Final character has been decided

Classic Fighting Games Skullgirls
21

This is exactly what Ravidrath said on Neogaf forums

In Vulpes Model the winner beats the runner up by 5%-8% more of the vote. That’s not a huge lead. Definitely not something you can be confident saying that you know who’s going to win with 40% of the vote still undecided. Also in both scenarios the vote would have to be split evenly between the 2nd and 3rd place character, and that’s surely not going to happen.

Now we need to talk about top 8. In order to get into top 8 each character would need to have averaged about 12.5% of the total vote. To get top 4 the average total vote would have to be around 25% per character. If 2nd-5th was nearly tied then that would mean top 5 percentage would have looked around 25%, 14%, 14%, 14%, 13%. That would mean that Isaac, Stanley, and Panzerfaust would have gotten less than 20% the total vote combined (6.67% each), and that’s impossible. You would have to say crazy stuff like Minette and Aeon are twice as popular as Isaac and Panzerfaust.

22

Keep in mind that not 100% of eligible voters will bother to vote.

It was said that there are 11,000 active voters, out of 15,000+ people who donated and are eligible.
Technically, 9000 votes is 60% of 15,000 – leaving about 2000 probable votes to make up whatever difference there is between the leader and current runner up.

23

Yes, but what I’m talking about is voter popularity trends relative to characters themselves, and not amount of people that voted. If those 4,000 out of 15,000 people never voted at all, they would have had no effect on the voting trends between the 4 rounds. You need to look at your percentages in relationship to the global voter chart. If 27% of eligible voters never voted at all, then that makes it even more impossible for any one character to reach 40% of the total votes. Ravadrith would have to be referring to 60% of the 11,000 voters.

24

He said “60% of all votes are in”. You can draw whatever conclusions you want from that.

Also, I don’t think it’s crazy at all to say Minette or Aeon are twice as popular as Isaac or Panzer.
I wouldn’t be surprised if Minette was 3x as popular as both of them :wink:

25

I order for Minette to be 3x as popular as Isaac or Panzer, she would have to be 2x more popular then Annie, or Eliza, and your rationality doesn’t explain Aeon, a character Mike Z repeatedly told people NOT to vote for, receiving twice as many votes as Panzer and Isaac.

I can say with confidence that Minette is as least hated as much as she is liked, making her definitely less popular then Eliza who I’ve heard nobody complain about at all.

26

Did I miss some statistics or something? How do you know how popular Isaac or Panzer are?
We know beowolf was almost at the same level as Aeon (was it?) in the range of 9 votes, but I never saw anything similar for these other guys.

re-edit:
Oh geez, you guys are working on some theoretical model that I skimmed through…
Disregard my previous edit I guess.

27

Correct me if I’m wrong here on the subject of Minette vs Isaac/panzer popularity though. With no voting statistics other than Beowulf’s proximity to Aeon, round 3 could have gone like this:

Character | % of total votes cast

Minette 29.7%
Eliza 20%
Annie 10.2%
Aeon 10.1%
Beowulf 10%
Stanley 10%
Panzerfaust 5%
Isaac 5%

28

Someone said that in the top 4, the 1st place character was way ahead, and places 2nd-5th were relatively close, so I drew up a quick model of what that would look like
1st place 25%, 2nd 14%, 3rd 14%, (Aeon) 14%,
(Beowolf) 13%, (Stanley, Panzerfaust, Isaac) 6.67% each

That’s what the model would have to look like if this were true. Keeping in mind that to even get into top 16, you need a minimum of 6.25% popularity. Even in this model the difference between 1st place and 2nd place is 9% not a huge lead. But if this was the case, that would mean certain characters were twice as popular as the male characters, and the 1st place character was 4x as popular the male characters.

Taking a look at Vulpes 1000 vote model again: Chars ABCD, votes split A:250(41%) // B:150(25%) // C:100(17%) // D:100(17%) =600
with 400 votes still undecided.
I couldn’t really figure out Vulpes math after that, but it sounded like the point was being made that in order for character B to overtake A that the trends would have to shift drastically in B’s favor. And he’s saying that this is unlikely. I’m arguing the opposite.

The most votes that any of the top 4 characters can be guaranteed is up to 250 votes or 25% of the votes. After that, in order to continue winning you would need to take potential votes from other candidates. It’s very likely that Aeon will not get more than 10% of the votes because Mike Z specifically told people not to vote for her. Even if she did go above she wouldn’t be able to go above 14% popularity. So that leaves a good 400-360 votes up for grabs by the characters A, B, and C.

If you look at that voter by country chart I linked earlier you will see that the US makes up 65% of the voters and all other countries make up the other 35%. It’s safe to assume that all these other countries will not vote for the same way the US does. This makes is harder, not easier for a character that’s already gotten 250 votes (Character A) to get even more. Character A’s voting trend will drop drastically as she reaches about 400 or 40% of the votes which I feel is the max any one character will be able to obtain.

