Hello. This topic is about a challenge run that I did for Pokemon Blue. I intend to do one for each Pokemon game in chronological order, although that may take a while. The run has been recorded and uploaded onto Youtube, here is the link:
The main purpose of the topic is to provide a more detailed explanation on the process of developing the run than would be possible to do so with CC or a video description. With that in mind, here's the basic outline of the goal I set out with routing the run: Using a minimal reliance on RNG, complete the game in as few combat turns as possible. Of course, in-battle items outside of Balls are also banned. Also, no party members are to be KO'd. I didn't want to speedrun the game because of the heavy reliance on RNG, the abuse of in-battle items, and the general lack of strategy involving party construction, so this is the compromise I came up with.
Regarding what constitutes as minimal RNG, this is the guideline used for the purpose of the run.
With the rules established, here is the party I used.
Final level: 64
Earthquake
Blizzard
Hydro Pump
Surf
Starter. Used in all major battles except for Lt. Surge. Squirtle is selected as the starter over Bulbasaur and Charmander mainly due to access to Bubble Beam after defeating Misty. Having a 65 STAB move at this point of the game is very powerful, especially when Ivysaur and Charmeleon are working with 35/40 BP STABs. From there on, Wartortle steadily becomes stronger with access to better TMs (Dig, Body Slam, Ice Beam) and later on Surf and Blizzard as Blastoise. In general it's a very balanced Pokemon with steady growth from start to finish.
Final level: 57
Agility
Drill Peck
Thunderbolt
Thunder
Along with Blastoise, Zapdos is the only other Pokemon used until the end of the run. Zapdos is used for Rival 6/7/final, Sabrina, Lorelei, and Lance. Zapdos is caught before Rival 6 (Silph Co.) and the Master Ball because it would take fewer turns to catch it and use it for that battle than grinding Blastoise until it can solo Rival 6 would. Zapdos' Thunderbolt nails Water types while Drill Peck covers the Rival's Venusaur and anything with high Special or a resistance to Electric. This gives Zapdos excellent coverage for Blastoise's shortcomings.
Final level: 7
Used for Rival 2 (Route 22). Catching a Spearow for that battle to use against the Rival's Bulbasaur takes fewer turns than grinding Squirtle until it can solo the fight consistently. After that, its only use is being traded for Farfetch'd for HM utility.
Final level: 32
Used for Lt. Surge, Rival 5 (Pokemon Tower), and Koga. Caught to save turns against Lt. Surge, whose Raichu requires quite a bit of grinding for Wartortle to otherwise beat. A 100 BP STAB in the form of Dig along with having a high catch level of 31 is excellent, but it doesn't offer much else past that, and Wartortle needs to hog all the EXP it can get for the end of the game, so Dugtrio doesn't see much use. Only other thing to note is that catching a level 31 Dugtrio is an absolute pain due to being a 1% encounter rate, the level 29 ones are a 4% encounter rate and much more easily encountered. Having to use the Repel trick isn't too bad, the longest drought I had resetting for DVs was 15 minutes of no encounters.
The above 4 are used for battle. Farfetch'd (from the Spearow trade) and Lapras are used for HM utility.
And lastly, notes on major battles.
I think that's about all there is to say about it. IGT at completion was 3 hours 56 minutes, the total turn count was 483. The goal was something of a showcase of the game being played "efficiently" while relying minimally on luck and in-battle items, and in that regard I feel like I was able to accomplish that satisfactorily. The next project is Pokemon Crystal, but it'll take a long time to finish that due to juggling other things. I hope that some people here might actually find this interesting.
Regarding what constitutes as minimal RNG, this is the guideline used for the purpose of the run.
1) Assume that the AI always makes the most unfavorable decision it can make. This rule was formed on the premise of another Pokemon-style game that I played that genuinely did have AI that read player inputs. For Pokemon Blue, the AI is bad enough that this rule is basically overkill to have, but it's still a good guideline to avoid losing due to bad luck.
2) Assume minimum damage ranges on damage done by the player, and maximum damage ranges on damage done by the opponent. This is so as to not rely on damage ranges for success. This isn't relevant to this run, but the rule is more precisely the least favorable damage range, for instances where I might need to hold off on KOs or rely on something like Torrent (and other similar abilities) ranges.
3) Assume that any attack by the player with an accuracy less than 90% misses, until the odds of landing the attack exceeds 90% (that is to say, a 70% accuracy attack used twice has a 91% chance of connecting once, so I assume it misses every other attack, starting with the first). 90% is used because that is the highest accuracy available to all attack types, Rock capping out at 90% for Rock Slide. I also feel that 90% in general is a sufficiently high probability to assume it will work.
4) Assume that any attack by the AI with an accuracy over 30% always hits. Mostly irrelevant to this run, but it may be important in later runs. Basically, I'm not ever counting on something like Hydro Pump from the rival's Gyarados missing. 30% is low enough that I feel that I can assume it'll miss, but OHKO moves are still scary. They are a nonissue in this particular game though.
5) Similar to the above rule, secondary effects with a greater than 30% chance of occurring are assumed to always occur. This is almost entirely irrelevant to the run, but may come up in later games.
