I'm adding another "Keep reading" link here just in case you accidentally opened the first one. Don't press it if you don't want to keep reading!
Next, it's back to Silph Co. to face Giovanni. This fight is often brutal, because all his team have very high IVs (Pokemon genetics) for some reason. But it can also be trivial because it also seems that he selects his moves at random.
His opening Nidorino is a bit of a problem. It will take two Thunderbolts to KO this thing, and it can choose between Horn Attack (which will take away around 40% of our HP), Poison Sting (15%), and Fury Attack (a semi-random move that can do anywhere from 20%-50% but more likely the lower end). We naturally want Poison Sting or a weak Fury Attack. The plan is to use Thunderbolt, hope to sustain minimal damage, set up a substitute, hope it doesn't get broken, then follow up with a second Thunderbolt from behind the substitute.
That will ensure we have a substitute for when Nidoqueen comes out. It will probably take 3 hits from Hidden Power to bring down the big rabbit, so we really need that substitute in case she goes for a powerful Body Slam or Double Kick which would be game over if they hit. But if the move selection really is random (as it often seems) then there's a good chance she will just go for Poison Sting or Tail Whip, which would be ideal. We can keep setting up a new substitute if she breaks it, as with Koga's Koffing, and hopefully KO her with not too much damage and a substititute still up.
Kangaskhan is more of the same. The marsupial will take 3 hits, and we need a substitute at all times in case of a horrific Mega Punch that could almost one-shot Rémy from full HP, but the kangaroo could equally go for Tail Whip or a weak Rage. The hope is that the odds land in our favour enough times that we can keep resetting the substitute when it breaks. Another Sitrus Berry might give us just enough of a chance to get away with this. The final Rhyhorn is a nearly-guaranteed one-shot with Hidden Power.
It's not a neat strategy, far too much can go wrong, but when you're against a very strong opponent who seems to have no move-selection logic whatever to take advantage of, it's very hard to play around. In fact, the strategy goes out the window immediately when a critical hit against Nidorino leaves us exposed to Nidoqueen without a substitute up. Thankfully she makes the poorest move selections imaginable, and Kangaskhan does little better, so we manage to win on our first try without really breaking a sweat. Weird battle.
Rémy demonstrates his skill
OK, now it's time to reveal why this whole previous section of the game was played underlevelled. It's because at Level 41 Rémy learns the move Thunder by level up. It's a really powerful Electric type move that I am going to need imminently, but I also needed to learn it specifically by level up so I could replace it later with another move and then relearn it even later using the TM available from the Power Plant (I am always planning ahead!). But, I needed both Thunderbolt and Hidden Power for the Giovanni battle you just saw, so I couldn't get to level 41 before that otherwise I would've had to overwrite one of those moves while I still needed it! A complicated strategy, let's hope it pays off.
We will find out against Sabrina. Thunder has replaced Thunderbolt, and we have also replaced Hidden Power with Rain Dance (the TM can be found on route 15), a move which makes it rain on the battlefield for 5 turns. Thunder has an accuracy of only 70%, but in the rain it always hits. We also got a Magnet from a wild Magnemite at the Power Plant, which slightly boosts the power of Electric type moves. That will be really necessary here, because Sabrina's team is really powerful.
Here's the plan. Her first Pokémon is Kadabra. On the first turn it is very likely to use Calm Mind or Reflect, two set-up moves that don't really matter to us, but there's a slim chance it uses Future Sight which would be really annoying. (Side note: in many hundreds of battles against this Kadabra, I have never seen it select Psybeam on the first turn. I think the AI gets confused with Future Sight and sometimes considers it to be a status move). We can just set up the rain with Rain Dance. Then our Thunder is going to one-shot that thing no problem.
