Elo World REDUX: Scientifically Ranking the Pokémon Red Trainers AGAIN - Level 50 AI Tournament
ฝัง
- เผยแพร่เมื่อ 4 มิ.ย. 2024
- In this video, I fix some of the issues with my original "Ranking the Pokémon Red Trainers" video, and run a brand new tournament with those changes. Most notably, I flatten all the Pokémon to level 50, the way competitive Pokémon tournaments are run.
Some links that might be of interest:
---
The original video: • Pokémon Red/Blue AI To...
The instructions on how to run battles from Hash IDs: (this was broken before, should be working now, grab the latest) github.com/jsettlem/elo_world...
Nic Dobson's blog post: nicidob.github.io/nba_elo/
It would be rude to steal @tom7's title again without linking his excellent video: • 30 Weird Chess Algorit...
My Twitter: / pimanrules
Music credits:
---
Pokémon remixes by @DJMykah. I commissioned the "Hall of Fame" remix from them, and as you can hear, it turned out great. Check out their channel here: / djmykah
"Heartbeat" from "The Music of 3Blue1Brown" by Vincent Rubinetti. Licensed with permission. Check out the album here: vincerubinetti.bandcamp.com/a...
Convenient timestamps:
---
0:00 Intro
2:11 The plan
2:52 Bug squashing
8:02 WONTFIX
11:03 Tournament enhancements
16:05 Elo system enhancements
18:33 How Elo works
24:11 Changes to orchestrating battles
26:13 Ranking of trainers
38:32 "Before we get to the top 16..."
39:30 Sweet 16
48:33 Grand finals
50:01 Trainers in game order
50:39 Frequently asked questions
52:12 Outro
Corrections:
15:27 - This process of fixing the moves went wrong for some trainers, specifically when they're in the enemy position, resulting in them actually having the moves they would have at level 50. This is what some people wanted anyway so I'm gonna try not to beat myself up over it (but uuuuugggghhhhh)
16:50 - That is so obviously not the sound ə makes. That's like the one IPA symbol everyone learns in elementary school. My excuse was that my brain was on total autopilot--I just copied the phonetic transcription of "hello," not realizing that "hello" is really pronounced more like "hullo" in most cases
33:17 - I say Wrap does "basically no damage." Of course, it's actually doing no damage because Gastly is Ghost type, but the effect still triggers because Gen I.
50:51 - Obviously I meant EVs instead of IVs - เกม
Since there's no way I'm making a follow-up to the follow-up, I'll use this pinned comment for mistakes and/or addendums:
15:27 - The biggest mistake in this video... this process of fixing the moves went wrong for some trainers, specifically when they're in the enemy position, resulting in them actually having the moves they would have at level 50. This is what some people wanted anyway so I'm gonna try not to beat myself up over it (but uuuuugggghhhhh)
16:50 - That is so obviously not the sound ə makes. That's like the one IPA symbol everyone learns in elementary school. My excuse was that my brain was on total autopilot--I just copied the phonetic transcription of "hello," not realizing that "hello" is really pronounced more like "hullo" in most cases
33:17 - I say Wrap does "basically no damage." Of course, it's actually doing no damage because Gastly is Ghost type, but the effect still triggers because Gen I.
50:51 - Obviously I meant EVs instead of IVs
And by EVs I assume you mean stat exp, bc gen 1 :p
(This comment is a joke, I love both videos)
Clearly, Wrap is dealing emotional damage, hence why the opponent can't act during it~
@@lykillcorreli6740 after playing blue kaizo, wrap, fire spin and clamp have left me emotionally scarred.
Repin this m8
What about a Linux-native emulator such as SameBoy?
“Hiker Number 3 partook in the 4th longest battle, his opponent was himself” that’s deep
If I'm ever counting Hikers when I'm playing in the future, I'll remember to tell him to have a good day.
The Charizard Champion's Greatest win was against himself
@@clarktheavali3959 he’s achieved enlightenment
(Rocking back and forth in the corner, staring out into the void) yup, understandable
He took a hike of self-reflection and discovery
I’ve seen the code for this and totally understand if none of the other generations never get done.
There are disassemblies for almost all pokemon games. Doing this would be a matter of emulating the games partially (only the AI subroutines), which would also be orders of magnitude faster.
i mean, i totally understand the reason why it could be REALLY time consuming, but tbf i'm still rooting for the GSC tournament. I just want more of this xD
Maybe aliens could do it for us. It could be their gift to humanity.
@@Sealed_Chamber i'd take that over aids cure tbh
Well GEN2 can run on the same emulator, so it should only require modifications to the same code. As opposed to starting from scratch again. Still, I'm not surprised these changes took a year.
Dang, Oak was slapping Blue in every round in the top 16.
"You gonna learn today son."
Until they hit Grand Finals, then Blue had his anime clarity moment and took it all.
"Smell ya later pops."
The mark of a true champion.
