Predict is so commonly misunderstood. The table above indicates which one of the predictions will be used, and then the effect of that prediction in the second list. Delphi had 110 MP at this point in time and therefore pulled Cleansing as his first prediction. That's one of the best ones to get, fortunately, and it was enough to defeat Karlabos immediately. Blessing and Healing Wind are also quite useful for healing the party, while Divine Judgment heals the party and deals holy element damage to the enemies. Then there are a bunch of predictions that hurt both the party and the enemies at the same time, most of them attuned to specific elements. Eruption (fire), Deluge (water), and Hurricane (wind) could potentially be useful in some situations even with the self-damage effect, especially with a Flame Ring equipped. Other predctions are random, with Starfall and Rockslide hitting targets by pure chance. The absolute worst prediction is Pestilence, which deals poison damage to the party and doesn't hurt the enemies at all. It would be fun to use with the Bone Mail equipped, except that the Oracle is one of the classes that can't wear the thing!
There are officially three different types of predictions. First Degree (1 MP) has a five second timer and is supposedly the strongest; Second Degree (3 MP) has a four second timer and is supposed to be weaker, while Third Degree (7 MP) has a 3 second timer and is supposedly the weakest. I'm throwing "supposedly" a lot in there because as far as I can tell there's no difference in damage at all between these three predictions beyond the length of the prediction timer and the amount of magic points consumed. Even the timers don't seem to matter that much, since the First Degree prediction is only slightly slower than the Third Degree prediction. I found that it took about four seconds, or the length of time of the Second Degree prediction, for the action bar to fill up on Delphi. Anyway, the three types of predictions were mostly the same. The key to unlocking the Prediction skill was using their variable magic point cost to cycle through to the predictions that I wanted. When starting on 110 MP value, Delphi could use a Second Degree (3 MP) prediction of Cleansing to leave himself on a MP total ending in 7, then use a Third Degree (7 MP) prediction of Blessing, which would take him back to the same starting point for another Cleansing again. This was the first combo sequence that I figured out, leaving Delphi only pulling the predictions that I wanted out of this skill while skipping all the bad stuff.
Now as far as the damage output of Predict goes, check out this use of Cleansing against a random mob in the Ship Graveyard:
The Skeleton and the UndeadRusk in the back took the same 750 damage as Karlabos, but the Carcurser was absolutely flattened with nearly 5000 points of damage. Whoa - what was going on here?! As it turns out, the damage multiplier for Predict works in its own highly unusual way. This can also best be explained through the use of a table:
As the table indicates, the magic damage multiplier (M) is based on the last digit of the target's HP. This is totally crazy as a gameplay mechanic, and I suspect it was their way of coming up with something that felt "random" without being random in actuality. This also explains the mystery of why the Carcurser took so much more damage than everything else from Cleansing. Most enemies have a maximum health total that ends in a zero, like Karlabos with his 650 HP or the Skeletons with their 70 HP. However, the Carcurser has 75 HP and that meant that the magic damage multiplier was M = 18 instead of M = 3. Sure enough, the math worked out to about six times as much damage on the Carcurser, just as predicted. This is another reason why the general opinion thinks that Predict isn't very useful. The first Predict generally doesn't do much damage because most enemies start out with a zero value as the last digit of their health. It's the SECOND Predict that deals huge damage after the first one scrambles the last digit of their HP. Or the Oracle can smack an enemy with their Flail just to randomize that HP total and get a number with a higher damage multiplier. If the Oracle can get their opponents in a position where the last HP digit ends in a "9", that yields a Cleansing with a base damage of 250 and a multiplier of 30 = 250 * 30 = 7500 damage. Not such a weak skill now, right?
Delphi picked up the Flail in the Ship Graveyard and mostly used that to bonk the monsters over the head, using Rejuvenation self-heals from Condemn to keep his health total up. I did notice something important here with regards to Condemn: the timer was going down as Delphi gained more job levels in the Oracle class. It had been 20 seconds initially on the Condemn timer, then 15 seconds, and now it had dropped to 10 seconds. That made the Condemn skill much more useful, especially with the ability to pull a 1000 HP heal on a ten second delay. Delphi gained a level from this fighting, and that changed his max MP from 110 up to 119. Uh oh, that meant he was now set up to pull Pestilence out of the Predict skill. There was no way to avoid this because the story forces a heal at the save point in this area. So what did Pestilence look like in action?
It looked like an instant game over, that's what it looked like. 
