Ok, so I'm probably not going to make another follow-up to this, but I did update the code a bit to address some things you all have brought up in the comments: First, I only generated one ally space per game, but in reality most boards have 4. So, I've updated the code to generate 4 ally spaces instead of just 1 to hopefully help Daisy and Shy Guy out a bit more. Second, it is possible to plan ahead in this game and try to position yourself so you've in a better spot for next turn. So, for example, Shy Guy can attempt to land on a space that's 4 spaces away from an Ally space so that he can guarantee himself an ally the following turn. Daisy can't do this as reliably, but she since he has a very high chance of rolling a 3 she can attempt to land on a space 3 spaces away from an ally. So, I've updated the code to allow for a tertiary priority of targeting these spaces. After rerunning the simulation, I found that these... barely changed anything. In fact, Bowser's win rate actually went up by 2%! (Probably due to just random variance), and Daisy and Shy Guy's win rates actually went down below that of Steve! I'd have to look closer at the data for a more thorough analysis, but my guess is that this is largely due to what I mentioned in the video. Since Bowser's die is so much more powerful than Daisy, Shy Guy, or the standard die, anyone else getting allies isn't all that helpful. And while adding in more ally spaces gives Daisy and Shy Guy more opportunities to utilize their unique abilities, it also gives Bowser a higher chance of just happening to land on an ally space anyway. It's also important to note that the turn limit is a serious factor here. Games are relatively short, and if you're spending 2 turns setting up to get an ally, in a 10 turn game that's 20% of your total actions, leaving you less time to get stars. In order for Shy Guy and Daisy to be significantly better than Bowser, they need to have 2 more allies than him and they need to get those allies fast. Even in a 20 turn game, this seemed to be a challenge for the AI to pull off. It seems that, in this updated simulation, Daisy and Shy Guy ended up wasting a fair amount of time trying to get allies that didn't help all that much in the long run. As for bad luck spaces, as I mentioned in the video they make the AI very complicated in edge cases where both an ally and bad luck space are within range, but also every character with the exception of the control is very good at avoiding specific spaces. Since they affect all characters equally, they likely wouldn't have much of an effect on the final result. Also, for everyone clamoring for me to add in coins, that's a LOT easier said than done. Simply adding in red and blue spaces and awarding coins from minigames at the end of each turn wouldn't truly capture the coin's effect on a game, I'd also need to factor in everything that one can spend coins on. That would mean programing in every item in the game and AI on when each character should purchase and use each item. At that point I'm not really "simulating" Mario Party, I'm rebuilding the whole game from scratch! And that's a lot of work for one guy to do in the span of, like, 2 days... So, to try to approximate this, I've added in yet another feature where if Bowser reaches a star, there's a 5% chance that he fails to pick it up. Based on my experience, if you're reasonably skilled at minigames you'll probably have more coins than you can hope to spend by the end of the game, but there is a chance that Bowser's minus 3 will screw you over. Adding this in did drop Bowser's overall win rate by around 3%, but he was still a solid 10% ahead of Daisy in 2nd place! Hope that answered some of your questions/concerns, I'll update the spreadsheet in the description to include the data form this updated simulation as well!
You put in so much work into both these videos AND this follow up, super impressed. Glad I know now who to pick in the mario party game I don't even own!
I'm sad to hear that you aren't planning on following this up with more videos. You've just opened a whole can of worms that I would love to see someone eventually calculate. Thank you for these videos though! The hammer bro vid got me *die*ing of laughter, and the calculations provide a very interesting analysis of the game. I hope you find other cool stuff to calculate as you move on from the Super Mario Party stuff!
I like the work you've put into this, real dedication and I think it shows the best lesson from this: that sometimes, adding more variables and complexity, just doesn't change the outcome. Sure, we can't say for 100% certain (I mean it was all chance to begin with) but Bowser has remained #1 from the start of your last video to this pinned comment, only sorta losing the spot if you really cared about coins. The only thing I could really see impacting the outcome would be social dynamics, players ganging up on one another or making truces or rivalries, but like... fuck simulating that to any relevant depth.
I’m gonna predict in two weeks you’re gonna come back again due to community backlash and reveal that you used an emulator to simulate 1000 real all expert CPU games of Super Mario Party only to still conclusively show that Bowser is best
A lot of people here are unconvinced by statistics giving unintuitive results, which is understandable, but doesn't mean they're correct. Also, it's a game, and how it feels to play as a character is far more important than the mathematically optimal one. Particularly as it's a casual game, choosing to play as a character that minimizes bad experiences (Daisy/Shy Guy) can be far more important because of human biases in memory and similar factors. Missing a star 2 spaces away because you're playing Bowser and you failed the coin flip will stick in your mind far more than the times that you were playing Daisy and failed to get close quickly because you're losing half a space a turn relative to other players. I think that people might be more convinced by a different type of analysis which doesn't rely on statistical analysis of simplified simulations, but I'm probably wrong about that, and doing such an analysis would be really hard, let alone presenting it in a digestible way.
Mfs complaining that bowser is still the decisive #1 here despite this test not even displaying Bowser in the best light possible. He has more options than any other character (having the ability to roll every number except a 7), he reaches shops quicker than other characters which bolsters his movement even more because he can lean on golden shrooms to move him when his dice fails him, he gets more burst movement meaning that he has more options when he DOES roll big, which indirectly can actually increase his odds of landing on key spaces like ally spaces since more move means more branching paths which means more options to land on. Daisy and Shy Guy are solid dies but even in the consistency department they get absolutely dwarfed by Wario who can pull the shy guy move of guaranteeing you don't roll on any harmful spaces 1-5 spaces in front of you and can also guarantee one particular number to secure early allies, all while having a die that beats the normal average, so if anything the consistency test should have included him to show that he decisively beats everyone not named Bowser as well. Bowser is the best and it isn't even close. Someday people will stop trying to say that he isn't.
Please remember that this is all entirely hypothetical and simplified to an extraordinary degree. This simulation has only 2 real components, which, to be fair, is 2 more than the last simulation, but is still not all that much in comparison to the actual game.
Yeah, but this simulation is mostly about movement factors to be able to buy stars, and the factors that are going to make the most difference are here
I kinda think super Mario party is just a broken game and more movement = better chance at winning. Some stuff left out would have favored the consistent characters, but some stuff left out would have favored bowser. Moving more also allows bowser to access item shops and star steals more often than his opponent. And doesn’t the characters consistency fly out the window when they get an ally?shy guys consistent 4 then becomes a 5/12 chance of getting a 5 or a 5/12 chance of getting a 6 when he gets an ally that can’t be turned off.
@@Georgeavocados An SPM review I saw years ago pointed this out that the function of reliable allies like Daisy, Shy Guy and Wario to target certain spaces spaces specifically goes out the window as you don't have a choice to not make allies roll their bonus +1 or +2, turning the strategy tumbling "like a house of cards", meaning it really is virtually is all about movement lots of times
Possibly, but since they give away coins so freely in this game and stars are so cheap, I feel like using Daisy or Shy Guy just to try to land on a blue space probably isn't worth it
@@TheChiptide Stars aren't the only thing that coins are used for; there are also item shops which you briefly mentioned but didn't really delve into. I understand that it is very hard to simulate these things, but it's a bit dishonest to tell everyone to play bowser just because he did well in the simulation.
@@jakemichal1220coins are not an issue in super Mario party. Winning a mini game gives you 10 coins, but second a third in this game also got coins. The golden pipe is 10 coins in this game, vs 25 to 30 in other games. All the prices are basically half or less of what they would usually cost in other games, while mini games handed out more coins. Inflation is so bad xD
It’s a logical result tbh. In SMP, the tool that’s made to give players more consistency IS items not allies : since an ally always provide you with a bonus die each roll, it ruins any kind of consistency, while something like a mushroom can garantee to any player that he we’ll travel at least 3 spaces for their turn
That still means that even under ideal circumstances he only has a 25% chance of hitting the sweet spot. Bowser's die, meanwhile, is so good (on average) that it's nearly a free ally unto itself.
@JETAlone12 My argument is not that Shy Guy is better than Bowser. Any accurate tier list will and should have Bowser, Wario, Boo in the top 3 (in that order). I'm arguing that Shy Guy may be good enough for that 4th place slot.