Blue avenged his younger self and defeated Pops once and for all
I imagined this in my head before even reading this comment
This comment made me laugh, thank you
And still won't remember his name...
and the best part is oak's Win in the first tournament was rigged because the rival won the most of all trainers and the venusaur champion being the best of all of them in the first tournament losing only 2 times and oak was still declared number one when the rival with the venusaur deserved the most and this hits even harder because the venusaur champion is the one who defeated oak
You know it's a good video when the video about something scientifically accurate actually spends most of the time talking about methodology rather than results.
I love this comment
I love scientific video-papers. Normally it’s so boring to read scientific studies but when it’s read for me with fun visuals it becomes a more pleasant experiences
It's nice that it appeals to both audiences in that way
This, very much this. It's important!
@@frownyclowny6955we need more papers in the form of videos
Conclusion:
- The second strongest gen 1 Pokémon is Venusaur because it won the most
- The strongest gen 1 Pokémon is Bulbasaur, because it beat Venusaur
LMAO BULBASAUR SUPREMACY
Nah Charizard is #1
@@bigballoftape5343 Please look at the number 1 elo score team and its champion blue charizard team.
@@Saxton_Hoovy Rock Throw go brr
@@Saxton_Hoovy that is because
A) It is the only one without that awful Arcanine.
B) Rhydon doesn't have a rock attack.
C) Venusaur moveset sucks with 3 grass attacks and growth (not leech seed, toxic or sleep powder) , unlike blastoise that has blizzard and charizard that has fire spin and slash.
"Smell ya later, pops," Blue says as he takes out his grandfather's last Pokémon with tears in his eyes.
F
Well considering how OP his Alakazam was, I doubt he had tears
Yeah "gramps"
Because he knew, he knew that if Oak lost a pokemon battle, that his heart would stop. It was the reason Oak stopped battle all those years ago and became a researcher.
Got to give a shoutout to Rocket #22 who managed to perfectly slot in at exactly #215 again. A true legend in sheer mediocrity.
It's so insane to me that SABRINA, mid-tier gym leader, is in the top 16 alongside a bunch of champs, elite 4s, and a victory road trainer. Psychic was so OP
Well Blaine and Giovanni were just really bad in red/blue
It’s definitely reflective of the experience of playing against her and yes realising that she had no weakness since bug was effectively non existent as a counter
And the Juggler who also had a bunch of psychic pokemon
It's a combination of the Psychic-types being generally stronger in Gen 1 (see how the Psychic-type Jugglers do in the round Robin?), her AI being a bit smarter than the rest AND having options for a stall. No wonder she's often touted the real difficulty spike in the original game.
@Elbow Drop Gaming No. Your team was Mewtwo, and a couple backup stab psychic users Incase anything went way too wrong.
36:50 She lost against the first rival battle, but won against Lance....oh gen 1 AI, oh you...
She basically just needed to beat Gyarados and Aerodactyl, since the Dragonair and Dragonite just spammed Agility and Barrier because _Psychic is good against Poison, right?_
she is the ultimate i can defeat anyone and lose to anyone
Would have been funnier if it was champion with venusuar she won against
@@paulnewton2284 The All Terrain Venomoth strikes again.
Hilariously, she won through the exact same cheese tactics that she lost by. She is truly the most chaotic trainer in this whole list.
I cried when Juggler lost he literally was seeded against the second worst matchup for him
Some might say he...
...
...
Dropped his balls.
Truly the people's champ
I liked when he disabled recover as a true stragtegist
@@loulou3676 YES!!! YESS!!!! YESSS!!!
Seed rigging
By jumping into the tournament bracket, we missed out on hearing how their battles in the main tournament went. I'm especially curious about that juggler who made it in despite wasting half of their turns.
When he said what "seed" they were, that's their placement. So the juggler was #15 in the main tourney, and Charizard Champ was #1.
@@drthvrdr But I want to know why the Juggler did so well in the first place.
@@harley2938 Well he has 4 Psychic types which are all OP in Gen 1
@@RipperJack77 Also there are absolutely tons of poison trainers
@@harley2938 most team rocket grunts have poison types nearly all bug catchers have poison types nearly every single grass type trainer is stuck with with odish line bellsprout line and bulbasaur which are part poison half the swimmers have tentacruel as their ace and most hikers only have 2-3 pokemons 1 of which is a fighting type also there are a lot of channelers with a single ghastly and the fisherman only carry magikarps with a single garidos bird catchers are fast but lack the moveset bikers only have poison types there aren't as many bulky water type users and nearly no electric type users aside from surge and there's about seven trainers that only use fighting types so it makes sense he made it to top 15 with only psychic types
The 3B1B segment had me laughing because of how much fidelity there is to his style. Well done on that joke.
@@LF-cm5wm Those two things aren’t mutually exclusive.
It also had me highly amused.
That segment started and my parody senses started tingling, despite not recognizing the style
"The All Magikarp strategy proves ineffective."
I just love the way you say that so seriously.
