RabiDrones simulated by a Markov Text Generator

General Discussion
41

Just had to do a double take to make sure we were looking at the same chart.

This has been pointed out in the thread already, but here we go. You have a basic understanding of ELO right? if someone wins there ELO goes up? Right? Still with me? If Zerg is suddenly buffed and players start getting boosted up, their elo will increase as they start winning more. This graph and your analysis assumes that they will still have a lower ELO than the people they are now evenly matched against. ELO updates to reflect this change.

So what about the top end? Surely, the best of the best will now win so much more that they will drag the average ELO up, right? Well, no. ELO graphs are asymptotes. They give the top players increasingly stringent point increases for repeated wins. This prevents the top players from dragging up the average. Why? Because ELO was NEVER DESIGNED FOR ASYMMETRICAL GAMES They were designed to compare one player to another player in a mirror match! This has been talked about in the thread you link, and mentioned ad nauseam, but not responded to with any defensible argument.

But not to fear! Some people did design a system to overcome this problem! Even better, you get all you stats from them! Behold:

http: //aligulac .com/misc/balance/

Dota, LOL, SC2, OW, virtually any competitive asymmetrical game uses some form of adjusted-for-asymmetry algorithm. You chuck them all out the window, because they adjust for race and therefore don’t provide the predetermined results you are so thirsty for.

Pick any one of them, so many are publicly available on the stack exchange, and use them to find results. Really, we know you won’t! At no point will you acknowledge that any of these very successful game designer know how to balance better than you, and defer to their tried-and-true techniques.

It doesn’t matter though. Blizz looked at their data and determined Zerg gets nerfed next patch. You seem to insist Blizz is wrong here, which is fine, but then provide evidence for your claim.

So I say once again. Provide proof for your claims! Put up or give up.

Edit: Magical internet fairies told me Dota abandoned their adjusted algorithm and now goes exclusively by win rate. Really cool to learn that players figured out the formula early on and abused it to boost their mmr. I imagine LoL may have also had a similar problem. It doesn’t matter because Batz won’t accept exclusive win rate either as it still makes Zerg look too strong.

1 Like
42

Correct. In fact that’s the exact crux of my entire point in that thread. There is no measurable skew in ZvX and XvZ elos thus we eliminate the possibility that you describe. If Zergs were winning more, there would be a skew. There would be a large cluster of zergs with great ZvT and ZvP ratings just above a big cluster of Terran and Protoss players that had bad TvZ and PvZ ratings. Yet this effect doesn’t exist. In fact the opposite happens.

Thanks for yet another glaring instance of how poorly you understand my argument. You literally just tried to use my exact point against me to falsify a prediction that my argument doesn’t make.

Elo is a logistic regression model and it applies to any data that is logistically distributed which skill is. Time to stop spouting nonsense.

If you bothered to read Aligulac’s FAQ, he destroys your point. His balance chart measures changes in balance, it does not measure where balance is. When his chart shows Zerg going up, all it means is that Zerg performance is going up which is fine if Zerg is underpowered. Also, his model doesn’t differ in any way from mine with regards to accounting for game asymmetry. You’re simply lying through your teeth on that point.

Your inability to understand a point is not a rebuttal to that point. I would think that would be common sense but here we are. Zerg factually has 26% lower ELO on average and this effect is not dependent on skill level. Those are the facts.

Given these facts, any argument that says Zerg is advantaged is wrong. Any argument that says Zerg’s performance is tied to skill level is wrong. Any argument that says zerg’s number of players is tied to Zerg’s performance is wrong. This is true on average for all players in Aligulac. This is what the facts say.

You claimed that Zerg is over-represented at the pro level because of balance. That theory is factually wrong.

1 Like
43

Wait, so you are saying that to give proof you would have to lie, as in the only available proof would be made up?

2 Likes
44

Imaginary arguments are trash and go right into the trash can.

45

Thank you, but don’t you think the dude who refers to himself as a greek god or refers to himself in third person as an different individual agreeing with yourself shouldn’t be the one to talk about imaginary?

Correct. In fact that’s the exact crux of my entire point in that thread

Ok. I respect you here, and should have given you the benefit of the doubt. I made a few assumptions about your argument because you have no legend or breakdown. The chart is also fairly messy. I also made assumptions that it was poorly done because this is clearly a correlational study over two points in time (at least), and you use it as evidence for causation.

Elo is a logistic regression model and it applies to any data that is logistically distributed which skill is.

You are well versed in statistics enough to know that ELO tops off, but I will explain it as not everyone does. Serral could win 5 games in a row against other players but only get 1-3 MMR (and the loser only loses 1-3 MMR) each win because of the skill difference. This is because the mmr system can’t find anyone of his level and has to pair him against weaker players who only give him sparse mmr. This skew is one of the main reasons that ELO isn’t commonly used as statistical data but rather for comparing individuals, which is what it was exclusively designed to do. On top of that you take data from Aligulac which has different active players from one period to another. Correct me if i’m wrong, but you don’t use the same sample size, which is a big no-no in stats because it taints the reliability of your results. Sure, attrition forces people to drop out over time in longitudinal studies, but here Zerg getting buffed then winning more and more which pushes other non zerg players out of the Aligulac sample pool is a confounding variable and not attrition. As new Zerg players push out other races to make the bottom threshold to get on Aligulac, they greatly skew the numbers. This means calculating averages will always make Zerg appear at best average or at worst weak.

If you want to see which race is really leading and which is lagging by raw numbers go Aligulac > Ranking > History

If you bothered to read Aligulac’s FAQ, he destroys your point. His balance chart measures changes in balance, it does not measure where balance is

The chart is described as supplementary data for the other graph, both of which show how powerful Zerg is. Without confusing correlation with causation, I would like to draw attention to how the races always hover around the “0” baseline in the center. Blizz seems to really want races to not deviate far from the baseline.

Provide proof for your claims!
providing evidence for Zerg having a drag

Your inability to understand a point is not a rebuttal to that point

I don’t care about a random point if it is unrelated to your original claim.

You claimed that Zerg is over-represented at the pro level because of balance. That theory is factually wrong.

You are the one claiming regression to the mean shouldn’t apply to all three races here. Because you make this claim going against conventional wisdom, the onus is on you to prove your claim. You admit that ~43% are Zerg but Zerg can’t be overrepresented because you gave me a totally unrelated graph of means?

If you can’t provide relevant data but instead your entire argument boils down to “Because I said so, and I have data that doesn’t back it up!” then why bother debating you?

1 Like
46

The problem with you saying this is that you already did it.

You claimed that your argument was too advanced for others to understand, and that was your best defence you offered. My imaginary argument was made only in mock imitation of yours.

1 Like
47

Don’t you see? Batz’s argument is so advanced even he doesn’t understand it. But take his word, its obviously correct, in spite of the fact that no explanation for the various flaws is forthcoming, nor any attempt to actually demonstrate any flaws in the opposing arguments.

3 Likes
48

Wrong. I thoroughly obliterated their rebuttals and when they failed to comprehend basic math I then concluded basic math is indeed too hard to understand. When someone repeatedly states that 2+2=7 that is the only conclusion that can be drawn.