Yeah, but games in SMP are really short. Even in 20 turns, 2 turn is 10% of the game where you are just focusing on getting 1 ally. Plus, allies hurt consistency. Shy Guy moves from having a 5/6 chance to roll a 4, to a 5/12 chance of rolling a 5 and a 5/12 chance of rolling a 6. And if you want to ditch consistency for movement, than why would you even pick Shy Guy in the first place?@@owenaspinall2046
which still doesn't matter as you'll be gunning for a 50/50 chance whether you land on the space you want after getting an ally really odds and evens don't matter except in partner party, and even then as soon as you get a single ally it doesn't matter
Honestly, I think you're correct, but your reasoning seems very flawed. Shy Guy being able to switch between parity really doesn't matter when talking about benefit. I still think it's a bit dishonest to just say "they benefit everyone equally" without actually justifying it. The amount that's added is the same for everyone, but I don't think it's at all obvious that that results in it "not mattering".
“I’m gonna make this more fair for consistent characters like daisy and shy guy!” (Gets rid of 7/8ths of the situations you would want consistency such as lucky spaces, event spaces, chance time spaces, etc…)
Note that targeting spaces is not what you're doing most of the time as players are likely to get allies more often than not and they destroy any targeted strategy a special dice a player or an ally can have.
Props to you for doing the follow-up video, I think this adds a lot to your initial analysis, and I think you've done a really good job operating within your constraints. There's always more things you could theoretically add to the simulation but ultimately until you're perfectly simulating the exact game itself down to every bug, /someone/ will be able to claim you haven't done a detailed enough model. Particularly taking into account the modifications you made in your pinned comment, I think that you've made a good enough model to give a good idea of what to pick. If you wanted to explore this subject more, I think you'd have to get into a more theoretical space with more high-level game theory concepts for weighing expected value vs standard deviations and how much you value each aspect, with the idea of trying to find a formula you could tune to find what dice best fits your criteria, which doesn't rely on simulations but on the set-up of the game itself. That would be pretty complex and hard to present in an accessible manner, like you've managed to do brilliantly with your approach in these 2 videos. All that said, great job, thanks for making these videos, I enjoyed discovering your channel through them!
Honestly, I think people expect way too much from this simulation. This simulation in my opinion is pretty good. Even if all the missing things were added to the simulation, Bowser would still be a top-tier character. He even explained that no simulation would ever be perfect unless you basically recreate/code the whole game. It is just the basic fundamentals that statically can give you an advantage if everyone has EQUAL skills. But no one in the world will ever have the same amount of skill. This game isn't all about luck, it is mostly about individual skills that no A.I. can ever replicate. No one in the comment section factors "skill level" to be a variable. Hammer Bro would dominate everyone if he is a gamer, even though he may have A BIT harder time winning statistically. Yall just be hating on my glorious turtle.
I actually did try updating the code to add in some things that people mentioned were missing, and as you said it didn’t really affect the final results all that much, bowser’s just too damn good!
I think one important thing to consider is how spaces (aside from star spaces) are generally consistent every playthrough. This can ultimately affect whether you get an ally as early as even turn 2 or happening spaces on turn 1. For example, if I want to go for the happening star early or block the path to the star on Whomp Domino Ruins or King Bob-Omb's Powder Keg Mine, I would want to roll a 4 - so using Shy Guy (for example) to target that space with a very strong chance of success.
Really fun, and I liked that you did mention the shortcomings at the end - Although you didn't mention that you could plan more than one turn ahead in a normal game xP
Having 1 good space and no negativity spaces out of 32 i feel skeelws this for just movement. While i do agree that coins don't matter as much in this game as others are saying... There are more than one ally space per map, as well as more than one bad luck space per map. There are 4 ally spaces, and 5 bad luck spaces on whomp's ruins. Bad luck spaces should have a 1/8th chance to loose you a star, and 1/8th chance chance to give a star to the person in last. Flip side, there should also be good luck spaces. Good luck spaces have a 1/7 chance to steal an ally for an other unit, or to boost your next dice roll by +3.
Mario Party has so many variables to it that this simulation is unable to do it justice. Of course Bowser will win most of the time when everything favors him. That is textbook definition of a bias.
If you check out the pinned comment, I did make some updates to hopefully address some of these issues! TLDR, though, I added some of this stuff in, and it didn’t really make too much of a difference.
I can firmly say as someone who played bowser a lot that the coin penalty was negligible, they really do hand out tons of coins in that game. Also I’m pretty sure that if you included items, that would only help bowser more. Bowser moving the fastest means that he can reach the shops first and cycle the board more times, effectively getting more items than everyone else. Which can in turn make up for some of the problems with bowser’s die.
I think the consistency effect is largely being offset by the fact that non bowser characters have no chance in situations where bowser has a 1 in 6 chance. Example: the target square is 9 spaces away bowser has a 1/6 chance of getting it while others have no chance in the same situation. This situation is also more common as bowser's 8,9,10 window is much larger than daisys 3,4 or shy guys 4. Speed give oppertunities that the other characters simply lack. I would like to see the resaults of a modified bowser vs standard bowser. Like 0,0,1,9,9,9 to add consistency while keeping speed and seeing if it makes a diffrence.
Oh this is really neat! Undisliked the prev video, liked this one, and subscribed! Even though the simulation was still very simplified I think it did do the maths justice! I'd love to see a full win/tie/lose% list based off this video's logic!
An interesting follow up might be a tier list for the partner party mode! It introduces interesting considerations such as needing to get either an even or an odd roll in order to pick up a star
The only use case I can think of for certain dice, like daisy or shy guy, is in the partner mode where you can walk anywhere. You often need to roll an odd or even number just to stop on the star. Shy guy can always roll an even number, but once you get any allies it becomes a 50/50 on every character. You're still gonna take fewer turns to reach anywhere with bowser though.
TBH I feel that avoiding spaces is probably going to be more often useful for consistent dice. It's not that common that the space you want to be is exactly 4 away, but if there's something you DON'T want to land on 1, 2, 3, 5 or 6 ahead of you. Well then it's useful to roll a dice that's 4 or nothing and that's going to happen much more often I'd think.
I think the biggest flaw with the simulation that favors high movement characters is breakpoints. For a non-Mario example, imagine a typical RPG where you have an enemy with 20 HP. If you do 9 damage, it will take you 3 hits to kill. If you deal 10 damage you will 2 hit to kill, however if you increase your damage to 11 all the way to 19 it will still take 2 hits to kill, so 10-19 damage is practically the same because it does not reach a breakpoint. Going back to Mario Party, imagine a star is 4 spaces ahead. Bowser would have a 50% chance of reaching the star while Shy Guy has 83% chance, so in this case Bowser's higher expected value doesn't help because the relevant breakpoint is 4, you only need 4 and anything above that is overkill. Now consider if the Star is 5-8 spaces away, Shy Guy would have to roll 4 twice to reach it in 2 turns which is 69% while Bowser has a 50% to reach the Star turn 1 + 25% chance to reach the star turn 2 (low roll turn 1 high roll turn 2) for a combined 75% so now it favors Bowser. By doing this analysis instead of making a tier list based on game length, you can instead make a tier list based on how far the star is. If there are restrictions on the board state where stars can be, then there will certain important breakpoints that will arise where Bowser's high average value might be overkill while a character with lower average value will perform better due to being more consistent while still hitting the relevant breakpoint. Also another thing not taken into account is the potential of the next star being placed behind you, in this case Bowser's high speed can be a downside because he's more likely to have the star be placed behind him. Bowser's high speed is only an advantage if the star is in front of everyone, but if he will have to wraparound because the star is behind him then his speed is probably offset by the higher distance he needs to travel compared to the rest.
Great video, but I might say that it’s not just the probability of a dice roll on one turn that determines your chances of landing on a specific space. If you need to land on a space 10 spaces away, you have a higher chance of landing on that one than a space right in front of you, since you have more rolls. I don’t know how the AI accounted for this as well, but allies can also increase the chances of rolling specific values because of the extra dice they roll, because you can have multiple outcomes with the same roll, such as you rolling a 3 and your ally a two, or you rolling a 4 and your ally a 1.