Mr Mime is significantly bulkier, with the Magnet equipped, Thunder is around 50/50 to get the one-shot, we hope it does. If it doesn't we should still be fine because the clown will typically use a pointless Barrier to needlessly boost its Defence, the issue is more a question of wasting precious turns of rain - if we narrowly miss the kill, we will be stuck in a healing loop with Sabrina spamming healing items on her dying clown as the rain fades, at which point we'll have to set it up again, maybe manageable but definitely suboptimal. The Venomoth which follows is much weaker, and a single Thunder will definitely finish it off.
If all goes to plan, Alakazam will appear with one rain turn remaining. Here, the (ridiculous and overcomplicated) strategy comes in. Although we are at full HP, Alakazam's Psychic has a decent chance to straight-up one-shot little Rémy. We want Sabrina to simulate her turn, see that kill, and go for it. The plan is to do a massive amount of damage with Thunder, while Alakazam's Psychic rolls slightly lower than the simulation and leaves Rémy with a sliver of HP. You know what's coming. Sabrina will use the next turn to heal her Alakazam, giving Rémy two turns in a row to use a maximum power Reversal and finish the job, nevermind it being "not very effective". Now that would be a delicious meal from the little chef!
There were really too many things that could go wrong with this plan, though. We're wrong-footed immediately when Kadabra goes for an unlikely Future Sight instead of a status move. Then Thunder gets a low roll twice on Mr Mime, forcing us to stall turns with Substitute while tanking the damage from the Future Sight attack. We're so far off-script now there may as well not be a script. Finally we're able to reset the rain and get the favourable roll on Mr Mime.
We're back on script, except that we were supposed to have full HP at this point, wherease we have barely 1/4. Venomoth goes down easily as expected, but after our first Thunder does solid damage we are sure to get destroyed by Alakazam's Psychic...unless it selects Future Sight for some reason. Sabrina has already wasted her two Hyper Potions on her clown, so we don't even need to worry about Future Sight, another Thunder gets the win on our very first attempt at this battle. I guess sometimes all the planning in the world is not good enough, but you can still win if you keep calm and pray your opponent makes a silly mistake.
Ok, that Sabrina battle made me wonder if there is any point making a plan at all. But next up is Blaine, who can be easy if you have a Surf user, and can be near-impossible otherwise. We are in the latter category, so we really will need a plan of some kind.
Here's what I came up with. Blaine's opening Growlithe will always go for Fire Blast, because it's by far its strongest move, doing roughly 2/3 damage. But Pikachu is faster and can set up Rain Dance after Growlithe has selected its move but before the Fire Blast hits. Rain halves the damage from Fire type attacks, so once the rain is up Take Down would've been a better option from Growlithe. A single magnet-boosted Thunder apiece will then take out Growlithe, Ponyta and (50% of the time) Rapidash.
The real problem is Arcanine, whose Fire Blast is a guaranteed one-hit KO on even a full-HP Rémy. We have a single turn of rain left when it comes out, which means we can do solid damage with Thunder before being hit with a Take Down (its best move in the rain). There is a slim chance that Take Down just KOs the fragile Rémy after Growlithe's Fire Blast damage, but far more likely Rémy survives with a sliver of HP, and between Thunder and Static (Pikachu's ability) we actually have an over 50% chance of paralyzing Arcanine on this turn. If Arcanine gets paralyzed, Blaine will take two healing turns, one to cure the paralysis, one to replenish the HP (he does not have a Full Restore to do both at once) giving us time to reset the rain and hit two 100% accurate Thunders. If Arcanine does not get paralyzed, Blaine will take one healing turn, in which case we must select a low-HP Reversal which will do solid damage (in spite of the two attack debuffs from Growlithe and Arcanine's Initimidate ability) then pray we can hit a 70% accurate Thunder to get the KO.
All seems to be going to plan, until we get to Arcanine where for once the plan fails in a beneficial way - Arcanine gets one-shot by a critical hit Thunder, sparing us all the drama. Lucky!
Rémy serves an early main course
Next up is the final gym battle against Giovanni. As the Ground type gym, this is likely to be Rémy's worst match-up of all. And they said Brock was impossible... Is there any way at all for this little rat to scrape a win? It took a long long time to come up with a possible strategy, but I think it may be one of the best things I've ever thought up.