It does make me wonder how an All Gyarados team would place.
me encountering sinnoh fisherman andrew with 6 magikarps all with dragon rage:😂
@@casoul993 Or with a Budew with growth, that works as well.
a fun side effect is that this would *sort of* make a comprehensive list of how difficult a trainer is to battle for the player at that point in game, considering the player's level would generally be around the same as the opponent. very cool
The only thing I would say is that the trainer can use better strategy though so some fights would be completely different for players rather than a limited AI. That and the fact that most teams focus on one type while a player would hopefully have variety so quite a bit of these rankings would just be type vs type while a trainer would sweep both types with a super-effective.
Yeah. It obviously couldnt be all encompassing (there being a bunch of rock type and normal type teams would mean that the types good against those are able to get extra elo score due to having more beneficial matchups) but the data is interesting to look at. If this was gen 4 for instance, the rarity of fire types would help grass types out and might give water types less advantage
Some would probably still be easier or harder based on which Pokémon you have available. Brock and Misty are much harder for players who started with Charmander for example as there aren't many other Pokémon you can catch and train up to counter them to make up for Charmander being countered (Mankey and Pikachu are a bit rarer to catch and the Nido's don't learn double kick till a point you'd be a bit overleveled anyway)
No definitely not, because you can have any evolution, stats, and anything else that could complicate the number of possibilities
I love how the rival changes clothes and poses throughout the game. I wish that level of detail carried on into the games after it
They do? Hau and Hop do that.
Lillie and Hop do! Hop has so many unique battle animations depending on which it is, it's actually very charming. His league battles mimic Leon's animations but his non league ones have unique poses and throws
You're definitely enter that category of TH-camrs I define as: "I don't expect them to upload, but when a video is up its always a must watch"
Great Job, you're making the nerd and Pokémon fan in me happy
Kinda proud that the 4 team Juggler in S Tier got so far, what an actual chad
psychic type is just strong in gen 1
The true underdog!
Is like switching pokemon is an actual good idea huh
@@cd7677jugglers tend to switch at random, but I imagine it helps in most cases, especially with a team full of psychic types
Bad tactics, but smart team building
Real talk the like scrolling display of all the trainers when discussing the rankings is just phenomenal. It is so aesthetically pleasing and feels like I could just reach out and navigate it myself
Agreed
Honestly, with the music and the rankings going up it's definitely a transcendent experience
it is so cathartic seeing Vensaur rival as the strongest trainer. he was my final boss for my first ever pokemon game and it took way more tries than i wanna admit to beat him. all i'm gonna say is I'm glad a caught that Venomoth
Dang a venomoth killed him that must be a great memory did u nickname it
My personal memory is that I beat Venusaur Champion with a Fissure.
I was really rooting for Juggler #2, he believed in himself and almost pulled it off
If anyone else was curious about the infamous Self-Destructing Rock Tunnel Hiker, he is Hiker #9, ranked at number 213 with a 1485 Elo score. You can see him at around 31:30.
Nice find!
Presumably he just won most of the matchups with non-channeler, 2 pokemon or less trainers
How the flip did that dude got such a good score with basically s*icide tactics?
@@danielsurvivor1372 bigger party size and the fact that any pokemon not rock or ghost type pretty much eat shit when self destruct happens
@@darkness74185 Didn't that move also half defense when used until gen 5?
Half-baked idea for a (maybe) shorter video to make: Which AI can beat the Elite Four?
Just match up trainers against the Elite 4 in order, if they lose they don't move on. Just do trainers that make sense, like the stronger Gym Leaders, Rivals, maybe a handful of randoms from Victory Road.
What, you don't think Fisher#9 could beat the Elite 4?
I love the idea/concept of using the stuff he already programmed to try different scenarios/challenges.
I think we already have an answer tbh, since the data is all there. It's just a matter of cross referencing every loss of the E4 members and see which ai beat all 4 of them
Cool idea! I hope he sees it
The challenge for the AI is any trainer that abuses lances agility spamming dragons is going to also have to be strong enough to kill Lorelei's resting Dewgong
This. Any way he can find to recycle this for more content will be worth it because God damn he clearly put a shitload of effort into making this video
Rest In Peace Juggler #2 you were so close to being the Champion.
Honestly, the fact that each trainer's team composition's difficulty level corresponds so well with the order you encounter them is really huge props to the design skill of the original creators of the game. It's nice to see videos that highlight their achievements other than just the financial success of the franchise in a game that is running on such constrained hardware and is so often panned for its abundant glitches and exploits. Making a game that is this good and that is still so beloved on Gameboy hardware before Pokemon was ever a "thing" with no existing fanbase to reassure its success despite its technical or design failures is a wonder that goes unappreciated.
Could we get a google doc file of all the trainers by rank... I liked when you cycled through them all in the video with all there stats but I sort of ... just want to look at them at my own pace.
I agree, or getting to see the excel spreadsheet mentioned would also be good
@pimanrules Pretty please? 🥺
So last year’s video is more like the characters from the games fighting eachother, while this video is more tournament official Pokémon fights, where it’s more like set Pokémon teams are battling it out.