The game truly is dependent on moving quickly, but for the same reason getting allies is so much better (since they add to your roll) and thus getting low variance rolls are also broken if they get you another ally. Since each ally adds to your roll (1-3 I think) and you can have up to 4 allies being able to force allies quicker can net you with the basic dice a higher average of up to 11.5 on average which out competes everyone tremendously. I think this study would be better if certain numbers (like however many spaces the ally ones are away from the start in Thwomp ruins with options to go slower to force an ally) could be calculated to in the data. Something like every 26 spaces is an ally and can be collected also at 28 with a loss of movement of 2. Characters can choose to use their dice over the average dice on the move before an ally space. I would bet Wario would come out on top in this case.
I'm so sorry to do this to you... but I won't be satisfied with the sim until you take into account coins. Those two -3s on Bowser could cost him the star.
I don’t know… there are mini games every turn and you get coins even for losing plus you often get coins for landing on spaces. I feel like you have to try to not have coins by the time you get to a star
Since stars are so cheap in this game, and you get some coins even if you don't get first in a minigame, I figured that the odds of you not being able to afford a star are so low that it wasn't worth all the complexity of coins. Like I mentioned in the video, if you're at the point where you're trying to min-max your die, you can almost certainly get enough coins that the occasional -3 won't actually matter all that much. Based on my experience, players usually end the game with a bunch of coins that they weren't able to spend, so it would really only come into play in a tie situation. Including coins may have cost Bowser a small handful of games, but in the grand scheme of things I don't think it would have affected the final rankings at all
@@TheChiptidein the official tier list made by the competitive community of this game Shy Guy is considered S tier while Bowser is considered really bad if I remember correctly
At first I thought my comments were auto deleted because I included a link, but very LOL at my comments being deleted just because I'm trying to offer more help and data on this project. I actively want more people to come together on solving this interesting problem as thoroughly as possible.
But if you land on an Ally Space, you can’t roll 0 anymore, since the Ally will always add 1 or 2 to your total. Being able to trigger the same space over and over again can potentially be useful in an actual Super Mario Party game, but would not be a factor for this simplified simulation.
It does strike me that either playing as Daisy, and having Bowser as an ally, or playing as Bowser and having Daisy as an ally, would probably be an optimal combination. Bowser's die to cover long distance and having Daisy's die for a high minimum roll when close to a star would be the optimal combination. Of course, you can't choose which ally you get, and this strategy is unworkable if one of the other players is Bowser/Daisy, so it's not exactly reliable. I do wonder if Daisy becomes more optimal if no other player is playing Bowser, Wario, or Boo, since her die poses her well to get an ally, and snagging one of them early is hugely beneficial to her. Obviously it's easier to just pick Bowser, but if you know that no one else is playing as him (or the other distance characters) it might be worth trying.
I personally like Boo's dice the best. It is in the top 3 best ones and has some pretty high numbers while still being decently consistent 2 fives and 2 sevens. And seeing as I am better at the minigames against most of my friends the -2 coins faces don't really matter to me.
Another fact about Bowser's dominance is that his 0's are actually beneficial if you consider that the space they are on is a positive space, for example, a lucky space, which if they landed the turn before they get a chance to get that same lucky space twice. And given the fact that most board spaces are blue, they give 3 coins when landing on them, which nullifies the -3 of Bowser's 0, which the idea of "but you lose coins" becomes "you do not gain coins", which is a huge difference if you compare them. Given what we see in the video, Bowser does not gain allies that frequently, which in a normal game is actually way more beneficial for his 0 values.
Love to see consistency accounted for this time. Was it accounted for perfectly? No. Am I satisfied with how it was accounted? Kinda. I also don't really care because nothing can convince me to not choose Shy Guy. I think a way to make allies not a liability for characters with consistent dice is to have an option that makes it so characters don't roll the extra di(c)e. Or just remove that and stick to the extra dice option. Either way, I do hope we get a third Mario Party with character dice on the Switch 2. Speaking of a third Mario Party with character dice, I personally believe the first Mario Party with Character dice (Star Rush) did it better, where allies spawn on spaces and you play as different coloured toads that have their character dice of the standard dice. I also believe these two games are the only ones with a grid system. Super Mario Party had allies pop up on spaces in the grid system, but that's in the Partner Party side mode. I do think Partner Party is better, for a multitude of reasons. You (kinda not really) can have a character's dice without them adding one or two to your role. Also, consistency is more important and also less useful for dies (that does in fact make sense in this context), which is really neat. Conclusion: I really like Mario Party games with Character dice and grid systems :D
The most accurate you could get in simulations is by actually simulating the game itself. There are 20 characters in the game, which makes the number of 4-player character combinations 4845. To test each of these combinations for 10, 15, and 20 turn games 1000 times each (total 3000 per party combination), you would end up with 14,535,000 simulations to run. Even if you sped up each simulated game to take just one second, it would take just over 168 days to finish computations. This would be the most rigorous method.
@@Neonlaserz I have code that runs each simulation in roughly a millisecond that I would love to work with others to get more eyes on or help this video series, but my comments about the code keep being deleted.
Okay, now to wait which Jamboree Buddy from Super Mario Party Jamboree is objectively better at winning, and which maps are the best to have them because Peach is pretty powerful in Rainbow Galleria.
As someone who does like to go back and play their favorite Mario Parties, this is one of my faves, and I will be using the information from these vids to my advantage. 👌🏻
If I had to guess, it would be whatever 4 have the least overlap in their dice, or one where each character good at landing on a specific number (Shy Guy for 4’s, Wario for 6’s, etc) so you can always try to target specific spaces. Interesting idea, though, I might have to look into that one more!
Extremely odd to only put one ally space on the board. Unlike most other ways the simulation can be improved, putting an accurate amount of ally spaces seems really easy. And it would definitely benefit Daisy and Shy Guy.
You’re totally right! I ended up updating the simulation to add more ally spaces that I detailed in the pinned comment, but tldr it didn’t really help Daisy and Shy Guy all that much. It did make it significantly easier for them to get allies, but it also made it a bit easier for Bowser to randomly get allies and undo all their hard work. While allies are powerful, they’re not enough to overpower the might of bowser!
@@TheChiptideyeah, it's pretty predictable that adding more ally spaces helps bowser more. Each ally you have, makes your rolls more inconsistent, so while they have an easier time getting their first ally, it becomes harder to aim for the next ally space. As a result, it's very expected that Bowser is the one that benefits from more ally spaces. The more ally spaces there are, the less consistency is needed to land on one.
You're still assuming very limited strategy for the consistency, only considering the special dice when a special thing is in range isn't enough, consistency is most valuable when you can plan ahead. They need to not just consider whether it will be more likely to get them to a special spot, but also which die is more likely to put them in an ideal spot to get to the special spot the next turn, or the turn after, etc. Of course, the further away it is, the less it will matter and the more you just want to get distance. Getting more coins easily will also make it easier to vuy something from the shop without screwing yourself on a future star, whereas losing coins does the opposite, so I'd argue that a character like Bowser is less likely to get stuff from the shop.
You see all of the code that he showed off in the video is basic computer science and only someone that doesn't know or recently learnt computer science would do a video about this and even if this video and code done in this video is basic it is still a very interesting and fun video 10/10
I feel a lot of the commentators object mostly due to feelings rather than facts (or at least a statistical implementation of one) -- lots of people who object here seems to really prioritize the flight over the fight response so I can see why that risky behavior even if it works in the long term even statistics wise isn't seen as effective IMO (similar PoV when someone says Focus Blast is "really 50:50" IMO).