6) Status effects (namely Confusion and Sleep) are assumed to have the maximum duration, Confusion is assumed to always inflict self-damage. Basically always try to avoid these status effects. This assumption is only at play for major battles, as otherwise I would have do things like grind until I can OHKO Zubat and the Goldeen in Cerulean Gym and that's pretty dumb.
7) What isn't assumed is critical hits, outside of boosted critical hit moves. If I were to assume these were in play then there would be a lot more grinding to avoid KO ranges.
8) Also not factored in is accuracy reduction, as if I factored this in then I would have to do things like grind for Rival 2 until I can OHKO his Pidgey and that's just really dumb and painful.
9) Because of the tremendous amount of randomness to capturing, I routed captures as taking the # of turns equivalent to their catch rate at full health using a Poke Ball, added up until it reaches/exceeds 100%. For most Pokemon, this is merely 3-8 turns. For legendary Pokemon with a catch rate of 3, this is a whopping 63 turns (1.6% catch rate at full HP using a Poke Ball). Of course, in practice the battle wouldn't even last anywhere close to that long, that's just for routing purposes. This formula is accurate to modern games, but not to gen I. I tried to manually calculate it and was unable to get accurate results.
10) All Pokemon used are assumed to have a DV of 7, and for the purpose of the run, I reset with each Pokemon acquired until I get stats that correspond to this.
2) Assume minimum damage ranges on damage done by the player, and maximum damage ranges on damage done by the opponent. This is so as to not rely on damage ranges for success. This isn't relevant to this run, but the rule is more precisely the least favorable damage range, for instances where I might need to hold off on KOs or rely on something like Torrent (and other similar abilities) ranges.
3) Assume that any attack by the player with an accuracy less than 90% misses, until the odds of landing the attack exceeds 90% (that is to say, a 70% accuracy attack used twice has a 91% chance of connecting once, so I assume it misses every other attack, starting with the first). 90% is used because that is the highest accuracy available to all attack types, Rock capping out at 90% for Rock Slide. I also feel that 90% in general is a sufficiently high probability to assume it will work.
4) Assume that any attack by the AI with an accuracy over 30% always hits. Mostly irrelevant to this run, but it may be important in later runs. Basically, I'm not ever counting on something like Hydro Pump from the rival's Gyarados missing. 30% is low enough that I feel that I can assume it'll miss, but OHKO moves are still scary. They are a nonissue in this particular game though.
5) Similar to the above rule, secondary effects with a greater than 30% chance of occurring are assumed to always occur. This is almost entirely irrelevant to the run, but may come up in later games.
6) Status effects (namely Confusion and Sleep) are assumed to have the maximum duration, Confusion is assumed to always inflict self-damage. Basically always try to avoid these status effects. This assumption is only at play for major battles, as otherwise I would have do things like grind until I can OHKO Zubat and the Goldeen in Cerulean Gym and that's pretty dumb.
7) What isn't assumed is critical hits, outside of boosted critical hit moves. If I were to assume these were in play then there would be a lot more grinding to avoid KO ranges.
8) Also not factored in is accuracy reduction, as if I factored this in then I would have to do things like grind for Rival 2 until I can OHKO his Pidgey and that's just really dumb and painful.
9) Because of the tremendous amount of randomness to capturing, I routed captures as taking the # of turns equivalent to their catch rate at full health using a Poke Ball, added up until it reaches/exceeds 100%. For most Pokemon, this is merely 3-8 turns. For legendary Pokemon with a catch rate of 3, this is a whopping 63 turns (1.6% catch rate at full HP using a Poke Ball). Of course, in practice the battle wouldn't even last anywhere close to that long, that's just for routing purposes. This formula is accurate to modern games, but not to gen I. I tried to manually calculate it and was unable to get accurate results.
10) All Pokemon used are assumed to have a DV of 7, and for the purpose of the run, I reset with each Pokemon acquired until I get stats that correspond to this.
With the rules established, here is the party I used.
Final level: 64
Earthquake
Blizzard
Hydro Pump
Surf
Starter. Used in all major battles except for Lt. Surge. Squirtle is selected as the starter over Bulbasaur and Charmander mainly due to access to Bubble Beam after defeating Misty. Having a 65 STAB move at this point of the game is very powerful, especially when Ivysaur and Charmeleon are working with 35/40 BP STABs. From there on, Wartortle steadily becomes stronger with access to better TMs (Dig, Body Slam, Ice Beam) and later on Surf and Blizzard as Blastoise. In general it's a very balanced Pokemon with steady growth from start to finish.
Final level: 57
Agility
Drill Peck
Thunderbolt
Thunder
Along with Blastoise, Zapdos is the only other Pokemon used until the end of the run. Zapdos is used for Rival 6/7/final, Sabrina, Lorelei, and Lance. Zapdos is caught before Rival 6 (Silph Co.) and the Master Ball because it would take fewer turns to catch it and use it for that battle than grinding Blastoise until it can solo Rival 6 would. Zapdos' Thunderbolt nails Water types while Drill Peck covers the Rival's Venusaur and anything with high Special or a resistance to Electric. This gives Zapdos excellent coverage for Blastoise's shortcomings.
Final level: 7
Used for Rival 2 (Route 22). Catching a Spearow for that battle to use against the Rival's Bulbasaur takes fewer turns than grinding Squirtle until it can solo the fight consistently. After that, its only use is being traded for Farfetch'd for HM utility.