I honestly like both videos and the insights you give to the code in Pokémon and changes that happen based on the battle scenario constantly changing in real time, gives me a better appreciation for the Pokémon franchise in its complex coding.
this makes me absolutely horrified of how the code of a modern AAA game looks like. Like, pokemon red is a really REALLY simple thing compared to any game from the last 5 years
I actually really liked this as the events leading up to the top 16 were essentially qualifier matches
I love that you basically made a home and away system for the tournament
38:00 I love how the mention of Giovanni and Lorelai is enough to make the Bellsprout in the bottom-right corner finally faint.
laughed out loud when the 3b1b music kicked in, the presentational touches here are fantastic
So, would this be "3 Blue 1 Red"?
@@KnTenshi2 "3Blue1Brown" is a famous divulgation youtube Channel about Math. It is known for its amazing visualizations of theoretical concept. In the "Hardcore Math" section, the screen changed to black as usual in 3B1B videos, the velocity and tone of the voice imitates 3B1B and the music is the one always used.
@@magnamon88 I know - I was making a pun on the name.
I did too. It was great. Really good emulation of 3b1b visual style as well
so anyway, when you solve for when two trainers exactly tie, you get these beautiful fractals...
"I've got more ideas than motivation" is such a fucking mood widjuebfbg
The motto of any and every artist
I like how even hyper genisues like this guy devotes a significant chunk of their lives to Pokemon just like the rest of us. You'd think with his brainpower and understanding of coding and advanced mathematics he wouldn't be entertained by a game meant for children, but nope he's just another dude smashing that A button.
We are all just another person smashing that A button on this blessed day
And I'm just another dude smashing the like button on the video
The last sentence made me die irl
doing stupid bullshit with smart bullshit is a good way to keep your skills sharp
Is not much a matter of what he did was that hard or he is intelligent.
I could have done this... But i would be bored of it midway and it would join my pile of unfinished stuff.
Persistence is his real superpower. And it is more op than psychic in in gen 1
Gen II would be pretty interesting considering it's basically two games in one and if you level the levels the Kanto Gym leaders should be a lot more powerful than they are at the point of the game were you fight them.
Great video by the way.
I like how perfectly suited Gen 1 is for this kind of analysis. The AI is (relatively) simple, Pokemon are always sent out in a specific order, every trainer has the same DVs, and switching out is based on a flat probability for select trainer types rather than a reaction to the battle-state.
I'd understand if the subsequent gens never got their own videos, because the AI just gets more complex and inscrutable the further along you go into the main series' timeline. There are lots of fairly obvious general rules you pick up just from playing casually (e.g. good AI prioritises status moves when the player's Pokemon is at full HP, but will always go for the KO when in range, selecting randomly between the moves that are capable of KOing you), but there's some fuzziness around the edges and it's very common for the AI to behave unexpectedly.
Even more interesting, however, are all the fun quirks in later gens, like how Gen 4 AI will continue to use Status moves against a Substitute and Water-type moves against a Pokemon with Dry Skin, but not Water Absorb.
Tbh the easiest one to tackle would be DP (not Platinum) because DP's AI is random for every trainer, at least after discounting moves that would have no effect. Obviously that's not a very interesting premise for a ranking video though haha.
How much more complex is Gen 2?
@@skibot9974 I don’t know all the details, but here are a few examples of the added complexity as I understand it:
-move selection is more intentional but fuzzier, like trainers with ‘good’ AI will usually but not always select the most powerful move (if they elect to deal damage rather than inflict status/set up/etc)
-starting in gen 2 trainers have a chance of switching out their active Pokémon if it just took damage from an attack and they have a Pokémon in reserve that both resists that attack and has a move that hits your Pokémon super-effectively
-trainers don’t replace fainted Pokémon in a set order, instead having some kind of process for determining what to send in next (which doesn’t always seem to make logical sense)
-there are also just a whole lot of new move mechanics (e.g. weather, Spikes, Protect/Detect/Endure, Snore/Sleep Talk) which all seem to have specific AI behaviours associated with them and require specific code
EDIT: the biggest practical difference in Gen 2 compared with Gen 1 isn't an AI thing, but rather the fact that enemy Pokemon consume PP. This has two major implications, as I see it:
1) Obviously it makes it impossible for battles to last as long. Struggle damage and recoil will eventually force a result.
2) A Level 50 tournament would make less sense with Gen 2, because a lot of battles would be reduced down to which Pokemon can most effectively spam and withstand Struggle (not an especially interesting factor in an AI ranking project)
My memory is old, but I still feel like the gen 4 AI has some pretty complex decicions it makes when it comes time to go for KOs.
Like, it seems to favor perfect accuracy or priority moves if those will kill, and favors non-stab, non-SE attacks if available. It seems to run by something like simulated eyeballing, rather than hard numbers, so the AI will *try* to KO with a priority move, but fail, instead of using another move.
@@aprinnyonbreak1290 Plat/HGSS has some interesting ways of approaching KOs, but I don’t think it ‘favours’ less powerful moves in general. You’re right about the priority/eyeballing stuff though!