Like I said in the first video, I never said you were wrong. I only ever said there was more to the game. Now with this video, I will say, even if Bowser's die is the best, I still will never use it, because I have such bad luck that I would probably only ever roll those "-2"s (my luck isn't why I was pushing Daisy so much though).
to put an end to those saying the -3 coins on bowsers die matters, you can actually just time the dice rolls in this game and just never have that matter everyones die is consistent
i think you should really add coins, red spaces, blue spaces, lucky spaces, and unlucky spaces. make it so every four turns the players all recieve 15 coins (equivalent to getting 1st, 2nd, 3rd, and 4th in minigames) Make it so stars cost 10 coins. I think ignoring coins and only having a single advantageous space put bowser at a large advantage. And i dont think that would be hard to simulate at all. adding coins is super super easy, red spaces and blue spaces are easy, and lucky/unlucky spaces just require adding another randomizer for when u land on it. Should be simple to implement. Makes me want to do it lol
The problem with adding in coins is that they wouldn’t really matter unless I also add in items for people to spend their coins on (otherwise they would be mostly negligible, as red spaces alone won’t be enough to prevent you from buying a star, even if you landed on them basically every turn). That makes things way more complicated, because you have to also program in ai for when each character uses each item. If I do all that, I’m not really simulating the game anymore, I’m basically rebuilding it from scratch! Kind of a lot of work for one guy in a couple of days…
@TheChiptide yeah I agree! The items are definetely well to hard for the scope and purpose of this video. I do still think adding more different spaces would help mitigate the advantage bowser has and make it less of a "whoever roles more wins", but yeah
The analysis still felt lackluster unfortunately :/ I don't really think there is a feasible way to do this where it doesn't feel lackluster tho. Like, in a real game, every single space does something, while this board only has 1 space you probably want to land on, and this 1 space is exactly the space that destroys your consistency, because it randomly adds either a 1 or a 2 to your throw. The simulation's purpose was to see whether consistent throws can give you an edge over bigger movements, but the way you choose to do this, was by making the only possible reward for consistent throws, being able to do bigger movements at the expense of consistency. Ignoring coins also ends up ignoring Bowser's dice's main downside, which doesn't feel all that fair. For the purposes of this simulation, it does make sense, because as you accurately stated, you earn enough coins to reliably always have enough for the star, but things like golden pipes in the shops, are pretty much equivalent to a star, and once you factor in those, coins do start mattering. Coins are also used as a tie breaker. So in games where bowser ends up with the same amount of stars as another player, bowser is more likely to loose due to having less coins.
Let's ignore coins, sure Bowser is losing 30% of the value of a star in coins every three turns, while being able to reach any star in 2 to 4 rolls, but whi cares about such thing, right?
While I agree that coins are not an issue, I think that there's a couple things to take into account regardless of that 1. Bowser spaces: These are likely to take away a star or remove econ, especially with communism. However, not only is Bowser probably likely to avoid the space unless it's right in front of him (in which case he should probably go for standard), because bowser has the -3 coin penalty and is picking up stars/items more often, communism will likely benefit him the most. Oh, wait. This is super mario party. Communism doesn't even exist for this, and the rest of the options are a lot more tame. dang 2. Bonus stars. Rich, sightseer, ally, event, there's a lot of factors that can matter with bonus stars. Minigame star is separate from boardplay, money is usually determined from who wins the most minigames, same with item, and the sightseer/slowpoke ones have already been determined, so the only ones that are likely to change things are happening ally/buddy, and misfortune. What you could do is put like 3 event spaces scattered across the map that ONLY contribute to happening star, (because misfortune is essentially the same thing) and cycle through 3 of a random list of sightseer, slowpoke, happening, rich, item, minigame, and ally at the end of a game, with the rich, item, and minigame stars randomly being assigned to single player, because they all have pretty much equal liklihood to gain those stars. This would then determine how impactful landing on specific spaces for allies and happening is compared to just going for speed and stars. 3. You love it, you hate it, you've all experienced turn 20 star swap into last place, it's CHANCE TIME! Absolutely useful for players in last place, maybe have the ai prioritize chance time when they're not in first, with a higher chance the closer to last they are (by stars, not just by placement) and also prioritize it the... wait. No, you're kidding me. SUPER MARIO PARTY DOESN'T HAVE CHANCE TIME??? WHAT? I guess that's less coding.
I agreed with you last time and still agree with your conclusion. The fans of this specific Mario party are trying to pretend like it’s more complicated than it really is. Super Mario party is one of the most broken Mario parties ever and moving more is all that matters. You didn’t even include items which would have made bowser’s win rate even higher.
If you move more, you can farm golden pipes or boo star steals more. The thing consistent dice can do is help you get an ally so that you can … move more.
@@Georgeavocados I agree that the mandatory bonus die allies give after getting them removes any consistency the Daisy and Shy Guy dice gives and devolves again to purely a movement only strategy for dice blocks...I don't think any board in Super benefits much on limited movement unlike say the boards in Superstars and Jamboree so long movement is almost always welcome strategy.
people really need to accept that bowser is superior, not only cause of statistics but also he is the supreme daddy so idk why everyeone is surprised xd
The fact that Bowser won the stripped down simulation is not a surprise at all. It is to be expected, since this simulation ignores the counter balance. The fact that Daisy and Shy guy have nearly the same win rate proves that there is a lot of merit to controlling rolls, a merit that only improves as you introduce the game's complexity, whereas Bowser's merits disappear as you add complexity. I suspect that you are a secret Hammer Bro enjoyer and are only making these videos to bias the public for greedy, less optimal Bowser picks. You understand the game is far too complex for movement as a victory condition.
Ok, so I'm probably not going to make another follow-up to this, but I did update the code a bit to address some things you all have brought up in the comments:
First, I only generated one ally space per game, but in reality most boards have 4. So, I've updated the code to generate 4 ally spaces instead of just 1 to hopefully help Daisy and Shy Guy out a bit more.
Second, it is possible to plan ahead in this game and try to position yourself so you've in a better spot for next turn. So, for example, Shy Guy can attempt to land on a space that's 4 spaces away from an Ally space so that he can guarantee himself an ally the following turn. Daisy can't do this as reliably, but she since he has a very high chance of rolling a 3 she can attempt to land on a space 3 spaces away from an ally. So, I've updated the code to allow for a tertiary priority of targeting these spaces.
After rerunning the simulation, I found that these... barely changed anything. In fact, Bowser's win rate actually went up by 2%! (Probably due to just random variance), and Daisy and Shy Guy's win rates actually went down below that of Steve! I'd have to look closer at the data for a more thorough analysis, but my guess is that this is largely due to what I mentioned in the video. Since Bowser's die is so much more powerful than Daisy, Shy Guy, or the standard die, anyone else getting allies isn't all that helpful. And while adding in more ally spaces gives Daisy and Shy Guy more opportunities to utilize their unique abilities, it also gives Bowser a higher chance of just happening to land on an ally space anyway. It's also important to note that the turn limit is a serious factor here. Games are relatively short, and if you're spending 2 turns setting up to get an ally, in a 10 turn game that's 20% of your total actions, leaving you less time to get stars. In order for Shy Guy and Daisy to be significantly better than Bowser, they need to have 2 more allies than him and they need to get those allies fast. Even in a 20 turn game, this seemed to be a challenge for the AI to pull off. It seems that, in this updated simulation, Daisy and Shy Guy ended up wasting a fair amount of time trying to get allies that didn't help all that much in the long run.
As for bad luck spaces, as I mentioned in the video they make the AI very complicated in edge cases where both an ally and bad luck space are within range, but also every character with the exception of the control is very good at avoiding specific spaces. Since they affect all characters equally, they likely wouldn't have much of an effect on the final result.
Also, for everyone clamoring for me to add in coins, that's a LOT easier said than done. Simply adding in red and blue spaces and awarding coins from minigames at the end of each turn wouldn't truly capture the coin's effect on a game, I'd also need to factor in everything that one can spend coins on. That would mean programing in every item in the game and AI on when each character should purchase and use each item. At that point I'm not really "simulating" Mario Party, I'm rebuilding the whole game from scratch! And that's a lot of work for one guy to do in the span of, like, 2 days...
So, to try to approximate this, I've added in yet another feature where if Bowser reaches a star, there's a 5% chance that he fails to pick it up. Based on my experience, if you're reasonably skilled at minigames you'll probably have more coins than you can hope to spend by the end of the game, but there is a chance that Bowser's minus 3 will screw you over. Adding this in did drop Bowser's overall win rate by around 3%, but he was still a solid 10% ahead of Daisy in 2nd place!