Final level: 32
Used for Lt. Surge, Rival 5 (Pokemon Tower), and Koga. Caught to save turns against Lt. Surge, whose Raichu requires quite a bit of grinding for Wartortle to otherwise beat. A 100 BP STAB in the form of Dig along with having a high catch level of 31 is excellent, but it doesn't offer much else past that, and Wartortle needs to hog all the EXP it can get for the end of the game, so Dugtrio doesn't see much use. Only other thing to note is that catching a level 31 Dugtrio is an absolute pain due to being a 1% encounter rate, the level 29 ones are a 4% encounter rate and much more easily encountered. Having to use the Repel trick isn't too bad, the longest drought I had resetting for DVs was 15 minutes of no encounters.
The above 4 are used for battle. Farfetch'd (from the Spearow trade) and Lapras are used for HM utility.
And lastly, notes on major battles.
Rival 1
Party: Squirtle (5)
Tail Whip -> Tackle offers the best turn/damage ratio. There isn't really much that can be done about Growl, so I kinda disregard it here and hope for the best. The projected turn count is 5 turns, in the run it was 8 due to a Growl and two misses. Ugh.
Rival 2
Party: Squirtle (12), Spearow (6)
Squirtle leads. Tail Whip -> Tackle offers a better damage/turn ratio than Bubble here. There isn't anything that can really be done about Growl/Sand Attack. Pidgey has STAB on its Gust so Squirtle needs a bit of a level advantage to compensate for its inferior move power. After KOing Pidgey, a Tail Whip on Bulbasaur will make Spearow's Peck a 2HKO, while Bulbasaur's Tackle is a 3HKO assuming maximum damage ranges. The projected turn count was 8 turns, in the run it was 9 due to Peck doing less damage than it should have. Either Spearow's Attack DV was lower than I thought it was, or a critical from Peck does less damage than when hitting -1 Defense. During the recording process, I actually lost this fight once due to a combination of Growl and multiple critical hits from Tackle.
Brock
Party: Squirtle (12)
Geodude and Onix are both 2HKO'd by Bubble. There's nothing to talk about here.
Misty
Party: Wartortle (24)
Wartortle needs to be level 24 for this fight in order to learn Bite. Bite is desired here because of Mega Punch's lack of accuracy. If Mega Punch was assumed to always connect, then Wartortle can win this fight consistently at about level 20-21. Otherwise, Wartortle just doesn't quite have the damage output to make it past Staryu and Starmie's Tackles consistently. Misty has smart AI, which is actually a handicap here as Wartortle would take more damage from Starmie's Bubble Beam than from Tackle. The projected turn count was 8, in the actual run it was 7 due to nailing a range to 2HKO Staryu.
Rival 3
Party: Wartortle (24)
Armed with Bubble Beam and Bite, this fight becomes what looks like a straightforward sweep that can quickly become a nightmare if Pidgeotto lands a Sand Attack. My first attempt at recording this section of the game ended after a Sand Attack caused 9/11 subsequent attacks to miss, ending in Wartortle getting pulverized by repeated Hyper Fangs from Rattata. Without Bubble Beam, Wartortle doesn't quite have the power to plow through Pidgeotto and Rattata, and will eventually get worn down by Bulbasaur. This makes fighting Misty first a priority. The projected turn count was 6, the actual run matches this.
Rival 4
Party: Wartortle (29)
Pidgeotto is still a 2HKO so Sand Attack poses some degree of threat, but the power difference is a lot higher than the previous fight so the odds are better. The risk of Sand Attack aside, it's a straight sweep. The projected turn count was 6, a critical hit on Pidgeotto and Ivysaur made it 4 turns in the actual run.
Lt. Surge
Party: Dugtrio (31)
Wartortle needs a couple more levels to outspeed Raichu and avoid getting KO'd by Thunderbolt, so Dugtrio is caught instead. Dugtrio effortlessly sweeps the fight with Dig. The projected turn count was 6, and the run matched this.
Giovanni 1
Party: Wartortle (33)
The only fight against Giovanni that isn't a perfect straight sweep. Kangaskhan is faster than Wartortle and has both Comet Punch and Bite, which with some RNG shenanigans can be pretty dangerous, between Bite's flinch chance and a critical Comet Punch doing a lot of damage if it gets 4-5 hits. It's still a pretty safe fight though. The projected turn count was 5, and in the actual run it was 4 due to hitting a range to 2HKO Kangaskhan.
Rival 5
Party: Wartortle (34), Dugtrio (31)
Even with the level advantage, Wartortle 3HKOs Gyarados with Ice Beam, which has access to Dragon Rage. If Gyarados uses Dragon Rage twice, then Ivysaur can KO with Vine Whip, as Ivysaur is 2HKO'd by Ice Beam. Because of that, it's necessary to switch to Dugtrio to finish Ivysaur off. Dugtrio is only threatened by the possibility of critical Vine Whips. In the actual run, Gyarados only used one Dragon Rage, making it unnecessary to switch to Dugtrio, but I do so anyway due to not having a backup for routing. I need Dugtrio to be level 32 for Koga, and this fight gets it there. The projected turn count was 10, and the actual run matched this.