It turned out that there was no possible way for Delphi to survive a use of Predict, even when used with a health value ending in zero for the lower possible magic multiplier. That still did 180 * 3 = 540 damage and that was much more than the 300 or so max HP that Delphi had at the moment. Since he had no other way to reduce his own magic points, Delphi was completely unable to use the Predict skill until he gained another level and increased his max MP to something that didn't end in a "9" value. This wouldn't be an issue as he progressed further in the game and gained more health, but it was pretty amusing for the moment. Predict was off completely limits on pain of horrible self-death. (Later I learned that I could trigger a slow-moving First Degree prediction for 1 MP cost and immediately run away from the battle before it hit. That would have salvaged this situation had I known about it.)
Fortunately Delphi didn't need Predict to be able to defeat Siren. He couldn't use Condemn on her either, but nothing stopped him from using Condemn on himself. I used the simple tactic of having Delphi alternate Attack/Rejuvenation, Attack/Rejuvenation, over and over again. Siren quickly inflicted Slow status on Delphi and it didn't matter one bit. She could never deal enough damage to kill him before the next Rejuvenation would hit and restore all of Delphi's health. This was the old monster regeneration issue from Diablo 2 all over again, only this time it was working in Delphi's favor. Siren couldn't get through his health regen and that meant Delphi was guaranteed to win eventually. After several dozen rounds of chipping away with the Flail, she finally keeled over in a very satisfying victory.
Delphi mostly used the curses from Condemn to deal with the monsters on North Mountain. They did enough damage to kill the enemies in one hit and didn't require as much setup as Predict. To deal with Magisa, I prepared another Cleansing prediction ahead of time, then bopped the boss over the head with the Flail one time. It did 44 damage which paved the way for a followup 5200 damage Cleansing. Overkill succesfully achieved. After looting Tycoon Castle, Delphi attempted to grab the Elf Cape in Worus Basement. He had enough damage to kill the Garkimasras via Predict, but he lacked the HP to survive long enough for the prediction to go off. It was the same story with Shiva upstairs, and I decided to return later to finish both optional subareas.
Worus Tower was a very short and easy area with all of the random encounters finished off with Condemn. For the Galura boss at the end, it was time to break out another prediction:
Take a look first at the HP and MP totals at the bottom of the screen. These are both displayed right before using Predict, making it easy to see what's going to happen from the command. Delphi had 133 MP and that meant that he would use the Starfall prediction, one that deals damage to both the party and all opponents at the same time. I had already attacked Galura with the Flail to deal 32 damage and leave the boss with 1168 HP remaining, which setup a very powerful multiplier of M = 24 for the prediction's damage output. Finally, I waited until Delphi had his own health total ending in a zero value (310 HP) to minimize the damage that he would take from Starfall. The result was the boss taking an ungodly amount of punishment while Delphi survived with minimal hurting. This kind of setup would get easier as he gained more levels and could take more self-damage without dying; right now, most of the predictions that hit both sides of the battlefield would result in instant defeat. Starfall in particular combos well with Cleansing since you can chain them together by going 110 MP (Cleansing) -> 103 MP (Starfall) -> 100 MP (Cleansing) and so on, using the 7 MP prediction followed by the 3 MP prediction over and over again. However, Starfall has a secondary property of a small percentage chance to inflict instant death status, and this did kill Delphi on several occasions. For this reason, I began looking for other Predict combos that didn't involve the chance for an instant reset.
The Steamship dungeon was one of the last places where the elemental Condemn curses were able to carry the main offensive load. Most of the enemies in this dungeon were weak against either ice or lightning elements, and that allowed the curses from Condemn to continue eliminating them in a single blow. Delphi had no trouble keeping his health maxed out between Rejuvenation curses and self-attacks with the Healing Staff. It turns out that there are no predictions that use ice element damage, and the one that uses water element (Deluge) was useless for LiquidFlame since she is oddly immune to water damage. Delphi repeatedly attacked with the Flail until he was able to get a high multiplier for Predict, then used Cleansing again for 3800 damage and another one-hit kill. There was never any danger here since Delphi could heal himself back to full with ease; I could have done the whole battle with the Flail alone if needed.
The escape sequence from Karnak Castle was a bit tricky because of the slow pace at which Condemn operated. Delphi also wiped himself out several times with Starfall before I realized that the instant death effect was triggering and resolved to avoid using it. The standard curses that did elemental damage worked for most of the enemies in this area, but Delphi went for the big guns and broke out his Doom curse for the Gigas monsters. They would have required multiple other curses and had no resistance to the instagib potential of Doom. Delphi opted to skip the Guardian dagger for lack of time (since he could never equip it) and picked up all of the Elixirs and gold in treasure chests instead. The Death Claw miniboss at the end was handled through never allowing him to change forms, one attack with the Flail to set up a following Cleansing prediction for 4400 damage. That was becoming routine by now.