First we must prepare. We go with Bill to the Sevii Islands, where Rémy can visit the move relearner to reteach Thunderbolt in the place of Thunder (we still have the TMs for both of these moves, remember, and several Hidden Power TMs from Meowth too, one of which we teach over Rain Dance). While there, we also visit Mount Ember, where you can find wild Machoke which, if you're lucky, might be holding a Focus Band, a held-item which has a 10% chance of keeping you alive at 1HP if you are hit with an attack that would ordinarily KO.
The issue with Giovanni's team is not just the type matchup, although everyone on the team does know the powerful Ground type move Earthquake, and thus all of them can one-shot Rémy from full HP. Actually the bigger issue is that neither Nidoqueen nor Nidoking can be one-shot by any move Rémy can learn. We thus somehow need to have a substitute up before they are sent out.
The plan is to start the battle in a very unusual position, with Rémy slightly underlevelled at level 49, and with only 55 HP. Why? Well, because Giovanni's first Pokemon is a Rhyhorn which knows Earthquake. At 55 HP, we are just in the zone where Take Down might also KO Rémy (it has a 37.5% chance of doing so), which means Rhyhorn might select that move instead (if the AI detects that two moves can get the KO, it will choose randomly between them). And because Take Down has 85% accuracy, we absolutely want Rhyhorn to select it, because, unlike Earthquake, it could miss.
Why level 49 instead of 50? Well, at level 49, Rémy has max HP of 111. Substitute takes 1/4 of your max HP rounded down, that's 27. That means Rémy can use Substitute twice and still have 1 HP left. At level 50, Rémy has a max HP of 113, so Substitute would take 28 HP, killing Rémy on its second usage. And if we accounted for that by starting the battle with 57 HP, we would be completely out of Take Down kill range, so Rhyhorn would always use Earthquake. It is such a delicate balance, but starting in a weaker position at 55HP and level 49 is the only way to get two attempts to set up a substitute, which we really need.
By my calculations, there's a 9% chance for Rhyhorn to attempt and miss a Take Down on one of the two chances it will get to attack. If it misses, we can use Hidden Power to one-shot the monster while our substitute survives. We will either get past with 1 HP or with 28 HP, depending on whether it missed its Take Down on its first or second attempt. (We level up after defeating Rhyhorn turning this into either 3 HP or 30 HP subsequently).
The frail Dugtrio will easily fall to Hidden Power. Then, the reason we ran through all that nonsense against Rhyhorn is so we could have a subsititute up when Nidoqueen comes out. Hidden Power will kill the rodent queen in two hits, giving her only one chance to attack, and since we have a substitute up we will not get KO'd by her attack. The problem? After her comes Nidoking, who also takes two hits to KO, and now we don't have a substitute any more. What can we do to avoid getting steamrolled by an Earthquake from the giant rabbit? Well, there are three options:
Get a critical hit with Hidden Power against either Nidoqueen or Nidoking. That's about a 12% chance.
If we get past the Rhyhorn with 3 HP rather than 30 HP, then both Nidoqueen and Nidoking will see a kill with Poison Sting as well as their two other attacking moves. If they select this then the substitute will not get broken, because it is such a weak move (why does the 8th gym leader even have it???). Net, that's about a 20% chance.
If Nidoqueen breaks the subsitute, then Nidoking will attack an unprotected Rémy. Since we are holding the Focus Band, there is a 10% chance that attack will not kill but leave Rémy with 1 HP.
So, ultimately it is actually fairly probable that one or other of these things might happen, allowing us to get to the final Rhyhorn. If we have 30HP still, we can set up a substitute to safely get down to 2HP, a substitute which will be promptly destroyed. At which point we are perfectly free to fire off a max power Reversal, KOing the weird rock creature just like Brock's Geodude that started this whole adventure off. That's a neat bookend!
And guess what? Another plan is executed perfectly, and it materialises on only the second attempt. I am honestly really pleased with this specific strategy, because I genuinely thought this battle would be a true brick wall for Rémy!