While it’s getting harder for Character A to get past 250 votes, it’s going to be easier for character B to reach her 250 votes. Strategic voting is going to be a factor here. People that love Character C but know that their character will lose to Character A, will most likely give their votes to Character B instead, making it even easier for Character B to catch up to Character A in voting.

In conclusion, assuming Characters A and Character B are within 6% of each other as they are on Vulpes voting model. Character C has up to 150 votes they can give to either A or B. If you look at all the factors there’s no way Rav would be able to predict with confidence who the winner is with only 60% of the votes. It’d be foolish to even assume that 1st place person would be able to keep up their momentum as they get closer to achieving 40% of the total vote.

29

I see see what you’re saying, but I just don’t agree.

I don’t think anybody is going to give away their vote under the assumption that their favorite ( who made it into top 4) is going to lose to another character. I don’t think we can make any assumptions about how the rest of the world will vote for either.

However, if you are looking at past voting results, and you see that after 60% of the votes are in, the character that has been way ahead in previous rounds is once again way ahead, I think you can be confident in your prediction. It doesn’t really matter where the votes come from.

30

Example Case1 for A not winning:

  • Of the remaining 400 votes, A gets 40(10%), B 150(~38%), C 105(~26%), D 105(~26%) - For a total of A:290, B:300, C:205, D:205 votes.
    The most popular char suddenly drops from 40% to 10% popularity (!!! That’s less than 2/3rds of the chars with the lowest vote turnout from the first batch)
    The 2nd most popular char rises from 25% to 38% (a 50% popularity boost)
    The 3rd+4th char get bumped from 17 to 26 (again, two 50% popularity boosts)
  • So for this example, you have one char getting a 75% voting drop and three 50%+ voting boosts. That is not likely.

Example Case2 for A not winning (this time with less distance between the votes; 200-150-150-100 (33%-25%-25%-17%)):

  • Of the remaining 400 votes, A gets 60(15%), B 120(30%), C 100(25%), D 120(30%) - For a total of A:260, B:270, C:250, D:220 votes.
    The most popular char drops from >30% to less than half of that (again, less % than the least popular char had prior to this)
    Char B gets a moderate increase that doesn’t sound too unlikely!
    Char C stays the same
    Char D almost doubles it’s percentage
  • So for this example, you got the most popular char dropping by 50% (below the prior least popular char), and the least popular char dancing on a 3/4 increase in popularity (almost as high as the prior most popular char was before). That is… not likely!

Note that these are very MODERATE examples. The further you go with the percentages, the worse it gets. Imagine we have two very unpopular characters, one high, and one sorta close
Say, Example Case 3: 600 Votes split A:294(49%), B:240(40%), C:60(10%) D:6(1%)

  • This time I’ll do it step by step, to have it easier to follow.
    There are 400 votes left to distribute.
    What percentage for A in this batch is likely? Surely a 49% popularity character won’t suddenly drop down to 1%?
    Let’s assume he only gets 15% of votes of the last 400 (This is a >66% decrease and won’t ever happen, but I’m being nice).
  • That means: 15% of 400 distributed, 60 Votes on top of the 294 that A had from the first batch, to make 354 total.
    Now you have to look at the Char B. Currently it has 240 votes, so to beat 354, it needs at least another 155.
    What percentage would 155 votes be? ~39%. Now that sounds easy! …We can even just say that in the 2nd batch he also gets 40% of votes, to make it simpler.
  • That means, 55% of 400 distributed. 160 Votes for char B, he now sits at 400, he has taken over! No problem.
    Wait, yes problem. 55% distributed means there are 45% votes up for grabs. Reminder: CharC+D got 11% of votes in the last voting round.
    So to make that >66% drop of CharA happen, CharC+D get a >300% increase. That sounds… unlikely.

Okay, let’s change it up! All gears back to zero.

  • The 66% drop of CharA was maybe too much of a thing? Let’s reduce it! Say, in the 2nd round A get’s 40% of votes. That sounds like it’s within reasonable boundaries (a 10% drop sounds possible).
    With 400 votes left to distribute, 40% votes for A means 160 votes.
  • That means: 40% of 400 distributed, 160 votes for CharA, +294 which he already had = 454 Votes.
    Now, how many votes does Char B need to overtake? A minimum of 215. What’s that in %ages? 54%!
  • That means: 94% of 400 distributed, 216 votes for CharB, +240 which he already had = 456 Votes.
    No problem? Well, you already notice that this is pushing the boundaries. There are only 6% votes left to distribute, which means a 50% voting drop for C/D which already had pretty much no votes. This is, again,** not likely**.

tbc/etc/whatever.