The Brimstone curse from Condemn was enough to deal with all of the random encounters in the Ancient Library, and I was taking the time to fight them for the minimal ability points that they were worth. Delphi had reached the third level of the Oracle class and dropped the countdown on his Condemn ability to 7 seconds. He was going to need a lot longer to get the final 300 ABP to master the class and drop Condemn down to 5 seconds though. At least Ifirt was a simple boss encounter to deal with. Delphi hit him once with the Flail and then used another Cleansing for the one-shot finishing blow.
Byblos likely could have been handled the same way. However, I wanted to take advantage of his weakness to fire element by employing Eruption instead, the prediction that results from using the skill when the Oracle's magic point total ends in a "2" digit. Eruption is another one of the predictions that hits both the party and the monsters at the same time, which meant that Delphi had to use care when employing this move. I came into the battle with his MP total already preestablished and Eruption ready to go. For a next step, Delphi needed to hit Byblos and deal some kind of damage to change the ending digit of the boss' HP. Byblos starts out with 3600 health and that wouldn't do much damage at all (M = 2 for Eruption). Therefore Delphi needed to attack with the Flail and inflict random damage to get Byblos off of that health total ending in a zero, then use Predict and make sure that the subsequent Eruption didn't kill him at the same time. Byblos was not cooperative at all in this regard, Slowing Delphi over and over again and dodging many of the Flail attacks. Delphi attacked and missed four times in a row at one point - annoying. Worst of all was Magic Hammer; normally a fairly useless monster ability, Magic Hammer cuts current MP in half and immediately wrecked Delphi's careful preparations whenever it showed up. It took way too long for me to line all of these various conditions up at the same time, well over two dozen total attempts at this battle before I finally managed to land the 7200 damage Eruption to floor Byblos. In retrospect, it probably would have been better to use the normal Cleansing prediction instead. At least that one didn't deal damage to Delphi.
The Oracle was probably the single cheapest class to equip that I've ever run in this game. Delphi rarely needed to drink Potions and with his weapon choice limited to staves, there was nothing to buy in stores beyond the latest pieces of the light armor set. Along with selling all of the junk that Delphi kept turning up along the way, he now had about 40k gold and I decided to go grab a Flame Ring immediately. That meant doing a little fighting for the remaning 10k gold required, and the best place to do that was against the Black Flames on the little teardrop island north of Crescent. Each random encounter on that island was worth almost 1000 gold, and it was easy to use Predict to flatten the Black Flames in every battle. Delphi was alternating between Cleansing and Starfall here, each of which would one-shot the monsters. The Oracle's damage is completely independent of level and their Dark Shock moves did nothing to him. Of course, the reverse is true as well: leveling up the Oracle doesn't increase damage. Ever. It's a cruel joke: this is the class with the game's highest Magic Power stat and it never uses that Magic Power stat for any of its abilities. 
For the Sandworm fight, I had Delphi use another basic Predict combo: Cleansing (3 MP) followed by Blessing (7 MP), followed by Cleansing (3 MP) and so on repeatedly. Here's a screenshot of Blessing in action, which restores a large amount of health back to the entire party. The base amount is 210 HP and that gets multiplied by everything from M = 3 up to M = 30 depending on the last digit of the player character's current health. This was a very powerful usage of Blessing, almost the maximum amount possible, and far more than Delphi needed at the moment. Blessing also has a chance to inflict Regen status, which was quite helpful when it appeared. I wish Blessing would always kick in the Regen effect, but no, it's only some kind of unknown percentage chance to occur. The damage output from Cleansing was surprisingly underwhelming in this battle, never going outside the 500-700 range. I have no idea what was happening here since Delphi's predictions didn't behave like this in any other battle. Something about how the Sandworm fight is coded is very wonky, something that my solo Berserker had found as well. Anyway, the Sandworm is a simple boss to defeat so long as the character has some form of healing available. Delphi had three different forms of healing: the Blessing prediction, the Rejuvenation curse, and the Healing Staff. It was no trouble to keep healing away and using those low-damage Cleansing predictions until the battle was won.