In more recent games the AI looks for KOs in a weird way: it ‘rolls’ each of its moves for damage, including the random 0.85-1x multiplier at the end, then if any of those moves are rolled to KO it picks randomly between any increased-priority attacks or, if it has none, picks randomly between regular-priority attacks that are rolled to KO.
However, the actual damage roll is done after move selection, so it’s possible for the AI to pick a move with, say, a 6.25% chance of KOing (there are 16 damage rolls for every damage calculation) because it got ‘lucky’ with its initial roll, only to fail to KO when it actually executes the move.
Wait, what? Wrap still HITS and stuns ghost Pokemon, it just doesn't deal damage?
Gen 1, you did it again...
Yep, the trapping effect does still happen.
So hypothetically speaking if a ghost was leech seeded or burned you could still use wrap to stall them out
Kanto, you've done it again.
@@ultimaterecoil1136 considering it's only the gastly line in gen 1 they'll probably outspeed and put to to sleep between wrap traps
@@thomaswawrzyniak118
That would still be a good few turns of leech seed/burn damage, so not a bad trade depending on the situation
The amount of work that goes into this project is amazing.
The Rival may have won the tournament and the championship, but the Juggler won our hearts! Outstanding job! ☺️
The amount of effort to:
Write code to simulate battles
Run a round robin
Have an ELO system
Be able to play back battles deterministically
This deserves a lot of views.
And people still call HIM stupid for calling a round-robin tournament a "tournament." Insane.
@@B3Band probably something to do with language barrier, since round robin are usually called a league in sport, which in other languages has a different word for it compares to knock-out type tournament
Wow Kadabra is insane, clearly we knew its line was strong, but it carrying the worst AI in the game so hard was really impressive
Well in Gen 1, Psychic type was broken. Psychic had no weaknesses. (It was weak to bug-type moves, but there were no bug-type moves in gen 1. Psychic was also weak to ghost, but a glitch in the game made Psychic types immune to ghost!) Also virtually every bug and ghost type happened to also be poison type, which is weak to Psychic. (So it's only 2 'weaknesses' actually ending up being advantages, which explains why a trainer with 1 single Hypno could rank so high).
Twineedle and Pin Missile were both in Gen 1. Beedrill was sufficiently weak that even Executtor with its 4X bug weakness could still win a 1v1.
@@carl4889 Oh yeah. I forgot about those moves because they were insanely bad.
Pinsir doesn't get any bug stabs in Gen 1. It does get hyper beam, slash and bind and swords dance though, so could be pretty good in the right hands
@@justinmadrid8712 Weaknesses weren't really a problem since it's very frail and could be one shot without a weakness anyway. It's sheer speed and special stat was just able to shine since pokemon that would normally outspeed via level were wittled down so only certain exceptionally fast Pokémon could attack and stop it. Plus Alakazam having Recover meant there was a good chance it could waste a turn instead of attacking by Recovering at full health while Kadabra didn't have this so just attacked first time it could more often.
I love how the tournament at the end doesn't actually have to be a real one.
piman had the footage of all the matchups anyway, and could just cast and present them as though they were sequential.
brilliant
The 3D animations are so beautiful, especially with the font style from the original games. It looks so neat, well done!
Considering in the end, the Venusaur and Charizard Champions and Oaks tended to do better than the Blastoise ones, you may think Blastoise is just the worst of the three, but I think it's something else
I think Gyarados is simply the strongest of the other trio of Grass-Fire-Water Pokémon the rival and Oak have, and since they don't have Gyarados when they have Blastoise then that means they have the disadvantage of not having Gyarados, rather than it being from not having one of the other two starters
You can even see how well Gyarados tends to do in the battles described in the top 16 tournament, especially Lance's Gyarados in his first battle
So basically my take away is Gyarados and Alakazam are the star players in this whole thing and probably some of the best Pokémon to use as a player too (but we absolutely knew that for Alakazam anyway)
Nah Gyarados is absolutely an S tier pokemon, hits hard, bulky, good type resistances, etc
At 47:40 you can see the Blastoise Oak has Gyarados and Blastoise. This doesn't apply to the Blastoise champ though, so you're right about that.
The biggest reason is actually pretty simple: Venusaur has Razor Leaf. One of the fun Generation 1 bugs is that the global rate for moves to be critical hits is four times higher than it's meant to be, as if all Pokemon are boosted by Focus Energy (which in Gen 1 hilariously reverts the Pokemon that uses it back to the intended critical hit rate). Combine that with a move that has an increased critical hit chance (and Venusaur's respectable speed at base 80. Critical hit chances in Gen 1 were also modified by the base speed of a Pokemon) and you have a Pokemon that is doing 2x damage almost every attack.
So basicly all the non-blastoise teams just get a better blastoise to replace him.
"Where is Youngster Joey?"
"...In Johto. Youngster Joey is in Johto."
Honestly, this is my favorite part of the video.