I don’t get any of your points. I think we’re both talking to walls?
- Rav said "It’s statistically unlikely that we’re gonna see someone win who isn’t A"
I just showed that even with a tiny tiny lead (say, 8%) this is the case (even if the %age shifts slightly into B’s favour, since there are less votes to distribute (400 vs 600), A will come out on top.
Now imagine that the lead is BIGGER than 8% (eg 20%, with B/C/D being very close to each other - which actually sounds very likely, given the statements of “There is one char with a clear lead” and “Places 2-8 were all within 50 votes of each other”), and you quickly notice that this is not hard of a call to make at all.

:spades::spades::spades:

Last case, A/B/C/D - 40/20/20/20
A has 240 Votes
B/C/D 120 each

A drops to 30% in the 2nd batch, 120 more Votes for 360 total
For any of B/C/D to catch him, they need >240 votes, which is >60%
If B gets, random number 65%
That’s 5% votes left to distribute between C+D which added up to 40% before
Not happening.

31

To put it (far) simpler:

There are two candidates, A and B. You have a batch of 10.000.000 randomly shuffled votes.
Now you cut the batch at a random point where both pieces still have a respectable size (Take eg 100.000+ for the smaller stack - no 1 to 9.999.999 splitting).
Now you count the votes in one batch and get some result, A=63% votes, B=37% votes
What %age do you now expect to be in the second batch?
It’s POSSIBLE that all of the votes in the second stack are votes for B. It’s possible that all of them are gonna be for A. Most likely you’re gonna get 63-37.

Same principle applies here, with the added bonus that we have already counted the bigger stack, so even if the %age shifts, it most likely won’t matter.

32

That much is obvious. Equally as obvious as people’s seemingly nonexistent sense of sarcasm.

But in all seriousness…I’m super salty that a character like Minette got picked over Panzer. At least Black Dahlia made it. I would have flipped my cookies if both Panzer AND her didn’t make it.

33

As a Cammy player, you should be voting for Aeon :~

34

Is that sarcasm too?

35

You have to look at it from a Geographical perspective. 65% of all the voters are in the US and 35% are non US countries. It’s perfectly reasonable that the non US countries would vote for different majority of the characters then the US does. Let’s be nice and say that UK, Canada, and Australia all vote the same way that US does that’s 80.1% voting the same and 19.9% voting differently.

In my model, I’m calculating the chances of the 1st place character winning where she has a lead over both 2nd and 3rd place, but not a majority vote. This means that if you add 2nd and 3rd place characters together they would have more votes to defeat 1st place. This is realistic. Assuming that the 4th place character got the minimum of 10%-12.5% of the total vote to even be considered top 4, that only leaves 90% of the vote up for grabs. That means that the 1st place character would max at 40-45% of the vote while 2nd and 3rd would average 22.5%-25% of the remaining total votes each.

Now someone asks Rav if he knows who’s going to win and he replies that he does. So right there Example 3 Scenario
Example Case 3: "600 Votes split A:294(49%), B:240(40%), C:60(10%) D:6(1%)"
Would be flushed down the toilet. If the 1st place character was close to the 2nd place character, he would not say he was sure unless the 1st place character had a decisive lead over the 2nd place character.

Example Case 4 “A/B/C/D - 40/20/20/20” could not happen either. I had already whipped up a model of a possible top8 --> top 4 conversion that looked like this:

1st place 25%, 2nd 14%, 3rd 14%, (Aeon) 14%,
(Beowolf) 13%, (Stanley, Panzerfaust, Isaac) 6.67% each

Mike Z is specifically saying “Don’t vote for Aeon!” In order for this example to be true, not only would more than 14% of the total votes would vote for Aeon regardless, she would need another 30~% of the total vote of the males that lost. It’s most likely that Aeons vote would decrease this round and not increase.

Example 1 & 2 are the closest to explaining Rav’s logic.
Ex1: A:250(41%) // B:150(25%) // C:100(17%) // D:100(17%) =600 with 40% (400 votes) undecided
Ex2: A:200(33%) //B:150 (25%) //C:150 (25%) //D:100 (17%) =600 with 40% (400 votes) undecided

In both your examples the first place 1st place character’s trend would have to drastically decrease and the 2nd place character’s trend would have to increase in order for 1st place to be overtaken. And you’re saying that this is statically unlikely, but NOT impossible.

Okay so let’s go back to Geographic voters. Let’s assume that 80% of countries are going to vote the same way and 20% are going to vote differently.
The sweet spot for Rav to be confident that 1st place was going to win would have to be around 40% of the total votes.
Assuming 1000 votes, the first place character would need 400 votes. The last place character would get at least 100 votes meaning 500 votes would be up for grabs.