Due to the lack of a prediction attuned to ice element, Delphi used the familiar Flail attack plus a Cleansing to take out Adamantium. That did enough damage to land the instant kill even through Adamantium's Shell status. Still, the days of using a single Cleansing to destroy bosses were rapidly running out. Their health totals were continuing to ramp up as Delphi progressed through the game, and I needed to rely more and more on combinations of predictions, chaining them together in sequence to ensure that only the specific predictions that I wanted were the ones that appeared. I had the chance to test this in more detail against Sol Cannon:
The Rockets and Flameguns were easily defeated through equipping a Flame Ring and using the Judgment (lightning element damage) curse from Condemn. I thought that Sol Cannon would be equally straightforward: two predictions would take out the Launchers (one to get them off their starting zero digit HP totals and then the second prediction to dish out the heavy damage), then Delphi could use his Recuperation curse to remove the Aging/Old status and he would have plenty of healing on hand to dodge around the Surge Beams from the main cannon. Unfortunately, it wasn't quite so smooth in practice. I was going to start out by using a Starfall followed by a Cleansing, which together certainly should have done enough damage to destroy both Launchers. Instead, the Launchers remained alive and kept shooting their Missiles at Delphi, even after both predictions went off. What was going on?! The Launchers have 10,800 HP but are programmed to self-destruct after taking only 800 damage. For whatever reason the Predict move was doing strange things to their AI programming and preventing that self-destruct from ever going off. That meant that Delphi needed to do this battle with two Launchers that effectively had 11k health, and a main Sol Cannon that had 22.5k health. Uh oh.
This meant that the Starfall plan was immediately out. While Starfall did excellent damage, it also injures the party and has that small chance to inflict Death status. With the Launchers firing away and Sol Cannon charging up its Surge Beam, there was no room for a prediction that harmed Delphi. I decided that I needed to get as much healing as possible and tried out a new combination: Divine Judgment (3 MP) into Eruption (7 MP) into Divine Judgment and so on. Delphi was already familiar with Eruption from the Byblos battle, and with a Flame Ring equipped it would heal him while damaging all enemies. Divine Judgment operated similarly, attuned to holy element and damaging all opponents while healing him in the process. Once the initial Eruption went off to scramble the HP totals, the followup Divine Judgment put out massive damage:
You would think that that would be enough to take out the Launchers, but nope, they kept on firing. There was no chance to use the Recuperation curse to remove the Aging/Old status from poor Delphi. I had to keep using Predict on every round, with the predictions that I had chosen fortunately healing him at the same time that they dealt damage. It took some repetitions of this fight to get the timing correct and avoid a Missile going off right before a Surge Beam, causing an instant reset. I also inputted the wrong Predict commend more times than I'd care to mention; it's very important to cycle through them in the proper order. Used in the wrong order, I wound up getting Pestilence for an instant self-kill. Finally though, everything lined up correctly and the Launchers both died simultaneously to the fourth prediction. They did not explode like they're supposed to do and I only knew they were dead from the lack of new Missiles. Delphi was now able to clear the Aging status and alternate more predictions alongside Rejuvenation curses for healing until Sol Cannon finally died. It burned away into purple mist, again not exploding the way it's supposed to be programmed. This was a weird situation but fortunately the Oracle class was up to the additional challenge.
It was the Condemn skill that dominated the Lonka Ruins, and specifically just one of the curses. None of the monsters in the random encounters were protected against the Doom curse, and Delphi happily sent the demonic skull off to destroy them one by one. Even the mighty Hydras weren't immune and fell as readily as everything else. This was an easy area for the Oracle, with Doom clearing a path forward and the Healing Staff removing any injuries taken along the way. I was surprised at how Condemn was still coming in handy. While it was no longer useful in terms of raw damage, Condemn provided excellent healing and status repairing and an ability to take out random monsters with instant death. It was actually getting better with time as Delphi gained more levels in the Oracle class. Condemn on a 7 second timer was a lot better than Condemn on a 20 second timer.
The key to defeating Archaeoavis invariably revolves around getting as much physical evade as possible to dodge the nasty paralyzation and confusion-inducing physical attacks from the boss. This meant that Delphi wanted the Elf Cape here instead of the Flame Ring, and that ruled out Eruption as a prediction. I decided to use the Cleansing/Blessing combination again, since Cleansing does the most damage of any of the predictions and a single hit from the Flail plus Cleansing would be enough to cycle through each of the five forms of the boss. It didn't hurt either that Cleansing is a form of non-elemental damage and that would remove any need to worry about Archaeoavis' many elemental immunities. I snapped a picture of the process in action above, with Delphi getting a single hit with the Flail (ooh 102 damage, setting up 1498 HP remaining on the boss for a Cleansing multiplier of M = 27) and then the massive 7300 damage output from the followup Cleansing. Delphi would then use Blessing to cycle back to Cleansing again and repeat the process for the next form. It went very smoothly and Delphi's surfeit of healing kept him out of the danger zone against anything other than the Maelstrom move from the final form. Delphi was wiped out on his first attempt by a confusion attack + Maelstrom combo, then won on his second attempt with no further close calls. He even pulled off the stylish kill by having the last prediction land while Delphi himself was paralyzed! Heh.