And I'm honestly kinda surprised about the Charizard Champion and Oak losing in the semi-finals. As much as I love Charizard, I assumed the #1 Elo was because the Exeggutor+Gyrados combo was stronger than the Arcanine+Exeggutor/Gyrados combo, as their teams were otherwise the same outside of that(And their starter obviously). But since those won...maybe it's the other way around, and Charizard itself WAS the reason for the higher ranking. Interesting regardless.
I wish you the best of luck with getting future later gen tournaments up quickly. Both from experience and for your sanity.
that Charizard is a double edged sword in red and blue, it has strong stab in flamethrower and good speed... but it also has rage where once selected, it's stuck using it and that move is genuinely horrible in gen 1.
Venasaur has razor leaf which is a bit broken in gen 1 and solar beam which is a turn waster to be honest. Just to put things into perspective.
That and Gen 1 AI being kinda dumb and trying to use weird moves against Charizard because of the Fire/Flying type (trying to Earthquake it because fire is weak to ground and having it fail because the game doesn't realise it's also flying too)
I would personally refrain from labeling one team better than another because of a single tournament game result. The #1 elo rating, I think, is a better rating of consistancy. It's not unheard of that a weaker player beats a stronger player in a game. It's just that if they played many games in a row, the skill of the stronger player will shine through via law of large numbers. Pokemon especially has extra RNG involved that can disrupt the outcome of a single game.
That said, tournaments are really exciting - partially for this reason. This was no exception for me.
Charizard is much better than Arcanine, while Venusaur isn't better than Exeggutor and Blastoise is way worse than Gyarados. Plus, Exeggutor doesn't have the same type priority glitchiness as Venusaur, giving it a huge edge there.
@@darkphoenix2745 Charizard is also faster and does more damage with its fire moves. Plus, in this specific bracket, the more threatening foes rarely use rock-types, and when they do, they usually die easy to flamethrower anyway, AND the best rock move any foes's gonna have is Rock Throw, hurts Charizard a lot sure but it's as powerful as EQ is against Arcanine
I absolutely ADORE Both ranking videos. I’ve watched both multiple times. Insanely entertaining and incredibly informative. The videos are sort of a comfort!
Thank you for making these.
Wow, this deserves _way_ more attention.
Fantastic premise and execution, and the guy himself is pretty entertaining too. 10/10 video
This is one of the most unbelievable TH-cam videos I've ever seen in terms of consistent quality and attention to detail. No wonder it's been a year since the last one! This is such a silly and - let's be real - inconsequential topic, but thank you for giving it so much effort and polish. I had a ton of fun watching it!
P.S. gen 2 when ??? 😏
in the end he said one day so lets wait for the next tournament and a possibility of the 16 best trainers being our final battle against youngster Joey and maybe win the final battle on the tournament style battle of the 16 or 15 since normal types are op in gen 1 and 2
Agreed
A gen 2 would be interesting to see. Probably wouldn't be a shocker to see Red and Blue at the top of the list, even if levels were set to 50. Both of their teams compared to everyone else are head over heels much better.
"Your video is incredible, also useless, when's part 2?" XDXDXD
I actually screamed when he starting going into the math of Elo and we crossed into a 3blue1brown video, especially with the music
The only things left are the diagrams, the little thinking bubbles and characters and it'd be 110% a 3blue1brown video!
My guy even used Manim lmao, the amount of effort in this video is honestly mindblowing
I genuinely did too, and started laughing way too hard
Oooh! I wonder if comments like this are why this video was recommended to me 😂 I'm not any good at math, but "3 blue 1 brown, but about pokemon" would instantly get my attention as a video log line, lmao
Wow. I know of very few other TH-camrs who would go back and redo an incredibly polished, painstakingly edited, thoroughly enjoyable video of theirs in order to properly address the legitimate criticisms from the commenters and make the video just that much better. I cannot even begin to express how much respect I have for you right now. This took so much time and effort and clearly came from a desire to truly deliver the highest quality video possible. I'm legitimately shocked at how good this video is. You're awesome. Thank you for this.
This was wonderfully thorough and I just want to say how much I appreciate the time it took to make this video. My friends and I can't wait for another.
You didn't mention the third type of comment in the comments section:
Comments acknowledging your god-like editing and presenting skills.
But we all know that the real champion is the one, who brought us this educational content.
Can I just say that you take over analyzing mildly interesting concepts and add such a level of comedy that this truly becomes a masterpiece, this is my first video of yours and I hope to find many more!
So happy the algorithm gave me this video, this (and the previous video) have been really interesting insights into the Generation 1 games I never considered too much. Thanks for the awesome content and look forward to the next one some day 😄
I don't know what to say other than "thank you". Thank you for putting all the time and effort into creating this beautiful masterpiece of data. I never thought we'd see a revisit, and I couldn't be happier to stumble across this once again.
I laughed so hard at the 3blue1brown music, I thought it was going to be some parody of it but it was actually a great explaination.
This video quality is insanely good, I wish I had the same skillset.
Man I wish i could understand the mumbo jumbo he's saying in the math zone.