It's nearly time for the final gauntlet, the Elite Four. But first there's another battle against our rival Edgar. On paper, a single Pikachu has no chance against his team, but in fact we can quite easily plan a strategy with a 50% win chance.
His Pidgeot is comfortably one-shot by Thunderbolt, thought it is fairly likely to do some damage with Quick Attack first. The Rhyhorn will go down to a single hidden Power. Exeggcute's only attacking move is SolarBeam, a move that takes two turns so, even if selected, we will have set up a substitute and KO'd the eggs with a Hidden Power before they can attack. Alakazam is the reason why that substitute was necessary, because its Psychic would comfortably kill Rémy if he'd been softened up by Pidgeot's Quick Attack meaning the AI would always go for it. Instead, the Psychic will only destroy the substitute, leaving us free to KO Alakazam over two turns. Then, with no substitute up against Charizard, it's down to a coin-flip: a magnet-boosted Thunderbolt has exactly a 50% chance to slay the dragon.
The plan works perfectly, and even a turn-one critical hit Quick Attack from Pidgeot can't derail it. To be clear, according to my rules the level caps are based on the highest level Pokémon of the gym leaders and Elite Four member Lance, so the level cap for this battle is actually 60 to match Lance's Dragonite, not the 53 that Rémy stuck to. But sometimes it's more fun to give yourself an extra challenge!
The Elite Four is next, and that's going to be tough for a little rat. Thankfully, we will for once have something of a level advantage initially. That's because the final five battles (the Elite Four plus the Champion) must be fought back-to-back with no training available inbetween, and since the five battles all have increasingly high-level Pokémon it makes it a bit tricky to establish how we can keep up with the level cap. The standard way this is done in Hardcore Nuzlockes and similar Pokémon challenges is to start the Elite Four at the level cap of the fourth member, which means you will level up during the battles to be approximately at the same level as the Champion when you get there. It may look a bit odd seeing Rémy at level 60 when Lorelei's Lapras is only level 54, but this is why, and the advantage will be absolutely essential for the first few battles.
Lorelei's Dewgong, Cloyster and Slowbro will not be a problem for Rémy, they are all Water types and will be comfortably one-shot by a magnet-boosted Thunderbolt. The problems are her Lapras (bulky) and Jynx (not weak to Electric moves). These two will be sent out back-to-back and will both take two Thunderbolts to bring down, meanwhile they can both do over 50% damage to Rémy or ruin the battle with status conditions. And the fact that they are back-to-back means a Substitute set-up earlier in the battle cannot save us - it's superficially a similar situation to the Giovanni gym battle.
However, if we plan very carefully we can actually get a one-hit KO on both of these monsters. It just requires a little bit of preparation. First, a trip to the Sevii Islands' move relearner to relearn Thunder over Hidden Power (we still have plenty of Hidden Power TMs and the Thunder TM from the Power Plant). Then a journey to Viridian Forest to get poisoned by a wild Weedle and get our health bar down to 1HP.
A magnet-boosted Thunder has an 81.25% chance to get the one-hit KO on Lapras if it hits, which combined with its 70% accuracy give a 58% chance of success. We need the one-hit KO because at our very low HP Lapras is guaranteed to use an attacking move and all of its moves have 100% accuracy (there is also the 1% chance that Lapras is not KO'd but gets paralyzed by Thunder, preventing its attack, which would be a valid but very unlikely path to victory). Then, predicatably, a single low-HP Reversal will destroy the frail Jynx. Overall, that's a 59% win chance, pretty good odds honestly, and we get past on our second attempt.
Bruno is next, on paper this seems genuinely impossible given his two Earthquake-using Onix plus his three strong Fighting types that are not weak to any of our moves, and all of which can do massive damage to Rémy with their attacks. The plan, once again, takes advantage of quirks of the AI move selection.