Hope that answered some of your questions/concerns, I'll update the spreadsheet in the description to include the data form this updated simulation as well!
You put in so much work into both these videos AND this follow up, super impressed. Glad I know now who to pick in the mario party game I don't even own!
I'm sad to hear that you aren't planning on following this up with more videos. You've just opened a whole can of worms that I would love to see someone eventually calculate.
Thank you for these videos though! The hammer bro vid got me *die*ing of laughter, and the calculations provide a very interesting analysis of the game. I hope you find other cool stuff to calculate as you move on from the Super Mario Party stuff!
I'm wondering whether wario would fair better instead of shyguy
Since stars are the way you win I wonder if just making stars always the top priority gives better results.
I like the work you've put into this, real dedication and I think it shows the best lesson from this: that sometimes, adding more variables and complexity, just doesn't change the outcome.
Sure, we can't say for 100% certain (I mean it was all chance to begin with) but Bowser has remained #1 from the start of your last video to this pinned comment, only sorta losing the spot if you really cared about coins.
The only thing I could really see impacting the outcome would be social dynamics, players ganging up on one another or making truces or rivalries, but like... fuck simulating that to any relevant depth.
I’m gonna predict in two weeks you’re gonna come back again due to community backlash and reveal that you used an emulator to simulate 1000 real all expert CPU games of Super Mario Party only to still conclusively show that Bowser is best
Nah, I’m good
@ lmao good, don’t bend to the will of unsatisfiable commenters haha
A lot of people here are unconvinced by statistics giving unintuitive results, which is understandable, but doesn't mean they're correct. Also, it's a game, and how it feels to play as a character is far more important than the mathematically optimal one. Particularly as it's a casual game, choosing to play as a character that minimizes bad experiences (Daisy/Shy Guy) can be far more important because of human biases in memory and similar factors. Missing a star 2 spaces away because you're playing Bowser and you failed the coin flip will stick in your mind far more than the times that you were playing Daisy and failed to get close quickly because you're losing half a space a turn relative to other players.
I think that people might be more convinced by a different type of analysis which doesn't rely on statistical analysis of simplified simulations, but I'm probably wrong about that, and doing such an analysis would be really hard, let alone presenting it in a digestible way.
luigi has the highest ally win rate due to it probably just having by far the best dice to go with bowser LMAO
Mfs complaining that bowser is still the decisive #1 here despite this test not even displaying Bowser in the best light possible. He has more options than any other character (having the ability to roll every number except a 7), he reaches shops quicker than other characters which bolsters his movement even more because he can lean on golden shrooms to move him when his dice fails him, he gets more burst movement meaning that he has more options when he DOES roll big, which indirectly can actually increase his odds of landing on key spaces like ally spaces since more move means more branching paths which means more options to land on.
Daisy and Shy Guy are solid dies but even in the consistency department they get absolutely dwarfed by Wario who can pull the shy guy move of guaranteeing you don't roll on any harmful spaces 1-5 spaces in front of you and can also guarantee one particular number to secure early allies, all while having a die that beats the normal average, so if anything the consistency test should have included him to show that he decisively beats everyone not named Bowser as well.
Bowser is the best and it isn't even close. Someday people will stop trying to say that he isn't.
Sup.
You're wrong ☝️🤓
Please remember that this is all entirely hypothetical and simplified to an extraordinary degree. This simulation has only 2 real components, which, to be fair, is 2 more than the last simulation, but is still not all that much in comparison to the actual game.
Yeah, but this simulation is mostly about movement factors to be able to buy stars, and the factors that are going to make the most difference are here
I kinda think super Mario party is just a broken game and more movement = better chance at winning. Some stuff left out would have favored the consistent characters, but some stuff left out would have favored bowser. Moving more also allows bowser to access item shops and star steals more often than his opponent. And doesn’t the characters consistency fly out the window when they get an ally?shy guys consistent 4 then becomes a 5/12 chance of getting a 5 or a 5/12 chance of getting a 6 when he gets an ally that can’t be turned off.
@@Georgeavocados An SPM review I saw years ago pointed this out that the function of reliable allies like Daisy, Shy Guy and Wario to target certain spaces spaces specifically goes out the window as you don't have a choice to not make allies roll their bonus +1 or +2, turning the strategy tumbling "like a house of cards", meaning it really is virtually is all about movement lots of times
The more components you add and the less consistency is rewarded
16:52
Statistical proof that *Luigi wins by doing absolutely nothing.*
I feel like the simulation still favored characters like Bowser because you didn’t include coins and the only one important space to land on.
Possibly, but since they give away coins so freely in this game and stars are so cheap, I feel like using Daisy or Shy Guy just to try to land on a blue space probably isn't worth it
@@TheChiptide Stars aren't the only thing that coins are used for; there are also item shops which you briefly mentioned but didn't really delve into. I understand that it is very hard to simulate these things, but it's a bit dishonest to tell everyone to play bowser just because he did well in the simulation.
@@jakemichal1220 I think we're forgetting this is Super Mario Party, the game with the cheapest items in the entire series
@@jakemichal1220coins are not an issue in super Mario party.
Winning a mini game gives you 10 coins, but second a third in this game also got coins. The golden pipe is 10 coins in this game, vs 25 to 30 in other games.
All the prices are basically half or less of what they would usually cost in other games, while mini games handed out more coins. Inflation is so bad xD
Don't hurt this man anymore. Let him rest.
It’s a logical result tbh. In SMP, the tool that’s made to give players more consistency IS items not allies : since an ally always provide you with a bonus die each roll, it ruins any kind of consistency, while something like a mushroom can garantee to any player that he we’ll travel at least 3 spaces for their turn
I’m not too concerned about the coins, but you’ve neglected the biggest factor of all: bonus stars! The king makers!
Agreed. Don't forget event spaces such as the present box where you can potentially steal another player's star as well.
Shy Guy shouldn't just target an ally if it is 4 spaces away. He should also do that if it's 8 spaces, or 12 spaces, etc.
That still means that even under ideal circumstances he only has a 25% chance of hitting the sweet spot. Bowser's die, meanwhile, is so good (on average) that it's nearly a free ally unto itself.
@JETAlone12 My argument is not that Shy Guy is better than Bowser. Any accurate tier list will and should have Bowser, Wario, Boo in the top 3 (in that order). I'm arguing that Shy Guy may be good enough for that 4th place slot.
Yeah, but games in SMP are really short. Even in 20 turns, 2 turn is 10% of the game where you are just focusing on getting 1 ally. Plus, allies hurt consistency. Shy Guy moves from having a 5/6 chance to roll a 4, to a 5/12 chance of rolling a 5 and a 5/12 chance of rolling a 6. And if you want to ditch consistency for movement, than why would you even pick Shy Guy in the first place?@@owenaspinall2046
I beg to differ, items DON’T benefit everyone equally when you include Shy Guy. They let him switch from even number spaces to odd ones at any time.
Yes they do
which still doesn't matter as you'll be gunning for a 50/50 chance whether you land on the space you want after getting an ally
really odds and evens don't matter except in partner party, and even then as soon as you get a single ally it doesn't matter
Honestly, I think you're correct, but your reasoning seems very flawed. Shy Guy being able to switch between parity really doesn't matter when talking about benefit. I still think it's a bit dishonest to just say "they benefit everyone equally" without actually justifying it. The amount that's added is the same for everyone, but I don't think it's at all obvious that that results in it "not mattering".
“I’m gonna make this more fair for consistent characters like daisy and shy guy!”
(Gets rid of 7/8ths of the situations you would want consistency such as lucky spaces, event spaces, chance time spaces, etc…)
Check out the pinned comment, I’ve made some updated to the sim that hopefully addresses some of this!
@@TheChiptide in this game, lucky spaces aren't nearly as powerful as in Jamboree or Superstars, also chance time doesn't exist in SMP
Note that targeting spaces is not what you're doing most of the time as players are likely to get allies more often than not and they destroy any targeted strategy a special dice a player or an ally can have.
Props to you for doing the follow-up video, I think this adds a lot to your initial analysis, and I think you've done a really good job operating within your constraints. There's always more things you could theoretically add to the simulation but ultimately until you're perfectly simulating the exact game itself down to every bug, /someone/ will be able to claim you haven't done a detailed enough model. Particularly taking into account the modifications you made in your pinned comment, I think that you've made a good enough model to give a good idea of what to pick.