The Chimera Brain and Titan were both a cinch to defeat. Delphi equipped the Flame Ring for both of them and cycled through Eruption and Divine Judgment repeatedly for endless self-healing while simultaneously damaging the monsters. The Puroboros were the tricky battle as usual:
The biggest issue in this encounter turned out to be speed. The Puroboros will do nothing the first two times that their action bars fill up, and then on the third time they will either use a physical attack or Self Destruct. If I wanted to avoid running this battle dozens and dozens of times until all of the little bombs failed to use Self Destruct, I needed Delphi to defeat them before they had the chance to act. I experimented a bit and realized there was enough time to get off a pair of the Third Degree (7 MP) predictions, the one with the 3 second timer. If I tried to have Delphi use even one of the Second Degree (3 MP) predictions, he wouldn't be able to get both of them off before the Puroboros had a chance to act. My first thought was to use Cleansing and Starfall together here, since Delphi would only have to dodge a single death status from Starfall and that seemed doable. However, I ran afoul of the math required to cycle through these particular predictions. In order for both of them to be Third Degree (7 MP) predictions, Delphi had to start with Cleansing and then follow it up with Starfall. It wouldn't work to start with Starfall because the Third Degree prediction would then land him on Healing Wind as the next in the list and that was no good. And Starfall was no good as the second prediction because its damage hits twice. When I used Starfall, some of the Puroboros would die from the initial damage, then they would use Life 2 / Arise, followed by the second damage hitting and the subsequent Puroboros that died using Life 2 again. Obviously this was no good and I had to go back to the drawing board again.
If the Cleansing/Starfall combo was out, then the next best option was the Eruption/Divine Judgment combo with a Flame Ring equipped to protect from fire damage. It had to be these two predictions in this order, as that was the only way to use Third Degree predictions on both of them. If I were to start with Divine Judgment, then the next Third Degree prediction would be Hurricane and that would inflict too much self-damage against Delphi. So this was a very specific combination of predictions that Delphi needed to set up ahead of time, and fortunately it did work the way that I planned. Divine Judgment even had a special secondary ability that drained MP out of the target hit, leading to those messages displayed above. There's no formula in the new Algorithms Guide to explain how this works, I only know that it does work. There was one Puroboros left alive and Delphi kept healing himself until it Self Destructed to end the battle. Whew. This one required some real thought to navigate.
One whiff of grapeshot from the GilToss cannon was enough to get past the initial solo Galuf encounter with Gilgamesh. I would have used Predict except that it requires 75 ABP to learn on the Oracle, and I wasn't going to waste time grinding that out for one trivial encounter. Then it was on to the Big Bridge:
The random monsters on the bridge were surprisingly difficult! The real issue here was that Delphi's life total was simply too small, and because his curses took time to appear, he couldn't eliminate one enemy immediately at the start of the battle. The little chariot monsters with their Mustard Bomb ability were the worst culprits, since that attack does a flat 300 damage and Delphi had less than 900 total HP. I was wiped out here about four times while trying to use Condemn, at which point in time I had to switch over to Predict, using the Divine Judgment + Eruption combination (with a Flame Ring) so that Delphi was being healed by every prediction. That was enough to do the trick. For the second Gilgamesh fight, I queued up the same two predictions and used a smacking with the Flail to get Gilgamesh to 2848 HP remaining, where there would be a huge multiplier (M = 21) on the next Eruption. It came out dealing more than 3000 damage, and Gilgamesh used his Armor/Shell/Haste triple buffs... only to keel over immediately after jumping at Delphi. Apparently he always casts those three spells even if he's already dead. However, he did not print out his "retreat" message afterwards the way he's supposed to do, and instead simply faded into purple mist. Predict does some weird things to the battle AI scripting.