First, we'll restore our HP back to full and teach Hidden Power again over Thunder. The two Onix will be easily dispatched with a Hidden Power apiece. Bruno's Hitmonchan's strongest attack is Sky Uppercut, but at most that will do 87% damage so it is unlikely to select that move on turn one. Far more likely are moves that the AI seems to interpret as status move-adjacent: most likely Counter (which only damages us if we used a Physical move on our turn, and Thunderbolt is classified as Special, not Physical), but also possibly Rock Tomb (which additionally lowers speed but is only 80% accurate). We need to see Counter or a Rock Tomb miss, which will allow us to two-shot the Hitmonchan at full HP. This is actually very likely to happen, because Counter seems to be so strongly favoured by the AI.
Against the Hitmonlee, things get a biiit more complicated. In itself, it is trivial to get past, either by just using two Thunderbolts (none of its attacks can one-hit us, so we would survive) or by setting up substitutes over and over, letting it break them each time, to get down to 2HP, and then KOing with a max power Reversal. But we really need a substitute up for Machamp, so we can't just do that, we have to be really careful here. Things are about to get mathematical! (Please feel free to skip past the indented section if you're not interested! I have included it only to demonstrate how in-depth I go planning these battles)
Hitmonlee knows Mega Kick, which (as we know) is only 75% accurate, and will do between 106HP and 125HP damage to us depending on the RNG roll. It also knows Brick Break, which is 100% accurate and does between 100HP and 118HP of damage to us depending on the RNG roll. Plus the 100% accurate Facade which does 62HP-74HP. This is a problem for us because the possible damage ranges for Mega Kick and Brick Break overlap, and that makes the AI's move selection less obvious.
I made several spreadsheets for this one battle to try to predict the AI as best as possible. By my calculations, there is a 195/256=76% chance that the AI will predict that Mega Kick is a stronger move than Brick Break, and a 9/256=4% chance that it will predict they are the same strength. When we have high HP, that makes for a 78% chance it will choose to use Mega Kick and a 22% chance it will choose Brick Break. As our HP gets lower, this will tend towards 50/50 as the AI sees that both moves will get the KO. Between 75HP and 100HP, it will be exactly 50/50. At 74HP it starts to have a chance to select Facade, and at 62HP and below it is a three-way probability tie between all these moves.
Why do I need to know all this? Because I will keep setting up substitutes that Hitmonlee will immediately break, unless it misses its attack, at which point I can attack. I need Hitmonlee to miss twice so I can KO with two Thunderbolts, or else I need Hitmonlee to miss once but specifically at the time when I have set up my fourth substitute, at which point I will have exactly 2HP and can one-shot it with a max power Reversal.
Hitmonlee will choose its attack based on how much HP I have. Because I will only lose HP via Subsitute, there are only a few specific HP values that I can have: 138HP, 104HP, 70HP, 36HP, and 2HP. The table of pobabilities works out like this:
HP Mega Kick chance Miss chance
138 78% 20%
104 63% 16%
70 45% 11%
36 33% 8%
2 33% 8%
If you work out allllll the different possible paths, you will find that we will get past Hitmonlee with substitute up 20% of the time: 13% of the time at 36HP or above, and 7% of the time at 2HP.
The HP we have for Machamp matters because Machamp is holding a Sitrus Berry. The Sitrus Berry restores 30HP once Machamp's health bar drops below 50%. We really want to avoid that happening, because there is simply no way to two-shot a Machamp with 30 extra HP, we can barely manage just the Machamp! So we need to be able to hit Machamp for under 50% of its HP on turn 1, and over 50% on turn 2. Can we do that?
Well, both Thunderbolt and a max power Reversal do around 50% damage, sometimes a little less, sometimes a little more, and Reversal is slightly stronger. Two Thunderbolts have a 31% chance to KO, Thunderbolt plus Reversal has 68% chance, and two Reversals a 94% chance. So Reversal seems the best bet, right? But we can only use two Reversals if we got past Hitmonlee at 2HP! If we have higher HP we cannot use Reversal on turn 1. Furthermore, we haven't taken the Sitrus Berry into account.