Koga
Party: Dugtrio (32), Blastoise (42)
This is the first battle where it can be very dangerous, but only if the AI makes very selective choices. First of all, Dugtrio cannot safely sweep the fight if it repeatedly gets hit. Everything is at least a 2HKO from Dig at this level, and even a Sludge will do about 25-33% of Dugtrio's HP despite being resisted. Even if Dugtrio could sweep, it's fewer turns if I use Blastoise due to Blastoise having Earthquake. But Blastoise also cannot sweep if it takes repeated hits, as Weezing's Selfdestruct will do massive damage and KO if Blastoise has taken 1-2 hits from anything's Sludge. Barring critical hits, the strategy I developed that was sure to work was to have Blastoise use Reflect on the second Koffing. This ensures that even the worst case scenario of 2 Sludges from Koffing (1 on switch that Poisons, 1 on Reflect) + 1 Sludge and 1 Selfdestruct from Weezing will not KO Blastoise. This circumstance has a less than 1% chance of occurring, but there are other potential problems like critical hits from Sludge or especially Selfdestruct and accuracy issues caused by Smokescreen, and basically the probability of losing the fight is quite a bit higher than 1%. All in all, Koga is the most dangerous gym leader in the run, even if it is a very safe fight overall. At any rate, projected turn count was 15, actual count was 14 due to landing a 2HKO range on Weezing.
I forgot to mention this, but the reason I wanted Dugtrio at level 32 for this fight was to guarantee a 2HKO on Muk. At level 31 it's a range to 2HKO.
Erika
Party: Blastoise (42)
Victreebel has Sleep Powder and is not guaranteed to be OKHO'd by Blastoise unless I have Blizzard or the Volcano Badge to increase Special by 10%, which lets Ice Beam net a guaranteed KO. For that reason, Erika is fought after Koga. Erika is fought with Blastoise instead of Zapdos for EXP routing. The only reason Erika is fought before Blaine is because it's a shorter fight. Blizzard is used twice with a 19% chance of missing once out of two uses, whereas a straight Ice Beam sweep would have been a 1.2% chance of missing any hits due to the 99.6% accuracy thing gen I has. At any rate, both projected and actual turn count is 3 here.
Blaine
Party: Blastoise (44)
Blastoise simply sweeps with Surf here. In earlier versions of the route, Blastoise was at about level 38-39 and actually needed to use Reflect in order to be guaranteed to withstand the Normal attacks from Rapidash and Arcanine, which were both 2HKOs for Surf and faster than Blastoise. Arcanine is still a guaranteed 2HKO, but in the actual run a critical hit makes it a OHKO. The projected turn count was 5, the actual run is 4 turns.
Rival 6
Party: Blastoise (45), Zapdos (50)
Blastoise is used for EXP, beating Pidgeot before switching out for Zapdos against Gyarados. From there on Zapdos just sweeps the fight between level and type advantage. In an earlier version of the route, I instead used Electrode for fighting Pidgeot and Gyarados and switching Blastoise in. Blastoise was able to withstand Alakazam's Psychic and Venusaur's Razor Leaf, but only barely. However, Electrode was insufficient for the next Rival battle, so I had to redo the route using Zapdos instead. Zapdos is pretty much needed here because Gyarados is still a 3HKO for Blastoise, Alakazam is faster than Blastoise, and Venusaur can withstand a Blizzard, it's all too much for Blastoise to handle without being considerably more overleveled. The projected turn count and actual run is 6 turns.
Giovanni 2
Party: Blastoise (45)
Blastoise OHKOs everything, there is nothing to comment on. Projected turn count and actual run is 4 turns.
Sabrina
Party: Blastoise (46), Zapdos (51)
Blastoise is used for EXP again, KOing the Kadabra before switching out. Blastoise cannot OHKO the rest of the fight, so Zapdos is needed instead. Mr. Mime has both Barrier and Light Screen, but Zapdos can OHKO with either Drill Peck or Thunderbolt, so it doesn't matter which it uses. From there, it OHKOs the rest of Sabrina's party. Projected turn count and actual run is 5 turns.
Giovanni 3
Party: Blastoise (49)
Blastoise OHKOs everything again. Projected turn count and actual run is 5 turns.
Rival 7
Party: Blastoise (50), Zapdos (51)
This is where the first routing mistake occurs. During the last phase of rerouting, I took as much EXP as I could spare from Zapdos for Blastoise and forgot that I needed Zapdos to be level 53 for this fight. The reason for that is to outspeed the Alakazam here. Alakazam can otherwise use Reflect on the first turn it's out, which makes Zapdos' attacks 3HKOs. It's very unlikely to occur, but from there it can Recover stall Zapdos, should the AI actually pull it off. The odds of seeing a turn 1 Reflect at all, and Alakazam not getting KO'd by critical Drill Peck, are roughly 21%. This being an actual problem afterwards is even lower of a probability than that, but it is a mistake that I made in routing.