If you wanted to explore this subject more, I think you'd have to get into a more theoretical space with more high-level game theory concepts for weighing expected value vs standard deviations and how much you value each aspect, with the idea of trying to find a formula you could tune to find what dice best fits your criteria, which doesn't rely on simulations but on the set-up of the game itself. That would be pretty complex and hard to present in an accessible manner, like you've managed to do brilliantly with your approach in these 2 videos.
All that said, great job, thanks for making these videos, I enjoyed discovering your channel through them!
I definitely recommend future Super Mario Party statistic videos like this but with Partner Party mode like best duo, movement, consistency, etc
bro's game theory 2.0
My exact thoughts
I open UA-cam and was blessed to be the first to witness this. I loooove statistical analyses in Mario games!
Somehow less of a linear board than Kamek's Tantalizing Tower
Honestly, I think people expect way too much from this simulation. This simulation in my opinion is pretty good. Even if all the missing things were added to the simulation, Bowser would still be a top-tier character. He even explained that no simulation would ever be perfect unless you basically recreate/code the whole game. It is just the basic fundamentals that statically can give you an advantage if everyone has EQUAL skills. But no one in the world will ever have the same amount of skill. This game isn't all about luck, it is mostly about individual skills that no A.I. can ever replicate. No one in the comment section factors "skill level" to be a variable. Hammer Bro would dominate everyone if he is a gamer, even though he may have A BIT harder time winning statistically. Yall just be hating on my glorious turtle.
I actually did try updating the code to add in some things that people mentioned were missing, and as you said it didn’t really affect the final results all that much, bowser’s just too damn good!
Based on the responses you're either gonna have to devote a year to perfectly recreating Mario party or never talking about it ever again 😂
Add coins in a simple way, just give them all a coin meter and increase their coins by some amount per turn.
I think one important thing to consider is how spaces (aside from star spaces) are generally consistent every playthrough. This can ultimately affect whether you get an ally as early as even turn 2 or happening spaces on turn 1.
For example, if I want to go for the happening star early or block the path to the star on Whomp Domino Ruins or King Bob-Omb's Powder Keg Mine, I would want to roll a 4 - so using Shy Guy (for example) to target that space with a very strong chance of success.
Really fun, and I liked that you did mention the shortcomings at the end - Although you didn't mention that you could plan more than one turn ahead in a normal game xP
Having 1 good space and no negativity spaces out of 32 i feel skeelws this for just movement.
While i do agree that coins don't matter as much in this game as others are saying...
There are more than one ally space per map, as well as more than one bad luck space per map. There are 4 ally spaces, and 5 bad luck spaces on whomp's ruins.
Bad luck spaces should have a 1/8th chance to loose you a star, and 1/8th chance chance to give a star to the person in last.
Flip side, there should also be good luck spaces. Good luck spaces have a 1/7 chance to steal an ally for an other unit, or to boost your next dice roll by +3.
Mario Party has so many variables to it that this simulation is unable to do it justice. Of course Bowser will win most of the time when everything favors him. That is textbook definition of a bias.
If you check out the pinned comment, I did make some updates to hopefully address some of these issues! TLDR, though, I added some of this stuff in, and it didn’t really make too much of a difference.
I can firmly say as someone who played bowser a lot that the coin penalty was negligible, they really do hand out tons of coins in that game.
Also I’m pretty sure that if you included items, that would only help bowser more. Bowser moving the fastest means that he can reach the shops first and cycle the board more times, effectively getting more items than everyone else. Which can in turn make up for some of the problems with bowser’s die.
"accounting for all that would make my computer explode and take ages"
*looks inside*
*Python code*
yeah, that checks out.
I think the consistency effect is largely being offset by the fact that non bowser characters have no chance in situations where bowser has a 1 in 6 chance. Example: the target square is 9 spaces away bowser has a 1/6 chance of getting it while others have no chance in the same situation. This situation is also more common as bowser's 8,9,10 window is much larger than daisys 3,4 or shy guys 4. Speed give oppertunities that the other characters simply lack.
I would like to see the resaults of a modified bowser vs standard bowser. Like 0,0,1,9,9,9 to add consistency while keeping speed and seeing if it makes a diffrence.
This video delights me, more statistical analysis? **chef’s kiss**
We also missing the fact that Bowser has a one in three chance of losing coins which may cost him Stars.
Oh this is really neat!
Undisliked the prev video, liked this one, and subscribed!
Even though the simulation was still very simplified I think it did do the maths justice!
I'd love to see a full win/tie/lose% list based off this video's logic!
An interesting follow up might be a tier list for the partner party mode! It introduces interesting considerations such as needing to get either an even or an odd roll in order to pick up a star
The only use case I can think of for certain dice, like daisy or shy guy, is in the partner mode where you can walk anywhere. You often need to roll an odd or even number just to stop on the star. Shy guy can always roll an even number, but once you get any allies it becomes a 50/50 on every character. You're still gonna take fewer turns to reach anywhere with bowser though.
TBH I feel that avoiding spaces is probably going to be more often useful for consistent dice. It's not that common that the space you want to be is exactly 4 away, but if there's something you DON'T want to land on 1, 2, 3, 5 or 6 ahead of you. Well then it's useful to roll a dice that's 4 or nothing and that's going to happen much more often I'd think.
I think the biggest flaw with the simulation that favors high movement characters is breakpoints.
For a non-Mario example, imagine a typical RPG where you have an enemy with 20 HP. If you do 9 damage, it will take you 3 hits to kill. If you deal 10 damage you will 2 hit to kill, however if you increase your damage to 11 all the way to 19 it will still take 2 hits to kill, so 10-19 damage is practically the same because it does not reach a breakpoint.
Going back to Mario Party, imagine a star is 4 spaces ahead. Bowser would have a 50% chance of reaching the star while Shy Guy has 83% chance, so in this case Bowser's higher expected value doesn't help because the relevant breakpoint is 4, you only need 4 and anything above that is overkill.
Now consider if the Star is 5-8 spaces away, Shy Guy would have to roll 4 twice to reach it in 2 turns which is 69% while Bowser has a 50% to reach the Star turn 1 + 25% chance to reach the star turn 2 (low roll turn 1 high roll turn 2) for a combined 75% so now it favors Bowser. By doing this analysis instead of making a tier list based on game length, you can instead make a tier list based on how far the star is. If there are restrictions on the board state where stars can be, then there will certain important breakpoints that will arise where Bowser's high average value might be overkill while a character with lower average value will perform better due to being more consistent while still hitting the relevant breakpoint.
Also another thing not taken into account is the potential of the next star being placed behind you, in this case Bowser's high speed can be a downside because he's more likely to have the star be placed behind him. Bowser's high speed is only an advantage if the star is in front of everyone, but if he will have to wraparound because the star is behind him then his speed is probably offset by the higher distance he needs to travel compared to the rest.
You know, I never even thought to consider consistency last time, but I had fun with the last video so I'm watching this one too
The reason Luigi is the best of the avialable allies is that they have an effect that matters, they increase your dice options.
I did NOT expect you to make another video on this. Good!
Did the simulations account for bowsers -3 coin rolls in the way where he would not move if he landed on an ally space?
I will not stand for this Hammer Bro slander
Great video, but I might say that it’s not just the probability of a dice roll on one turn that determines your chances of landing on a specific space. If you need to land on a space 10 spaces away, you have a higher chance of landing on that one than a space right in front of you, since you have more rolls. I don’t know how the AI accounted for this as well, but allies can also increase the chances of rolling specific values because of the extra dice they roll, because you can have multiple outcomes with the same roll, such as you rolling a 3 and your ally a two, or you rolling a 4 and your ally a 1.