Thunderbolt+Reversal has a 68% chance to KO, but only a 33% chance to KO while not triggering the berry. Two Reversals has a 94% chance to KO, but only a 14% chance to KO while not triggering the berry. Aaargh, I'm getting a headache!
OK, so that settles it, Thunderbolt plus Reversal against Machamp it is. But just in case everything goes wrong I gave Rémy the Focus Band he held during the Giovanni battle. When it comes to actually executing this strategy, everything amazingly actually does go to plan, except that the Thunderbolt against Machamp got a really low RNG roll, so we miss the KO. The Focus Band saves us though, and after Machamp is healed we can demonstrate how it was supposed to go. Ugh, that was a lot, let's move on.
OK, we may be about to be punished for the Double Team spam against Brock on our first attempt with Roquefort. Agatha's lead Gengar loves to use Double Team, and many are the times I've lost to her because of being unable to hit Gengar with any move. Let's hope it doesn't happen this time...
The Gengar also knows Toxic and Confuse Ray, two status moves we really can't risk being hit by, so we absolutely must set up a substitute right away. Two magnet-boosted Thunderbolts will easily KO this thing, but we have to be careful when Double Team is involved. Thunderbolt is more likely than not to put Gegar below 25% HP, which will cause Agatha to use a healing item. We would then have to hit Gengar three times through however many evasion boosts it has set up by then. It is much safer to use Hidden Power first, which doesn't risk causing a heal, then bring out the Thunderbolt on the next turn.
Golbat is easily KOd by Thunderbolt, but Arbok will take two hits. After having set up the substitute against Gengar, we will be at 104HP, which means there's a 75% chance the AI will simulate a KO with Sludge Bomb and threfore select that move. If it doesnt see the kill, it will pretty much always use the status move Screech. We naturally prefer Screech, but we can handle having our substitute destroyed here, and it is more likely to happen that way.
The higher-level Gengar will be sent out next. If we have a substitute, we can just KO it over two turns with Thunderbolt, but if we don't we need to set one up. It should be OK: we will have levelled up and have 106HP here, which is enough health that the AI will never simulate a KO with any attacking move, thus it's very likely to select Hypnosis which cannot affect us behind a substitute. Two Thunderbolts will bring it down from there, and the final Haunter only needs a single hit.
Everything goes to plan on the second attempt of this battle - on the first the intial Gengar used Double Team so much that Remy missed ten times in a row! That shows me how Brock felt earlier...
Rémy takes on the head chef!
I am going to group the battles against Lance and the Champion together, because they are typically both incredibly hard in any challenge run. Here, I am going to be resorting to a little bit of what I previously called "cheese". No, not Double Team (never again, I hope), but rather I am going to be using Leftovers (the held item that restores 1/16 of your max HP each turn) for these battles. I have no rule against Leftovers, I just prefer not to use it because often that restriction makes battles more interesting. Here, it was actually the opposite: I could get by relying on lucky RNG falling my way, or I could have a more interesting and reliable strategy that used the Leftovers. I will always take the more interesting path to victory!
The reason I consider the Lance battle so difficult is that there are three Pokemon in his team that I cannot one-shot and who can all do massive damage to me. It's a similar situation to Bruno, except there my ability to one-shot his Hitmonlee if necessary was a key part of the plan.
Lance will first send out Gyarados, a single Thunderbolt will make quick work of it. Next will be come the first Dragonair. It will not see the KO with any of its moves, so it is very likely to go for a status move, either Safeguard (which I don't mind) or Dragon Rage (which is obviously not a status move, but for some reason the AI often treats it like one). Safeguard is ideal, but Dragon Rage always does a predictable 40HP damage so we should be able to play around it. Two Hidden Powers will KO this noodle.