The max number of votes the english speaking countries have in a 1000 vote model is 800 votes. The max the non english speaking countries have is 200 votes. Assuming that they will vote differently, it would be very difficult for one character to pick up 400 votes total. Rav’s 1,000 vote model would have to break down similar to this:

1000 votes (english 800 + non english 200)
344 votes in the english countries ( 43%) + 56 votes (28%) in non english countries. = 400 votes (1st place character)
188 votes in the english countries (23.5% each) + 62 votes ( 31% each) in non english countries = 500 votes (2nd & 3rd place characters)
80 votes in english countries (10%) +20 votes (10% each) in non english countries = 100 votes (4th place character: Aeon)

That is what a decisive victory would look like with 100% of the votes in. Now let’s take a look with only 60% of the votes

600 votes (english 480 + non english 120)
206.4 votes in the english countries ( 43%) + 33.6 votes (28%) in non english countries. = 240 votes (1st place character)
112.8 votes in the english countries (23.5% each) + 37.2 votes ( 31% each) in non english countries = 300 votes ( 2nd & 3rd place)
48 votes in english countries (10%) + 12 votes (10% each) in non english countries = 60 votes (4th place character: Aeon)

So the current difference between 1st and 2nd place in a 1000 vote model would only be 93.6 votes. And there’s 400 votes (320+80) still undecided! Yes if the current trends continued on their course the 1st place character would win, but that’s the same as saying if the remaining 40% of the population voted exactly the same way as previous 60% vote, 1st place would win. That simply is not going to happen, the margin of error is really too big (93.6 of 400 votes) to say that he knows who’s going to win.

36

This explains your strange maths

You’re running baseless assumptions

37

Now if you look at my 1000 votes model it’s extremely conservative. It’s assuming that UK, Canada, and Australia would have similiar voting trends then the US.
The second thing my model assumes is that last 40% of the voters is going to vote the same way the first 60% did, which is a another leap of fate. We don’t even know if 100% of the people that voted last round would even bother to vote this round.
The third thing my model assumes is that strategic voting will not take place and C will not dump all it’s votes on B to beat A.
The fourth thing my model assumes is that the US is a bunch of idiots. If one character was running around with 43% of the US popular vote we would know about it. Do you know who it would be? I sure don’t.

Between Elliza, Annie, and Minnete. One of these two would have to be almost twice as popular in the US as the other two characters. And even then, the other two characters would have to twice as popular as Aeon. That would mean the 1st place character would have to be 4x as popular in the US as Aeon. It’s way too much stuff going on for anybody to be certain who’s going to win, even statistically speaking.

You guys keep talking about the trends, I keep talking about the factors. There’s simply no way that one character would have a big enough lead where 40% of the undecided votes would be unlikely to tip the scales. And such a character did exist, there would be no way we wouldn’t be able to figure it who it was by now.

38

You think that MikeZ stating anything matters for any relevant part of the voting %age
HINT: MikeZ stated not to vote for Aeon SINCE THE BEGINNING (it’s actually right up there on the 3rd char description page), and Aeon managed Top4.
For all we know, Aeon is the char running away with the votes.

Then you come with the “US votes different than other countries”-argument, which is completely baseless and even if it was true WOULD NOT MATTER FOR SHIT because votes aren’t coming in on a per-country basis (it’s not Monday-All US votes, Tuesday-All Japan votes, Wednesday-All Africa votes…).

Yeah, precisely what Ravid was stating.
Nothing is impossible until one character has won. If you count 499 votes and have 100% A, 0%B/C/D - then your next 501 votes can be 100% D, 0%A/B/C.
It’s just not all too likely to happen.

I can’t follow your model one bit, and quite frankly I don’t care to try, because you don’t seem to understand the first and foremost thing (Unlikely != Impossible) and base your “model” on a romp of useless baseless speculation instead of just doing maths.

39

I went through the trouble of breaking down each and one of your examples and now you’re mad. I’m sorry to hurt your feelings. But don’t waste our time claiming anybody’s math is baseless unless you have better math to back up your claims.

Once again I apologize that you can’t follow my math. Maybe it’s better that you move on to a different topic so the rest of us can continue the discussion. The fourth place character’s percentage was actually the same as your own. I made my model based off your own numbers. This was an extremely conservative estimate, meaning that if Aeon’s voting percentage was any higher, that would mean the gap between the 1st and 2nd place characters would be even closer, making the scenario of knowing who’s going to win even less likely.

40

He’s not saying your math is baseless, he’s saying the assumptions you made in setting up your math are baseless.

In finance we have a saying: “garbage in = garbage out.” It doesn’t matter how good or how much analysis you do if your underlying assumptions are unwarranted or you have bad data.