I watched your trainer ranking the other day,that's a crazy idea for a video I loved watching it so much effort. Great work
Holy shit.... This is truly a labor of love. Gotta thank youtube for having me stumble on your channel. This was fantastic! Appreciate all your hard work!
Honestly the fact this sequel took over a year but the video it was a sequel to was immediately recognisable really speaks to the quality and originality of this series. Good on you for taking the time you need to make it memorable
"I'm the strongest trainer in the world!" - Blue, champion battle.
He is not entirely wrong, he's just uhh, one of the strongest trainers from Kanto. All three parallel versions of his peak, er, champion himself, hogged the top 3 ranks, the S+ tier spots.
Red: are you sure about that
Of fuck he's going to to need to do one giant video when its all said and done comparing all the champions, isnt he
@@DarthAnimal wait why did you have to fight Lance if he was just dethroned as the champion by blue. That's why I like gen 5.
@@waterierStone because he still had his elite four position to fall back on.
@@tonyhakston536 but who was the prior champion? before Gary reached there and beat him?
the quality of research, implementation of everything, animation, storytellingyand the explanations on the technical side are all top notch, and you deserve to get a bigger following off this, just hoping you don't set your standards too high and get in a cycle of never bringing out any content because it's not good enough, that would be a shame, because we need more of this :)
This is a great video, but holy HELL the WINNER sound effect scared the crap out of me.
Viewers: "When is Gen II coming?"
pimanrules: "Look! Over there! A distraction!"
I appreciate the transparency and the details of this video. The comments scouring, the mathematics, the programming, the trouble shooting, the presentation...
Well done!
"Johto. Youngster Joey is in Johto." I saw the answer coming but your delivery on that was perfect
I loved the first video and I loved this video too. Excellent work, I whole-heartedly commend you on your dedication and willpower.
actually, trying to reimplement the trainer AI in later generations would be a tall task, even considering the source code decompilation projects of DP, RS, etc...
I feel it could work for Gen 2 maybe Gen 3(?), but I understand if it can’t be done.
Actually the AI for Gen 2 is just a build off of the AI of Gen 1 really. Mostly dealing with bug fixes with Gen 1. Such as trainers picking Super Effective moves even if they do no damage.
As well as adding things. Such as certain Trainers now getting specific moves to each dedicated Pokemon. And adding that trainers have PP as well.
It would not be EASY. But it wouldn't be very difficult. At least as difficult as doing this with other Generations.
Plus how would you handle instances of forced double battles like Tate & Liza?
As such I feel Gen 2 is the only other one still possible
@@skibot9974 You don't make it forced. You just run it like a Double Battle without the script of Tate and Liza forcing you to double battle. There are other Double battles in the game that don't force you to use more then 1 Pokemon.
@@linhero797 what I meant more is what about trainers that are always pairs? Also double battle trainers tend to have movesets tailor made for doubles so making them singles wouldn’t give accurate results
This is one of the most insanely high-effort videos I've ever seen.
Just gotta say? I really do appreciate the fact you decided do do a redux even thought the original idea was still so solid to begin with. The changes and implementations are fascinating g to hear about in how you address them in code/methodology, and while I totally understand you deserve a break after all that? Even if it takes a while, I'd watch a similar video on another generation hands down. Earned my sub 9x over mate
Really enjoyed this video. Much respect to the work you put in on this. Loved the tournament
50% of the vid was me nodding along pretending I have any clue what is being said. While the other 50% is me loving a good Anime Tournament arc
You are definitely hitting that "I don't care what their videos are about, I just like how they're doing it" level of quality. Also code review would be very interesting.
I appreciate your thoroughness. It's always something that I've wanted to do. While I do possess the mathematical wherewithal to run elo based systems, I severely lack the programming that makes such an undertaking feasible. I anxiously await more Pokemon related content
You're insanely talented, man, so much hard work goes into this vids.
So what you're saying is that Bulbasaur is best starter.
Ah, victory.
A bit of a fact since from what I recall, the game is balanced around the player choosing Charmander
It always have been... Bulbasaur is the "easy" mode because it's the strongest.
@@USELESSDCK Nope, more like Bulbasaur -> Easy mode, Squirtle -> Regular, Charmander -> Hard mode. Bulbasaur has advantage against the first 2 gym leaders and resistance to the first 4, Squirtle has advantage against Brock and resistance against Misty, Charmander gets walled by the initial gym leaders.
heck yeah bulbasaur
Maybe.
It could also be that Exeggutor is holding the team back, except on teams that don't have exeggutor. In all the champion matches with it, it seems to be a really dead pick.
Omg this is so hyped
Same!
More hype than a Hyper Potion
Firstly, this was incredible.
Secondly, the amount of different skillsets and the depth of knowledge required to make these videos... the understanding of the game, the time and patience to invest into it, the idea for brackets and balance, the python coding, mathematical understanding, video editing, graphic creation and not to mention great humour throughout... this is one of those rare occasions where someone has invested so much to create a genuinely high quality video. Excellent stuff and a very easy sub from me to see whatever you might bring out in the future!