I continued to use Condemn for most random encounters in the wilderness near Rugor and then in the Underground Waterway. The Doom curse was effective against almost every random monster and made short work of them without needing to go through the careful setup for Predict. Delphi was also getting a lot of use out of the Healing Staff, allowing him to keep his health topped off in every fight. The Rejuvenation curse would heal about twice as much health back, 1000 HP as opposed to 500 HP from the staff, but that came on a seven second delay, and when you need healing, you often need it immediately. Both of these options had a place in Delphi's battle tactics. As for the Tyranosaur:
This boss is weak against fire element, and that allowed Delphi to set up an easy double damage Eruption prediction. I decided to opt into the real kill instead of the Phoenix Down cheese because the deadly ??? counterattack from the undead dinosaur only triggers on taking physical damage and wouldn't respond to Predict. I'll also use this opportunity to showcase a screenshot of the three predictions on the menu, the familiar First Rank (1 MP), Second Rank (3 MP), or Third Rank (7 MP) options. I was completely certain by now that there was no difference at all between the three ranks of predictions, aside from the MP used and the length of the countdown that followed. After extensive experimentation with this skill throughout the first world, I felt that I had a clear understanding of how it worked, and could now focus on maximizing its use in battle.
After progressing through the Moogle Village and Bal Castle cutscenes, Delphi made his way to the Hiryuu Valley. This was another area where the Divine Judgment + Eruption combination came in use for most of the random battles, helped by the weakness to fire element of the Zombie Dragons and Bone Dragons. The most annoying enemy in this area proved to be the Drippy monsters, who would break out Magic Hammer and mess with Delphi's carefully chosen MP totals. Anyone who could cut his magic points in half was a foe to be taken seriously. As for the Hiryuu Plant / Dragon Pod at the end, my initial plan was to use repeated predictions to take out the flowers as they spawned. Unfortunately, the predictions took too long to appear and Delphi started getting paralyzed repeatedly and then plinked to death. I found myself wishing for an Angel Ring here, as the Aging status of the first flower kept reducing Delphi's speed down to nothing. The Bone Mail would also have been a godsaver, but of course Delphi couldn't wear it.
Fortunately, the Hiryuu Plant lacks the "Heavy" flag in the code that most bosses possess, and that meant a straightforward one-shot kill:
Fire off the Doom curse and that was that. It was basically the same thing as the Blue magic Condemn spell: when that countdown hit zero, the big plant wilted for good. Of course, by the time that happened Delphi had already been paralyzed and aged. He only had to get off one more prediction to defeat the three little flowers, and fortunately Delphi did manage to make that happen before he was killed by a series of attacks. I'm not exactly sure what I would have done here if this boss hadn't been vulnerable to the Doom curse. As I said, an Angel Ring would have been super helpful to block some of those statuses.
Gilgamesh and Enkidou were an amusing battle for Delphi. Unlike the vast majority of bosses who have a health total that ends in a zero, Gilgamesh has exactly 8888 HP for the ship fight. This meant that he took a huge pounding from the initial prediction that Delphi used; I decided to continue to employ the Divine Judgment + Eruption combo with a Flame Ring still equipped. Eruption has a multiplier of M = 21 for a monster health total ending in an "8" digit, and that meant about 3500 damage from the first prediction, enough to cause Enkidou to appear immediately. Predict almost had enough damage to one-shot Gilgamesh with a max damage Cleansing (which would do about 7500 damage), but since it fell short I didn't try to set it up. Divine Judgment's second property of sucking away magic points also came in handy here, as the blast of Divine Judgment pictured above took away 510 MP from Gilgamesh in addition to the 5000+ physical damage. Gilgamesh only has 1000 MP in this battle so two such uses removed all of his magic and eliminated the chance for Death Claw to show up. However, before that could happen Delphi simply won the battle outright, with Eruption and Divine Judgment both healing him as they dealt damage. Delphi died twice in this encounter, once from Death Claw and once from a Vampire spell out of Enkidou when the monster was almost dead. Otherwise, this was a straightforward battle and not too tricky.
I fought the random encounters necessary to pick up a Wall Ring in the Barrier Tower and obtained the random drop without too much difficulty. It took roughly two dozen battles with Wall Knights to land the item... followed by landing a second Wall Ring before I could climb up to the top of the tower. Randomness at its finest. 
I also picked up the Gold Hairpin from one of the treasure chests in the tower, and I was disappointed to find that it does not cut the MP cost in half when using Predict. I had a lot of plans for using predictions that only cost 2 MP or 4 MP with the Gold Hairpin in place, and the item simply didn't work at all with this particular ability. Very annoying. Delphi also finally maxed out the Oracle job here, which granted him the mostly useless "Read Ahead" ability (reduces the rate of random encounters) and more importantly dropped the timer on Condemn down to 5 seconds. Delphi could now Condemn a target and have the curse timer run out before his action bar filled back up again. This was quite handy and I found myself using Condemn a lot in random battles. While it's too bad that it doesn't work against anything with the "Heavy" flag in the coding, Condemn was continuing to see a lot more use than I had been expecting.