Anyway, Blastoise KOs Pidgeot with Blizzard and Rhyhorn with Surf, then switches out for Zapdos against Gyarados. Zapdos OHKOs Gyarados and 2HKOs Alakazam and Venusaur with Drill Peck. Now, the other routing mistake I made was that I believed that "smart AI" avoided using NVE moves if it had alternatives in the same way that it prefers SE moves even if they don't cause damage, as that's what I understood an explanation on the AI to say. But this isn't actually the case. Key here being that I assumed Venusaur wouldn't use Grass attacks against Zapdos. Zapdos can withstand one Hydro Pump from Gyarados and one Psychic from Alakazam, but I didn't check to see if it had enough HP to survive a Razor Leaf from Venusaur afterward. I don't think it would. But at any rate the probability of this being an issue is very low, it's only roughly a 25% chance each that Gyarados/Alakazam use those moves to begin with, and Venusaur should preferentially use Growth/Poison Powder.
Projected turn count was 9, actual run was also 9 turns.
Lorelei
Party: Zapdos (55), Blastoise (57)
Zapdos needs to be at level 55 in order to net a guaranteed KO on Dewgong using Thunderbolt. This is necessary in order to preserve Zapdos' HP, it needs to later take a hit from Lapras' Blizzard. Zapdos cannot net guaranteed KOs on Jynx or Lapras, so when Jynx comes in, I switch to Blastoise. Jynx only has 3 Normal attacks + Ice Punch. Blastoise 2HKOs with Earthquake and weakens Lapras to put it into Thunderbolt KO range for Zapdos. Zapdos then switches in, tanks a Blizzard, and finishes the fight. Lapras does know Confuse Ray, but Lorelei has smart AI and should always use Blizzard instead of Lapras' other moves. Lapras has an 11.7% chance of getting a critical hit, paired with Blizzard's 90% accuracy this is a 10.5% chance of Zapdos being KO'd here, plus the 10% chance of a Freeze. Aside from the last two Rival battles, this is the fight with the highest possibility of losing the run at. Projected turn count was 9, actual turn count was also 9.
Bruno
Party: Blastoise (57)
Straight sweep here, there's nothing to talk about. Machamp is 2HKO'd by Surf, but a critical turns what was projected to be a 6 turn fight into a 5 turn fight, everything is OHKO'd.
Agatha
Party: Blastoise (63)
Agatha's level 60 Gengar has Confuse Ray, in order to avoid dealing with that Blastoise needs to be able to outspeed it. Accounting for stat EXP, Blastoise needs to be level 63 in order to outspeed this Gengar. I use Rare Candies and 3 Carbos to hit this. I could do it at a lower level with more Carbos but I didn't like the idea of spamming stat items to compensate for lower levels. Anyway, the bulk of EXP routing in the second half of the game is to get Blastoise at a level range that it can hit 63 for this fight using Rare Candies. With Blastoise at 63, everything is OHKO'd. Projected and actual turn count is 5.
Lance
Party: Zapdos (55), Blastoise (63)
Zapdos KOs the Gyarados and sets up Agility on the first Dragonair. The Aerodactyl outspeeds Zapdos and has Supersonic, it's roughly a 13.7% chance it'll use it and it'll connect. The Dragonair are 2HKOs with Drill Peck and don't really merit any comments, Aerodactyl is OHKO'd by Thunderbolt. Blastoise is switched in against Dragonite and KOs with Blizzard. Projected and actual turn count is 9.
Rival Final
Party: Zapdos (56), Blastoise (63)
Like with the previous Rival battle, there was an error in the routing due to misunderstanding the AI's choices with NVE moves. The intended battle strategy I had was to KO Pidgeot and Alakazam with Zapdos, switch Blastoise in for Rhydon, switch Zapdos again for Gyarados, switch Blastoise again for Arcanine, then switch Zapdos in for Venusaur. By this point, I estimated that Zapdos would be at about 10% HP from Alakazam's Psychic and Gyarados' Hyper Beam, but wouldn't be KO'd by Venusaur because its only non-Grass move was Growth. But Venusaur will use its Grass moves here so this strategy wasn't foolproof.
The alternate strategy that I actually did go through with was having Blastoise take out Gyarados instead of switching to Zapdos. The problem with this strategy is that Rhydon has roughly a 50% chance of using Leer/Tail Whip, which would put Blastoise in range for a KO provided Gyarados uses Dragon Rage followed by Hyper Beam and Arcanine isn't 1HKO'd and uses Take Down. If Gyarados uses Hyper Beam as its first attack, then it's safe, I switch to Zapdos and KO, and Arcanine is also 2HKO'd by Zapdos' Thunderbolt. The probability of Gyarados using that specific order of attacks and Blastoise missing the KO on Arcanine is very low, but it's something I would have altered the route on had I known about it. The alteration would be as simple as getting Blastoise to level 66 by this point, although 65 + some Calcium would also do this. At that point, Surf is a guaranteed KO. With the current route, Hydro Pump has a better shot of OHKOing Arcanine than Surf, but I actually forgot to use Hydro Pump.
Anyway, Zapdos sets up Agility on Pidgeot in order to outspeed Alakazam, then OHKOs Pidgeot with Thunderbolt and 2HKOs Alakazam with Drill Peck. Blastoise is switched in to OHKO Rhydon. Blizzard is a 3HKO on Gyarados, if it uses Dragon Rage it's a 4HKO, if it uses Hyper Beam I just switch to Zapdos. Arcanine is a range to KO but its most dangerous move is just Take Down, which would do about 25% of Blastoise's HP at max. After Arcanine, Zapdos comes in and 2HKOs Venusuar, which poses no threat to it. The projected turn count was 14, the actual run was 12 turns due to a critical on Alakazam and netting the range to 1HKO on Arcanine.