Now I'm curious to see how the FRLG remakes do with all their improvements to the battle system and AI.
Also I think the "expanded/refined/third/whatever" version (that is, GSC -> Crystal, RSE -> Emerald, and DPPt -> Platinum) should probably be used for gens 2-4.
well this was based on the "expanded/refined/third/whatever" version, so it makes sense to do the same with the following ones
This one isn't yellow, otherwise some specific important trainers would rank higher or lower depending on the moveset and Pokemon changes. Sabrina, Erika and Lt. Surge would become weaker while a few others would get stronger with Yellow changes but it would be an interesting video for sure
The production value on this video is insane.
46:45 "Agatha didnt really stand a chance"
One could even say she didn't stand a *ghost* of a chance
*wink wink*
“Yugi”
How do I delete someone else's comment?
@@newaccount3743 Delete your eyes
- 4kids
Uggggghhhhh
Great work, as before! I envy you for your hard work and for having the idea in the first place. Very clever approach to the many obstacles.
The amount of work you put into this is insane, this deserves a LOT more views!
I wish this had merch. Id love a acrylic stand with the trainer sprite and card. Good work over all, I look forward to what you do next year!
actually though, I'd buy the crap out of these. They're super clean and awesome to look at.
Forget those other games you listed, I wanna see the enemy AI ranked from Colosseum and XD. As fun as that would be to see, it would probably be insanely challenging since those games have some of the most complicated AI of all of them, especially considering the double battling mechanic.
This is exactly what I was thinking, I'm glad I checked the comments before commenting it.
This is your first video for me, I thoroughly enjoyed it. Thank you for your time and effort.
Daaayum, Blue so strong, he really deserves it. I really felt the tournament! Kudos to you, sir for making this awesome video.
It would be interesting to make a pokemon RBY battle simulator rom, which lets you pick trainers and even design a custom trainer and either watch battles with animations or just simulate them all on the GameBoy
Id kill for this
I mean it sounds like our dude here basically has made that, and he explained in pretty good detail how he did it. That said I wouldn't pressure anyone to release their code
I know I'm kinda late, but Pokémon Showdown has a Gen 1 format, which is similar enough to whar you're saying.
You know the captions on this video are particularly excellent! Thanks for that :)
With Gen 2 coming.... eventually, I can't wait for how the AI fairs against each other. With Psychic being nerfed, Dark and Steel being created, 100 new pokemon, 8 new gym leaders, and Whitney's miltank it's going to be a wild ride. I'm really hoping MandJTV is right with Karen as the champion because now looking at her "Dark" team in G/S/C she has some good type coverage.
I think whitney will rank lower as expected since she is pretty relying on that miltank lol.
She will still rank high but not as high as people expect her to be since she effectivly only really has 1-1,5 pokemon.
@@jeltevl8868 Kinda depends on the methodology chosen, too...
@@jeltevl8868 she will get some wins via metronome luck too. Heck she might get some wins against far stronger teams with bad ai off some status moves from metronome
@@ultimaterecoil1136 oh god how the hell is metronome gonna work in such a format
Fares* 🙂
I love that you used the International Phonetic Alphabet when discussing the appropriate pronunciation of Elo. It made the differences in pronunciation so much easier to see
If there's ever another entry in this series (completely fine if there isn't, by the way, the amount of time this took for a game as simple as gen 1 is staggering), platinum would definitely be my pick. Absolutely phenomenal video, hats off to you, and maybe take a rest on these fascinating yet very high effort videos.
Absolutely mesmerizing!
I was in a trance of pure concentration during this and the former video. Loved it!
me, listening to all the math and coding involved in the programming: i like your funny words magic man
Great video! I really like the "conveyor belt" animation for the trainers.
I think the most intriguing aspect of this sort of tournament is the possibility of a result that *doesn't* match the order in which each trainer is fought.
Case in point, the Juggler that made top 16, and Bulbasaur rival consistently being ahead of the 2 corresponding rivals.
@@magiclapras I probably would have been much further out of order if each Pokémon had the moves that it would have at level 50. That would be pretty neat since, to be honest, a level 50 that only knows splash isn't really a level 50.
I can't keep up with the math of this but I really admire all the work you put into this, and the presentation of the video itself is really well done too.
Utterly fantastic video. Good work.
"Someday" implies legally that there will one day be another on this. So good. That's what I want.
Since the last video I've started a career on computer science, now I can understand a lot better how much work went into this, it's amazing
Awesome work. Really loved the video and look forward to what you will make next.
i love hearing people talk endlessly about things they love, even if i don't understand it. i have no idea what you're talking about when you get into the math and codey stuff, but i still think it's really enjoyable to watch because it's so clear that you're having fun. i wish more people were willing to listen to long winded rants about topics people are passionate about instead of just telling people to shut up
The reason why the Bulbasaur line rival starts out ahead and then goes behind is that it's bad matchups are disproportionately high ranked (psychic types). In the lower end of the rankings this means it loses less elo on average per loss, and then when it's elo gets higher it loses more elo on average per loss.