Atomos threw a bit of a monkey wrench into the Divine Judgment/Eruption pattern that Delphi had been using against most bosses. Lacking Flame Rings for the other party members, one use of Eruption would instantly kill the other three characters and foil my normal pattern of offing them one at a time to buy additional time. Therefore I had Delphi fall back on his original Cleansing/Blessing cycle again, which meant that the Blessing predictions would be wasted time where no damage was being dealt. I suspected that there would still be plenty of time for Delphi to win before Atomos sucked up all of the other party members, and I was correct. Each Cleansing was doing around 4000 damage on average, with some of them hitting the boss when it had a low digit HP total and some of them hitting the boss when it had a high digit HP total. I think that Delphi needed five Cleansings in all to get the job done, and Galuf was the only one who had been sucked up at that point. Plenty of time to spare. This battle would actually be pretty doable for the Oracle if there was any way to get the solo character's HP up to about 3000 points. Divine Judgment and Eruption would both heal Delphi back to full health when used, so every character action could have been used for predictions. But the Oracle class has very low HP growth and I wasn't going to gain twenty levels just for this battle, so I followed the normal pattern for solo runs and removed the other three characters one at a time as described above.
After stopping in Mua to pick up slightly better armor, Delphi headed through Guido/Gheedo's Cave:
I wanted to fight all of the Metamorphas due to their chance to drop Light Staffs upon defeat, with the Light Staff being an object that casts the Holy spell when broken in battle. The Metamorphas are an otherwise useless battle to fight, since they do not provide more than a piddling amount of gold and experience. I've often mentioned how this whole dungeon is crawling with them in virtually every random encounter, one of the more puzzling design decisions in this game. To deal with them more speedily, Delphi broke out one of the predictions that he had rarely used thus far: Hurricane, a wind element prediction that hits both enemies and party members alike. Hurricane (8) was located rather conveniently in the prediction rotation in relation to Divine Judgment (5); Delphi could start on Divine Judgment and use a Third Rank (7 MP) prediction to cycle over to Hurricane, and then a Second Rank (3 MP) prediction to cycle back to Divine Judgment, and so on. The Metamorphas have a weakness to wind element and that caused Hurricane to deal double damage. Of course Delphi took damage in the process as well, but I could minimize that by using Hurricane when his current HP ended in a low digit number. This would cause Hurricane to deal about 200 damage and that was easily survivable. As Delphi continued to gain levels and more total health, it was opening up more avenues to employ these self-damaging predictions without suffering instant death. I was planning to get more aggressive with them as Delphi made it further into the game.
My expectation was that Delphi would be using a lot of Eruption in the Forest of Mua to take advantage of the weakness to fire element that many of the tree enemies have in there. Instead, this was an area where Condemn once again saw the most use against random foes, with Doom being the repeated curse of choice. Now that Delphi had maxed out the Oracle class that short 5 second Condemn timer was too good to pass up. Everything in this area was susceptible to Doom, and that meant "one curse, one kill" status throughout the whole forest. There was no need to go through all of the rigamarole of setting up the use of Predict when Condemn was flattening everything with ease. The Forest of Mua also had plenty of Mini Mage enemies who would steal away Delphi's magic points with their Psych/Osmose spell, and that also caused all sorts of trouble with the use of predictions. Normally those spells bounce off a character wearing the Bone Mail but no such luck here. Condemn was faster and safer to use.
The four Crystals at the end of the forest are one of the few bosses in Final Fantasy 5 that lack the "Heavy" flag in the coding:
And that meant they suffered under the grinning skullhead Doom curse as well. Four uses of Condemn, four dead Crystals. I don't think any other class has handled this boss so easily. For this one particular encounter, the Oracle crushes everyone else. 