The second Dragonair will be sent out next. This one knows Thunder Wave instead of Dragon Rage. If we were hit with Dragon Rage against the previous noodle, we will have had at least two healing turns from Leftovers leaving us with 116HP at a minimum. Otherwise we should be at full HP. The highest damage roll for Hyper Beam (this noodle's strongest move) is also 116. If we are at 116, there is a very slim chance it goes for that move immediately. Otherwise it will almost certainly go for Thunder Wave. It's essential, therefore, that we set up a substitute immediately.
That substitute could put us anywhere from 89HP to 113HP at the start of the next turn (depending what happened against the first Dragonair). At the lower end, this is squarely within Hyper Beam kill range, at the higher end it's a 19% chance for the AI to see the kill. The higher end here is much more likely, so there is a good chance we just see another failed Thunder Wave and KO with two Hidden Powers. At the lower end of our possible HP range, we are probably about to see our substitute destroyed by Dragonair's Hyper Beam. That's actually OK though, because a Pokémon who uses Hyper Beam cannot attack on the next turn, allowing us the chance to reset the substitute and KO with a second Hidden Power.
That will bring us to Dragonite with a substitute up. Dragonite's Hyper Beam will always one-shot Rémy, even at full HP, so we absolutely need the substitute. Since we have it, we can safely KO the Dragonite over two turns, seeing our substitute destroyed. That leaves Aerodactyl, a Pokémon so fast that I really wasn't expecting to outspeed it, but that timely level-up after defeating Dragonite will give us just enough of a boost to our Speed stat to let us outspeed by a single point - that's the payoff of all those level 2 Pidgeys coming good once more. A sinlge Thunderbolt will see of the fossil, giving us an effectively guaranteed win.
In fact, the battle starts off unluckily with a Dragon Rage from Dragonair 1. Then luck swings the other way with a Hyper Beam miss from Dragonair 2. Ultimately though, it didn't matter, this battle was effectively unloseable with the strategy outlined above, which is amazing at this stage in the game.
(A little aside about what Leftovers did for use here. Without them we could easily have been at 100HP initially against Dragoniar 2. It would almost certainly then use Hyper Beam turn 1 destroying our first substitute leaving us at 65HP, we would then have to set up another substitute leaving us at 30HP, and we still wouldn't've attacked Dragonair 2 yet. At 73HP or below, Dragoniar 2's other attacking move Outrage has a chance to one-shot, so we could no longer rely on Hyper Beam's recharge turns to get the win. Leftovers allows us to carefully manage our HP range, ensuring we can be within the ranges of certain attacks and not others, we can delay and use substitutes to get our HP lower and higher as needed, and in this case I think it made the battle far more strategic.)
OK, it's time for the final conflict. Edgar re-appears at the last, and his team has had a little upgrade since our last meeting: both his Rhyhorn and Exeggcute have evolved into their final forms, and they're both bad match-ups for Rémy. How are we going to handle it? Well, first we're going to have to say goodbye to the move that got us this far: Reversal has no utility for us in this battle, so it will be replaced by Thunder (we still have the TM from the Power Plant). It's sad to see it go, but it's all about getting this last win.
First out will be Pidgeot. This thing is a comfortable one-shot with Thunderbolt, but it will be followed by Rhydon and we really need a substitute up. On turn 1 it is most likely to use a status move, Sand Attack or Featherdance, but it will also sometimes use Aerial Ace which has a 31% chance to not break our substitute. Overall, we have a very good chance to get by with a substitute up.
Rhydon is next, and it could KO Rémy several times over with a single Earthquake, so we needed that substitute for sure. Two Hidden Powers will see it off, but we will be left unprotected for the Alakzam that comes out next.
Alakzam is a terrifying prospect, and we have no chance to one-shot. We should be at exactly 139HP at this point, and Alakazam's Psychic has a highest possible damage roll of ... 138. This very neat fact makes Reflect or Future Sight more likely here, and of course we would prefer Reflect. Future Sight may be manageable too but it would require a lot of luck against Exeggcutor. We can KO the Alakazam over two turns with Thunderbolt.