The game truly is dependent on moving quickly, but for the same reason getting allies is so much better (since they add to your roll) and thus getting low variance rolls are also broken if they get you another ally. Since each ally adds to your roll (1-3 I think) and you can have up to 4 allies being able to force allies quicker can net you with the basic dice a higher average of up to 11.5 on average which out competes everyone tremendously. I think this study would be better if certain numbers (like however many spaces the ally ones are away from the start in Thwomp ruins with options to go slower to force an ally) could be calculated to in the data. Something like every 26 spaces is an ally and can be collected also at 28 with a loss of movement of 2. Characters can choose to use their dice over the average dice on the move before an ally space. I would bet Wario would come out on top in this case.
I'm so sorry to do this to you... but I won't be satisfied with the sim until you take into account coins. Those two -3s on Bowser could cost him the star.
I don’t know… there are mini games every turn and you get coins even for losing plus you often get coins for landing on spaces. I feel like you have to try to not have coins by the time you get to a star
Since stars are so cheap in this game, and you get some coins even if you don't get first in a minigame, I figured that the odds of you not being able to afford a star are so low that it wasn't worth all the complexity of coins. Like I mentioned in the video, if you're at the point where you're trying to min-max your die, you can almost certainly get enough coins that the occasional -3 won't actually matter all that much. Based on my experience, players usually end the game with a bunch of coins that they weren't able to spend, so it would really only come into play in a tie situation. Including coins may have cost Bowser a small handful of games, but in the grand scheme of things I don't think it would have affected the final rankings at all
I suppose so, and won't ask you to make a whole other video, but I'm curious how often the issue would come up
Maybe I'll make my own simulation!
If you don't have enough coins in the game where Stars cost 10 coins, then it doesn't matter which character you pick: you're just bad at the game
@@TheChiptidein the official tier list made by the competitive community of this game Shy Guy is considered S tier while Bowser is considered really bad if I remember correctly
6:12 I am steeeeeeeeeve!!!!!!!!!!🗣🗣🗣🗣🗣🗣🗣🗣🗣🗣🗣
At first I thought my comments were auto deleted because I included a link, but very LOL at my comments being deleted just because I'm trying to offer more help and data on this project. I actively want more people to come together on solving this interesting problem as thoroughly as possible.
Reading the comments it seems as if you should basically code in a full game of mario party and simulate it to really determine who wins
You also forgot that rolling a 0 triggers the space again
But if you land on an Ally Space, you can’t roll 0 anymore, since the Ally will always add 1 or 2 to your total.
Being able to trigger the same space over and over again can potentially be useful in an actual Super Mario Party game, but would not be a factor for this simplified simulation.
Technically the AI does account for that, except once you get an ally you can never move 0 anymore, so it never really comes into play here
It does strike me that either playing as Daisy, and having Bowser as an ally, or playing as Bowser and having Daisy as an ally, would probably be an optimal combination. Bowser's die to cover long distance and having Daisy's die for a high minimum roll when close to a star would be the optimal combination.
Of course, you can't choose which ally you get, and this strategy is unworkable if one of the other players is Bowser/Daisy, so it's not exactly reliable. I do wonder if Daisy becomes more optimal if no other player is playing Bowser, Wario, or Boo, since her die poses her well to get an ally, and snagging one of them early is hugely beneficial to her. Obviously it's easier to just pick Bowser, but if you know that no one else is playing as him (or the other distance characters) it might be worth trying.
You could have just played one real game with 4 cpus and put the results into a random number generator for the other 2999 times.
I liked this vid before it even begun because that's part of what I was thinking about the first one.
Dasiy and shy guys are the S tier allies
When you roll dice, you are able to pick where the die stops if you have skill.
I personally like Boo's dice the best. It is in the top 3 best ones and has some pretty high numbers while still being decently consistent 2 fives and 2 sevens. And seeing as I am better at the minigames against most of my friends the -2 coins faces don't really matter to me.
Another fact about Bowser's dominance is that his 0's are actually beneficial if you consider that the space they are on is a positive space, for example, a lucky space, which if they landed the turn before they get a chance to get that same lucky space twice. And given the fact that most board spaces are blue, they give 3 coins when landing on them, which nullifies the -3 of Bowser's 0, which the idea of "but you lose coins" becomes "you do not gain coins", which is a huge difference if you compare them. Given what we see in the video, Bowser does not gain allies that frequently, which in a normal game is actually way more beneficial for his 0 values.
Do a simulation with Luigi to see if he is actually just the best by himself or if he is only the best ally.
Love to see consistency accounted for this time. Was it accounted for perfectly? No. Am I satisfied with how it was accounted? Kinda. I also don't really care because nothing can convince me to not choose Shy Guy. I think a way to make allies not a liability for characters with consistent dice is to have an option that makes it so characters don't roll the extra di(c)e. Or just remove that and stick to the extra dice option. Either way, I do hope we get a third Mario Party with character dice on the Switch 2. Speaking of a third Mario Party with character dice, I personally believe the first Mario Party with Character dice (Star Rush) did it better, where allies spawn on spaces and you play as different coloured toads that have their character dice of the standard dice. I also believe these two games are the only ones with a grid system. Super Mario Party had allies pop up on spaces in the grid system, but that's in the Partner Party side mode. I do think Partner Party is better, for a multitude of reasons. You (kinda not really) can have a character's dice without them adding one or two to your role. Also, consistency is more important and also less useful for dies (that does in fact make sense in this context), which is really neat.
Conclusion: I really like Mario Party games with Character dice and grid systems :D
This sort of makes me want to make a more accurate simulation. Sounds fun
Out of curiosity, what language do you think you'd use to write it in?
The most accurate you could get in simulations is by actually simulating the game itself. There are 20 characters in the game, which makes the number of 4-player character combinations 4845. To test each of these combinations for 10, 15, and 20 turn games 1000 times each (total 3000 per party combination), you would end up with 14,535,000 simulations to run. Even if you sped up each simulated game to take just one second, it would take just over 168 days to finish computations. This would be the most rigorous method.
@@Neonlaserz I have code that runs each simulation in roughly a millisecond that I would love to work with others to get more eyes on or help this video series, but my comments about the code keep being deleted.
Here's hoping for some justice for my boy hammer bros
he's not even in this video 😭
Because he sucks
@@TheChiptide at least he's better than yoshi waluigi and monty mole
Justice may have been served...it just isn't in your favor hehe.
Okay, now to wait which Jamboree Buddy from Super Mario Party Jamboree is objectively better at winning, and which maps are the best to have them because Peach is pretty powerful in Rainbow Galleria.
As someone who does like to go back and play their favorite Mario Parties, this is one of my faves, and I will be using the information from these vids to my advantage. 👌🏻
But now we need to know what’s the most optimal ally team.
If I had to guess, it would be whatever 4 have the least overlap in their dice, or one where each character good at landing on a specific number (Shy Guy for 4’s, Wario for 6’s, etc) so you can always try to target specific spaces. Interesting idea, though, I might have to look into that one more!
Now we need to see your tier list for doubles... Preferably with a more accurate simulation if possible.
7:07 this does, unfortunately, give Bowser a little buff given he has a ⅓ chance of losing 3 coins
I say its Rosalina, because... Have you seen her ride a bike? 😂
Coins are nothing when stars are only 10
Extremely odd to only put one ally space on the board. Unlike most other ways the simulation can be improved, putting an accurate amount of ally spaces seems really easy. And it would definitely benefit Daisy and Shy Guy.
You’re totally right! I ended up updating the simulation to add more ally spaces that I detailed in the pinned comment, but tldr it didn’t really help Daisy and Shy Guy all that much. It did make it significantly easier for them to get allies, but it also made it a bit easier for Bowser to randomly get allies and undo all their hard work. While allies are powerful, they’re not enough to overpower the might of bowser!
@@TheChiptideyeah, it's pretty predictable that adding more ally spaces helps bowser more. Each ally you have, makes your rolls more inconsistent, so while they have an easier time getting their first ally, it becomes harder to aim for the next ally space. As a result, it's very expected that Bowser is the one that benefits from more ally spaces. The more ally spaces there are, the less consistency is needed to land on one.
You're still assuming very limited strategy for the consistency, only considering the special dice when a special thing is in range isn't enough, consistency is most valuable when you can plan ahead.
They need to not just consider whether it will be more likely to get them to a special spot, but also which die is more likely to put them in an ideal spot to get to the special spot the next turn, or the turn after, etc. Of course, the further away it is, the less it will matter and the more you just want to get distance.