Party: Squirtle (5)
Tail Whip -> Tackle offers the best turn/damage ratio. There isn't really much that can be done about Growl, so I kinda disregard it here and hope for the best. The projected turn count is 5 turns, in the run it was 8 due to a Growl and two misses. Ugh.
Rival 2
Party: Squirtle (12), Spearow (6)
Squirtle leads. Tail Whip -> Tackle offers a better damage/turn ratio than Bubble here. There isn't anything that can really be done about Growl/Sand Attack. Pidgey has STAB on its Gust so Squirtle needs a bit of a level advantage to compensate for its inferior move power. After KOing Pidgey, a Tail Whip on Bulbasaur will make Spearow's Peck a 2HKO, while Bulbasaur's Tackle is a 3HKO assuming maximum damage ranges. The projected turn count was 8 turns, in the run it was 9 due to Peck doing less damage than it should have. Either Spearow's Attack DV was lower than I thought it was, or a critical from Peck does less damage than when hitting -1 Defense. During the recording process, I actually lost this fight once due to a combination of Growl and multiple critical hits from Tackle.
Brock
Party: Squirtle (12)
Geodude and Onix are both 2HKO'd by Bubble. There's nothing to talk about here.
Misty
Party: Wartortle (24)
Wartortle needs to be level 24 for this fight in order to learn Bite. Bite is desired here because of Mega Punch's lack of accuracy. If Mega Punch was assumed to always connect, then Wartortle can win this fight consistently at about level 20-21. Otherwise, Wartortle just doesn't quite have the damage output to make it past Staryu and Starmie's Tackles consistently. Misty has smart AI, which is actually a handicap here as Wartortle would take more damage from Starmie's Bubble Beam than from Tackle. The projected turn count was 8, in the actual run it was 7 due to nailing a range to 2HKO Staryu.
Rival 3
Party: Wartortle (24)
Armed with Bubble Beam and Bite, this fight becomes what looks like a straightforward sweep that can quickly become a nightmare if Pidgeotto lands a Sand Attack. My first attempt at recording this section of the game ended after a Sand Attack caused 9/11 subsequent attacks to miss, ending in Wartortle getting pulverized by repeated Hyper Fangs from Rattata. Without Bubble Beam, Wartortle doesn't quite have the power to plow through Pidgeotto and Rattata, and will eventually get worn down by Bulbasaur. This makes fighting Misty first a priority. The projected turn count was 6, the actual run matches this.
Rival 4
Party: Wartortle (29)
Pidgeotto is still a 2HKO so Sand Attack poses some degree of threat, but the power difference is a lot higher than the previous fight so the odds are better. The risk of Sand Attack aside, it's a straight sweep. The projected turn count was 6, a critical hit on Pidgeotto and Ivysaur made it 4 turns in the actual run.
Lt. Surge
Party: Dugtrio (31)
Wartortle needs a couple more levels to outspeed Raichu and avoid getting KO'd by Thunderbolt, so Dugtrio is caught instead. Dugtrio effortlessly sweeps the fight with Dig. The projected turn count was 6, and the run matched this.
Giovanni 1
Party: Wartortle (33)
The only fight against Giovanni that isn't a perfect straight sweep. Kangaskhan is faster than Wartortle and has both Comet Punch and Bite, which with some RNG shenanigans can be pretty dangerous, between Bite's flinch chance and a critical Comet Punch doing a lot of damage if it gets 4-5 hits. It's still a pretty safe fight though. The projected turn count was 5, and in the actual run it was 4 due to hitting a range to 2HKO Kangaskhan.
Rival 5
Party: Wartortle (34), Dugtrio (31)
Even with the level advantage, Wartortle 3HKOs Gyarados with Ice Beam, which has access to Dragon Rage. If Gyarados uses Dragon Rage twice, then Ivysaur can KO with Vine Whip, as Ivysaur is 2HKO'd by Ice Beam. Because of that, it's necessary to switch to Dugtrio to finish Ivysaur off. Dugtrio is only threatened by the possibility of critical Vine Whips. In the actual run, Gyarados only used one Dragon Rage, making it unnecessary to switch to Dugtrio, but I do so anyway due to not having a backup for routing. I need Dugtrio to be level 32 for Koga, and this fight gets it there. The projected turn count was 10, and the actual run matched this.