Exdeath's Castle was a fun area to traverse. There was plenty of danger here, and Delphi was wiped out by a series of Aero 3/Aeroraga spells from the Magic Dragon enemies at one point. His health total was simply too low to absorb multiple castings of that spell, and I needed to run away if there was the possibility of several of those spells landing. However, there was a solution for almost every other situation in the diverse selection of random encounters to be found throughout the castle. Most of the monsters could be dealt with via Condemn:
Even the dreaded double Yellow Dragon battle was no sweat for Delphi. Two uses of Condemn and they were both instant goners, with Delphi walking away 9000 XP richer for his troubles. Red Dragons were an interesting opponent to face, since they couldn't be Condemned and they were immune to fire element damage. Delphi handled them by attacking repeatedly with the Healing Staff, which dealt roughly 700 damage per attack into their undeadness. With access to multiple forms of infinite healing via the Rejuvenation curse and the Healing Staff, they were no problem. Another enemy with the "Heavy" flag were the Dark Wizards, and they had to be countered with various Predictions. Fortunately their starting health total was 1999 HP, which made them easy prey to both Eruption and Divine Judgment. The latter spell in particular always did 9999 damage into their holy element weakness. Overall then, the fights were lively and required some different tactics to puzzle out, with Delphi up to the task in virtually all cases. For a funky underpowered class with weird abilities, Delphi was handling himself quite well.
The final Gilgamesh battle was another place where Delphi relied on his Divine Judgment/Eruption cycle once again. I'd been using that a lot with the Flame Ring still equipped; it was hard to beat the combination of dealing heavy damage while also being healed at the same time. This is an unusual boss fight since Gilgamesh mostly relies on debuffing the party with various nasty tricks. Most of them were useless against Delphi, however: the Sonic Wave/Dischord halving of his level didn't matter because Predict damage is completely independent of the player's level or stats. Gilgamesh's other status ailments didn't matter either, as Predict can be used through Toad status, Mini status, Silence status, etc. I don't know if the programmers simply forgot to have the ability blocked by those statuses or whatever, but you can always use Predict no matter what. That made this an easy battle - or it should have, anyway. Gilgamesh actually has 53k health in this battle and it ends normally after dealing 13k damage, which triggers the dialog sequence where Exdeath banishes him to the Void. However, as I had previously seen with Sol Cannon, the Predict ability doesn't trigger events that are supposed to happen when enemies reach certain HP totals and that manifested itself again here. I had figured that a couple of predictions would be enough to win this battle, and started to grow curious when I was sure that Delphi had done something like 25k damage without the combat ending. Then I realized what was happening, pulled out the Morning Star to bop Gilgamesh once over the head, and that immediately triggered the end of battle text sequence. Whew. Predict is a strange ability.
Now it was time for what can often be a huge stumbling block for solo runs, the Exdeath battle:
Fortunately Delphi had enough tools on hand to make this a relatively straightforward encounter, almost verging on an easy one. The classic way to get a staff user past Exdeath is to farm Light Staff rare drops from the Metamorphas in Guido's Cave; they cast the Holy spell when broken and deal enough damage to get past Exdeath's dangerous first AI routine quickly. Delphi didn't need to use this crutch, however. I went into this duel with a Flame Ring equipped as usual, relying on the combination of Divine Judgment and Eruption once again to injure the boss while healing Delphi each time that a prediction was employed. There were two huge factors working in Delphi's favor here. The first was Exdeath's weakness to holy element, which allowed Divine Judgment to put up astronomical numbers on offense. The second factor was Exdeath's starting health total: 32,768 hit points, which meant that the first Divine Judgment would have a multiplier of M = 27, and then that would be doubled to M = 54 due to the holy element weakness. In other words, an instant 9999 damage from the first usage of Predict every single time. This sliced out a huge chunk of Exdeath's health and made the rest of the battle noticeably easier.
There were still challenges to this boss fight that Delphi needed to overcome. Exdeath's attacks did a lot of damage and the low Vitality of the Oracle class meant that Delphi wasn't working with a ton of health. Twice Delphi was wiped out by a Vaccuum Wave + Hurricane combo when his health was reduced to nil while in HP Leak status, and there was nothing I could do about it. There was also Exdeath's propensity to start 2/3rds of the fights with Condemn, as always, and I was disappointed to find that Delphi's own Condemn ability could not override that ticking death clock with another beneficial curse. There's one other challenge unique to the Oracle class: using Elixirs can be dangerous business, as an Elixir will restore MP to full, and that can knock your character away from the carefully managed last digit of the MP total. Your character winds up getting predictions that they have no intention of using intead of the desired sequence. Regardless though, Delphi was still able to manage this boss fight with a minimum of fuss. Divine Judgment was the prediction that dealt out nearly all of the damage, as Delphi was able to land three predictions that did 9999 damage, 9500 damage, and then another 9999 damage to secure the kill. Eruption was just there to keep cycling back to Divine Judgment again and contributed maybe 4000 damage in total. Delphi won in fewer than ten total attempts, and that was including the ones that were wiped out with Condemn before they could even begin in truth. This was one of the easier solo classes to get past the Exdeath boss fight.