Getting more coins easily will also make it easier to vuy something from the shop without screwing yourself on a future star, whereas losing coins does the opposite, so I'd argue that a character like Bowser is less likely to get stuff from the shop.
And you didn't even factor in chance time spaces where us Daisy mains get to steal all the hard work from bowser. Smh.
19:06 Now these videos make sense
You see all of the code that he showed off in the video is basic computer science and only someone that doesn't know or recently learnt computer science would do a video about this and even if this video and code done in this video is basic it is still a very interesting and fun video 10/10
Law of Averages means Bowser IS consistent if you play enough games, right?
I feel a lot of the commentators object mostly due to feelings rather than facts (or at least a statistical implementation of one) -- lots of people who object here seems to really prioritize the flight over the fight response so I can see why that risky behavior even if it works in the long term even statistics wise isn't seen as effective IMO (similar PoV when someone says Focus Blast is "really 50:50" IMO).
I don't see the point in the bowser comparison if you neglect coins. Straight buffing a character is going to make it look better than it should
It’s Bowser Time! Let’s go, Koopa King!
Like I said in the first video, I never said you were wrong. I only ever said there was more to the game.
Now with this video, I will say, even if Bowser's die is the best, I still will never use it, because I have such bad luck that I would probably only ever roll those "-2"s (my luck isn't why I was pushing Daisy so much though).
Simp
@@Inksploded If you're gonna call me a simp, just because a (fictional) woman was merely mentioned, I'm going to demand an explanation.
Pom pom's dice is still slightly better than daisy's imo
to put an end to those saying the -3 coins on bowsers die matters, you can actually just time the dice rolls in this game and just never have that matter
everyones die is consistent
Yoooo so I watched the other video earlier and finished it just only and just found this!! So excited
could you run the simulation again, but giving every character the option to roll Steve dice?
i think you should really add coins, red spaces, blue spaces, lucky spaces, and unlucky spaces.
make it so every four turns the players all recieve 15 coins (equivalent to getting 1st, 2nd, 3rd, and 4th in minigames)
Make it so stars cost 10 coins.
I think ignoring coins and only having a single advantageous space put bowser at a large advantage.
And i dont think that would be hard to simulate at all. adding coins is super super easy, red spaces and blue spaces are easy, and lucky/unlucky spaces just require adding another randomizer for when u land on it. Should be simple to implement. Makes me want to do it lol
The problem with adding in coins is that they wouldn’t really matter unless I also add in items for people to spend their coins on (otherwise they would be mostly negligible, as red spaces alone won’t be enough to prevent you from buying a star, even if you landed on them basically every turn). That makes things way more complicated, because you have to also program in ai for when each character uses each item. If I do all that, I’m not really simulating the game anymore, I’m basically rebuilding it from scratch! Kind of a lot of work for one guy in a couple of days…
@TheChiptide yeah I agree! The items are definetely well to hard for the scope and purpose of this video. I do still think adding more different spaces would help mitigate the advantage bowser has and make it less of a "whoever roles more wins", but yeah
1:13
Did TcNick3 saw the previous video?
The analysis still felt lackluster unfortunately :/
I don't really think there is a feasible way to do this where it doesn't feel lackluster tho.
Like, in a real game, every single space does something, while this board only has 1 space you probably want to land on, and this 1 space is exactly the space that destroys your consistency, because it randomly adds either a 1 or a 2 to your throw. The simulation's purpose was to see whether consistent throws can give you an edge over bigger movements, but the way you choose to do this, was by making the only possible reward for consistent throws, being able to do bigger movements at the expense of consistency.
Ignoring coins also ends up ignoring Bowser's dice's main downside, which doesn't feel all that fair. For the purposes of this simulation, it does make sense, because as you accurately stated, you earn enough coins to reliably always have enough for the star, but things like golden pipes in the shops, are pretty much equivalent to a star, and once you factor in those, coins do start mattering. Coins are also used as a tie breaker. So in games where bowser ends up with the same amount of stars as another player, bowser is more likely to loose due to having less coins.
it would be so cool if we got a statistics in marty party trilogy.5 before gta6 (0.5 comes from the hammer bro apology)
Let's ignore coins, sure Bowser is losing 30% of the value of a star in coins every three turns, while being able to reach any star in 2 to 4 rolls, but whi cares about such thing, right?
16:16 is my math not mathing?
Yeah no those don't add up
EVERYONE KNOWS ITS STEVE
16:20 the graph is missing 5
I'm rooting for Steve
Ok but what's the best pair of characters in team mode?
fuck yeah allies
While I agree that coins are not an issue, I think that there's a couple things to take into account regardless of that
1. Bowser spaces: These are likely to take away a star or remove econ, especially with communism. However, not only is Bowser probably likely to avoid the space unless it's right in front of him (in which case he should probably go for standard), because bowser has the -3 coin penalty and is picking up stars/items more often, communism will likely benefit him the most.
Oh, wait. This is super mario party. Communism doesn't even exist for this, and the rest of the options are a lot more tame. dang
2. Bonus stars. Rich, sightseer, ally, event, there's a lot of factors that can matter with bonus stars. Minigame star is separate from boardplay, money is usually determined from who wins the most minigames, same with item, and the sightseer/slowpoke ones have already been determined, so the only ones that are likely to change things are happening ally/buddy, and misfortune. What you could do is put like 3 event spaces scattered across the map that ONLY contribute to happening star, (because misfortune is essentially the same thing) and cycle through 3 of a random list of sightseer, slowpoke, happening, rich, item, minigame, and ally at the end of a game, with the rich, item, and minigame stars randomly being assigned to single player, because they all have pretty much equal liklihood to gain those stars. This would then determine how impactful landing on specific spaces for allies and happening is compared to just going for speed and stars.
3. You love it, you hate it, you've all experienced turn 20 star swap into last place, it's CHANCE TIME! Absolutely useful for players in last place, maybe have the ai prioritize chance time when they're not in first, with a higher chance the closer to last they are (by stars, not just by placement) and also prioritize it the...
wait. No, you're kidding me. SUPER MARIO PARTY DOESN'T HAVE CHANCE TIME???
WHAT?
I guess that's less coding.
I agreed with you last time and still agree with your conclusion. The fans of this specific Mario party are trying to pretend like it’s more complicated than it really is. Super Mario party is one of the most broken Mario parties ever and moving more is all that matters. You didn’t even include items which would have made bowser’s win rate even higher.
If you move more, you can farm golden pipes or boo star steals more. The thing consistent dice can do is help you get an ally so that you can … move more.
@@Georgeavocados I agree that the mandatory bonus die allies give after getting them removes any consistency the Daisy and Shy Guy dice gives and devolves again to purely a movement only strategy for dice blocks...I don't think any board in Super benefits much on limited movement unlike say the boards in Superstars and Jamboree so long movement is almost always welcome strategy.
But what about the bonus stars?
Let's goooo!
people really need to accept that bowser is superior, not only cause of statistics but also he is the supreme daddy so idk why everyeone is surprised xd
Is that a Worcester Polytech shirt you're wearing?
🐐
@TheChiptide It's one of the colleges I'm considering. What do you think of it?
@MeeBacon I personally had a great time! If you’re planning on doing anything STEM, I’d say it’s worth applying
You sound like Silver Spoon from Inanimate Insanity Invitational
Nice college.
🐐
you speech patterns remind me a lot of mumkey jones
Bowser #1 😎👍
The fact that Bowser won the stripped down simulation is not a surprise at all. It is to be expected, since this simulation ignores the counter balance. The fact that Daisy and Shy guy have nearly the same win rate proves that there is a lot of merit to controlling rolls, a merit that only improves as you introduce the game's complexity, whereas Bowser's merits disappear as you add complexity. I suspect that you are a secret Hammer Bro enjoyer and are only making these videos to bias the public for greedy, less optimal Bowser picks. You understand the game is far too complex for movement as a victory condition.
Soup Mario Party!!!!!!!!
Do it with mario party star rush
This whole video is super entertaining and extremely interesting, AND super funny! Sorry there's so many PICKY PANTS gamers in the comments.