Koga
Party: Dugtrio (32), Blastoise (42)
This is the first battle where it can be very dangerous, but only if the AI makes very selective choices. First of all, Dugtrio cannot safely sweep the fight if it repeatedly gets hit. Everything is at least a 2HKO from Dig at this level, and even a Sludge will do about 25-33% of Dugtrio's HP despite being resisted. Even if Dugtrio could sweep, it's fewer turns if I use Blastoise due to Blastoise having Earthquake. But Blastoise also cannot sweep if it takes repeated hits, as Weezing's Selfdestruct will do massive damage and KO if Blastoise has taken 1-2 hits from anything's Sludge. Barring critical hits, the strategy I developed that was sure to work was to have Blastoise use Reflect on the second Koffing. This ensures that even the worst case scenario of 2 Sludges from Koffing (1 on switch that Poisons, 1 on Reflect) + 1 Sludge and 1 Selfdestruct from Weezing will not KO Blastoise. This circumstance has a less than 1% chance of occurring, but there are other potential problems like critical hits from Sludge or especially Selfdestruct and accuracy issues caused by Smokescreen, and basically the probability of losing the fight is quite a bit higher than 1%. All in all, Koga is the most dangerous gym leader in the run, even if it is a very safe fight overall. At any rate, projected turn count was 15, actual count was 14 due to landing a 2HKO range on Weezing.
I forgot to mention this, but the reason I wanted Dugtrio at level 32 for this fight was to guarantee a 2HKO on Muk. At level 31 it's a range to 2HKO.
Erika
Party: Blastoise (42)
Victreebel has Sleep Powder and is not guaranteed to be OKHO'd by Blastoise unless I have Blizzard or the Volcano Badge to increase Special by 10%, which lets Ice Beam net a guaranteed KO. For that reason, Erika is fought after Koga. Erika is fought with Blastoise instead of Zapdos for EXP routing. The only reason Erika is fought before Blaine is because it's a shorter fight. Blizzard is used twice with a 19% chance of missing once out of two uses, whereas a straight Ice Beam sweep would have been a 1.2% chance of missing any hits due to the 99.6% accuracy thing gen I has. At any rate, both projected and actual turn count is 3 here.
Blaine
Party: Blastoise (44)
Blastoise simply sweeps with Surf here. In earlier versions of the route, Blastoise was at about level 38-39 and actually needed to use Reflect in order to be guaranteed to withstand the Normal attacks from Rapidash and Arcanine, which were both 2HKOs for Surf and faster than Blastoise. Arcanine is still a guaranteed 2HKO, but in the actual run a critical hit makes it a OHKO. The projected turn count was 5, the actual run is 4 turns.
Rival 6
Party: Blastoise (45), Zapdos (50)
Blastoise is used for EXP, beating Pidgeot before switching out for Zapdos against Gyarados. From there on Zapdos just sweeps the fight between level and type advantage. In an earlier version of the route, I instead used Electrode for fighting Pidgeot and Gyarados and switching Blastoise in. Blastoise was able to withstand Alakazam's Psychic and Venusaur's Razor Leaf, but only barely. However, Electrode was insufficient for the next Rival battle, so I had to redo the route using Zapdos instead. Zapdos is pretty much needed here because Gyarados is still a 3HKO for Blastoise, Alakazam is faster than Blastoise, and Venusaur can withstand a Blizzard, it's all too much for Blastoise to handle without being considerably more overleveled. The projected turn count and actual run is 6 turns.
Giovanni 2
Party: Blastoise (45)
Blastoise OHKOs everything, there is nothing to comment on. Projected turn count and actual run is 4 turns.
Sabrina
Party: Blastoise (46), Zapdos (51)
Blastoise is used for EXP again, KOing the Kadabra before switching out. Blastoise cannot OHKO the rest of the fight, so Zapdos is needed instead. Mr. Mime has both Barrier and Light Screen, but Zapdos can OHKO with either Drill Peck or Thunderbolt, so it doesn't matter which it uses. From there, it OHKOs the rest of Sabrina's party. Projected turn count and actual run is 5 turns.
Giovanni 3
Party: Blastoise (49)
Blastoise OHKOs everything again. Projected turn count and actual run is 5 turns.
Rival 7
Party: Blastoise (50), Zapdos (51)
This is where the first routing mistake occurs. During the last phase of rerouting, I took as much EXP as I could spare from Zapdos for Blastoise and forgot that I needed Zapdos to be level 53 for this fight. The reason for that is to outspeed the Alakazam here. Alakazam can otherwise use Reflect on the first turn it's out, which makes Zapdos' attacks 3HKOs. It's very unlikely to occur, but from there it can Recover stall Zapdos, should the AI actually pull it off. The odds of seeing a turn 1 Reflect at all, and Alakazam not getting KO'd by critical Drill Peck, are roughly 21%. This being an actual problem afterwards is even lower of a probability than that, but it is a mistake that I made in routing.
Anyway, Blastoise KOs Pidgeot with Blizzard and Rhyhorn with Surf, then switches out for Zapdos against Gyarados. Zapdos OHKOs Gyarados and 2HKOs Alakazam and Venusaur with Drill Peck. Now, the other routing mistake I made was that I believed that "smart AI" avoided using NVE moves if it had alternatives in the same way that it prefers SE moves even if they don't cause damage, as that's what I understood an explanation on the AI to say. But this isn't actually the case. Key here being that I assumed Venusaur wouldn't use Grass attacks against Zapdos. Zapdos can withstand one Hydro Pump from Gyarados and one Psychic from Alakazam, but I didn't check to see if it had enough HP to survive a Razor Leaf from Venusaur afterward. I don't think it would. But at any rate the probability of this being an issue is very low, it's only roughly a 25% chance each that Gyarados/Alakazam use those moves to begin with, and Venusaur should preferentially use Growth/Poison Powder.