Next will be Exeggcutor. If everything has gone to plan, we will be at full HP without a substitute, and since its best move Egg Bomb has a maximum damage roll of 101 (71% of our max), and because the AI favours status moves on turn 1, we are much more likely to see an attempted Sleep Powder. A new substitute will block that.
That will leave us at 115HP at the start of the next turn, out of KO range of any attack, but on turn 2 any move is possible. Sleep Powder would be ideal, since our substitute would block it, and two Hidden Powers would enable us to get past with a substitute up at full HP, which would be awesome. Egg Bomb would merely destroy the substitute, allowing us to KO with two Hidden Powers. Giga Drain is risky: it would destroy the subtitute, but in doing so it would also restore 17HP to Exeggcutor after our first Hidden Power, and that's just enough extra HP that our second Hidden Power is no longer guaranteed to get the KO, it's an 86% chance though, which is still good. The worst move to see by far, though, is Light Screen, a status move that halves the damage done by Special moves (that's all Rémy's attacking moves) for 5 turns. If that happens, there is honestly no point continuing the battle, we will likely not even be able to KO the Exeggcutor. We just hope not to see it.
Gyarados is barely worth mentioning, it will go down to a single Thunderbolt. That brings out Charizard. If we were lucky enough to see two Sleep Powders against Exeggcutor, we will still have our substitute up, so we will be able to KO over two turns with Thunderbolt. If not, we are dangerously exposed and need to take a risk for the one-shot. We added Thunder to our moveset before the battle for this very scenario. It is a guaranteed one-hit KO against Charizard, but with only 70% accuracy it is a risky option. However, we should be at virtually full HP, meaning Charizard is likely to select the only move that will KO us, Fire Blast, a move with 85% accuracy. Taking into account the possible seuqnces of misses (Thunder miss followed by Fire Blast miss followed by Thunder miss etc etc) it actually amounts to a 73% chance to get the win, even if we have to use Thunder.
This plan required luck to go our way on a few occasions, but when it came to executing it, we got a lucky break I wasn't expecting: a critical hit on the Rhydon made us face Alakazam a turn earlier than expected putting us within Psychic kill range and with a substitute up. That is brilliant, because there is no risk of Future Sight screwing things up later on, and we actually arrive at Exeggcutor totally back on track. Exeggcutor obliges by using the double Sleep Powder, the best possible option, and from there the win is guaranteed because with a subtitute up, Charizard is a comfortable KO with two Thunderbolts.
That was back-to-back first-try victories against Lance and the Champion, really nice to see!
So, we've done it. Little Rémy has proved that it really is possible for a Pikachu (albeit not an ordinary Pikachu) to beat this game under the absurdly harsh restrictions I made at the start. And I think the rat can be proud of its achievements. Some of the strategies it pulled off were really cool.
It was frustrating that Roquefort (a much more ordinary Pikachu) could only manage it if I edited the game to make a couple of overworld trainers non-mandatory (not because they were hard to beat, but because they gave us too much experience). In fact, Pikachu was actually really lucky here. The Pokémon games have something called Experience Groups, and which Experience Group a Pokémon is in determines how fast it levels up. Pikachu was actually in one of the groups which levels up more slowly at the start of the game, which made Rémy's run possible, even if Roquefort's technically wasn't. If I had tried this with Bulbasaur, for instance, it would have been totally impossible. Even facing Brock immediately at Level 6 and fighting the fewst battles possible, Bulbasaur would still be overlevelled for the battle against Erika.
If you enjoyed reading this (and if you got this far I hope you did) you may be pleased to know that I have done similar runs with Bulbasaur, Charmander, and Squirtle. In those, I had a particular strategy to get past the levelling rate issue, which I think works really well. I have also done the same in Pokemon HeartGold with Pichu, Chikorita, Totodile and Cyndaquil: that's a much harder game and those runs required some really intense planning.
I think what the experience of writing all this out has shown me is that I really need to publish this in the form of a YouTube video, rather than a longform text post that's totally wrong for this platform lol. Anyway, thanks for reading and I hope you enjoyed it!