Projected turn count was 9, actual run was also 9 turns.
Lorelei
Party: Zapdos (55), Blastoise (57)
Zapdos needs to be at level 55 in order to net a guaranteed KO on Dewgong using Thunderbolt. This is necessary in order to preserve Zapdos' HP, it needs to later take a hit from Lapras' Blizzard. Zapdos cannot net guaranteed KOs on Jynx or Lapras, so when Jynx comes in, I switch to Blastoise. Jynx only has 3 Normal attacks + Ice Punch. Blastoise 2HKOs with Earthquake and weakens Lapras to put it into Thunderbolt KO range for Zapdos. Zapdos then switches in, tanks a Blizzard, and finishes the fight. Lapras does know Confuse Ray, but Lorelei has smart AI and should always use Blizzard instead of Lapras' other moves. Lapras has an 11.7% chance of getting a critical hit, paired with Blizzard's 90% accuracy this is a 10.5% chance of Zapdos being KO'd here, plus the 10% chance of a Freeze. Aside from the last two Rival battles, this is the fight with the highest possibility of losing the run at. Projected turn count was 9, actual turn count was also 9.
Bruno
Party: Blastoise (57)
Straight sweep here, there's nothing to talk about. Machamp is 2HKO'd by Surf, but a critical turns what was projected to be a 6 turn fight into a 5 turn fight, everything is OHKO'd.
Agatha
Party: Blastoise (63)
Agatha's level 60 Gengar has Confuse Ray, in order to avoid dealing with that Blastoise needs to be able to outspeed it. Accounting for stat EXP, Blastoise needs to be level 63 in order to outspeed this Gengar. I use Rare Candies and 3 Carbos to hit this. I could do it at a lower level with more Carbos but I didn't like the idea of spamming stat items to compensate for lower levels. Anyway, the bulk of EXP routing in the second half of the game is to get Blastoise at a level range that it can hit 63 for this fight using Rare Candies. With Blastoise at 63, everything is OHKO'd. Projected and actual turn count is 5.
Lance
Party: Zapdos (55), Blastoise (63)
Zapdos KOs the Gyarados and sets up Agility on the first Dragonair. The Aerodactyl outspeeds Zapdos and has Supersonic, it's roughly a 13.7% chance it'll use it and it'll connect. The Dragonair are 2HKOs with Drill Peck and don't really merit any comments, Aerodactyl is OHKO'd by Thunderbolt. Blastoise is switched in against Dragonite and KOs with Blizzard. Projected and actual turn count is 9.
Rival Final
Party: Zapdos (56), Blastoise (63)
Like with the previous Rival battle, there was an error in the routing due to misunderstanding the AI's choices with NVE moves. The intended battle strategy I had was to KO Pidgeot and Alakazam with Zapdos, switch Blastoise in for Rhydon, switch Zapdos again for Gyarados, switch Blastoise again for Arcanine, then switch Zapdos in for Venusaur. By this point, I estimated that Zapdos would be at about 10% HP from Alakazam's Psychic and Gyarados' Hyper Beam, but wouldn't be KO'd by Venusaur because its only non-Grass move was Growth. But Venusaur will use its Grass moves here so this strategy wasn't foolproof.
The alternate strategy that I actually did go through with was having Blastoise take out Gyarados instead of switching to Zapdos. The problem with this strategy is that Rhydon has roughly a 50% chance of using Leer/Tail Whip, which would put Blastoise in range for a KO provided Gyarados uses Dragon Rage followed by Hyper Beam and Arcanine isn't 1HKO'd and uses Take Down. If Gyarados uses Hyper Beam as its first attack, then it's safe, I switch to Zapdos and KO, and Arcanine is also 2HKO'd by Zapdos' Thunderbolt. The probability of Gyarados using that specific order of attacks and Blastoise missing the KO on Arcanine is very low, but it's something I would have altered the route on had I known about it. The alteration would be as simple as getting Blastoise to level 66 by this point, although 65 + some Calcium would also do this. At that point, Surf is a guaranteed KO. With the current route, Hydro Pump has a better shot of OHKOing Arcanine than Surf, but I actually forgot to use Hydro Pump.
Anyway, Zapdos sets up Agility on Pidgeot in order to outspeed Alakazam, then OHKOs Pidgeot with Thunderbolt and 2HKOs Alakazam with Drill Peck. Blastoise is switched in to OHKO Rhydon. Blizzard is a 3HKO on Gyarados, if it uses Dragon Rage it's a 4HKO, if it uses Hyper Beam I just switch to Zapdos. Arcanine is a range to KO but its most dangerous move is just Take Down, which would do about 25% of Blastoise's HP at max. After Arcanine, Zapdos comes in and 2HKOs Venusuar, which poses no threat to it. The projected turn count was 14, the actual run was 12 turns due to a critical on Alakazam and netting the range to 1HKO on Arcanine.
I think that's about all there is to say about it. IGT at completion was 3 hours 56 minutes, the total turn count was 483. The goal was something of a showcase of the game being played "efficiently" while relying minimally on luck and in-battle items, and in that regard I feel like I was able to accomplish that satisfactorily. The next project is Pokemon Crystal, but it'll take a long time to finish that due to juggling other things. I hope that some people here might actually find this interesting.
Last edited: