DUDE. he literally just quoted you LOL
1 Like
DUDE, he just made a huge edit. Try to keep up.
Iām up. and still waiting for you to properly justify why aligulac is bad data, for starters
1 Like
I asked a while ago which winrates heās even talking about. GM? Masters? All of ladder? Aligulac? I donāt see the winrates in this thread.
And for ladder distributions I said multiple times that unless races are randomly assigned thereās no basis for assuming that every league should be evenly represented. The fact that thereās 29% more Terran players than Zerg on ladder shows that race selection is an obvious bias.
And if you want to take aligulacās winrates then YOU need to provide an accurate margin of error. Thatās not my job. I gave examples of aligulacās data where players matched have hundreds and even thousands of mmr difference between them to show that thereās obviously a significant margin of error. But again, itās not my job to calculate that, itās yours.
And every period this margin of error changes because the data set changes. Aligulacās data is all over the place and nobody bothers looking at the details.
not necessarily.
Thereās a reason he uses Elo, which, is closely related to aligulacās glicko.
Youāre ignoring the whole point of using Elo/Glicko as well as the aligulac methodology described in the faqs. In fact you are probably ignoring the entire Ladder-MMR system as well as the barriers that the tournaments and counted on aligulac are present.
YOU are the one who needs to prove why itās not sufficient
Batz doesnt just analyze win rates, he takes skill into account as well.
You finding a sprinkling of 1000s of MMR difference games doesnt mean anything thanks to Elo/Glicko.
The only major flaw that I can think of in aligulac is that it overrates the upper echeolon of non-koreans compared to the upper echelon of koreans, thanks to the affirmative action being implemented. But off the top of my head i doubt this affects batzās analysis in any way since itās only relevant for a handful of players
1 Like
Rating gap cap. The idea here is to prevent players from āfarmingā much lower rated players. It is possible to artificially inflate a playerās rating if he never plays other players close to himself in skill. The ideas usually consist of ignoring matches where players are farther away in rating than a given threshold. However, this can be seen as unfair to the lower rated players, who will have their wins against good players discarded. Capping the gap to a given value will make the problem worse. Say a 1700-rated player plays a 1000-rated player, but the cap is 500, so for the purposes of updating the stronger playerās rating the lower rated player is assumed to be rated 1200. Then it will be easier for the 1700-rated player to overperform than it was previously.
here is the PhD in mathematicsā response to your contention
but all you would need to understand this is a basic understanding of Elo/Glicko
Can you tell us why you think protoss is OP?
Youāve said that the data suggests itās āstrongerā (not necessarily OP) but, I , at least, havent seen your data
Yepā¦just clueless about stats.
- This is just false, there are INFINITELY many possibilities on race selection that would yield a perfectly even representation. More specific assumptions are needed which you have failed to mention. But I can see why you did this though, to give room for your nonsense.
- Manā¦your logic statements are just wrong on so many levels it is not even funnyā¦you even need a course on basic high school logic of what is a premise , what is a conclusion, what constitutes a proof, and what it means to be an implication.
Also:
- There are: 31724 Zergs, 30191 Protoss, 32298 Terrans diamond league or above currently.
Umā¦he did that already? LOLā¦he literally wrote threads and threads of statistical analysis and justifications. I aināt gonna dig that up for you rofl. If you canāt even spend the effort to do that.
LOL!!! This is beyond facepalm. Please read up on ELO also, the Central Limit Theorem.
Elo isnāt factored into Aligulacās winrates. Those are just tallies of which race won the matches in their data set. To use the winrates you need to calculate a margin of error to account for things like skill differences, which Aligulac doesnāt do.
Thatās why ladder is a better measure of balance. After balance patches you can see shifts on ladder but you almost never see significant changes in aligulac during those periods.
Huh? The 29% is literally proof that that race selection is a bias. If race selection wasnāt a factor then each race would have the same number of players on ladder but that obviously isnāt the case. Some races are more popular than others and that affects the distributions.
Heās talking about a playerās rating on Aligulac, not winrates.
That was actually painful to read. Iām going to have to block you.
2 Likes
yep typical. Saw that coming a mile away.
Man I canāt believe I even apologized earlier. What the hell was I thinking. Never again will I apologize to someone on these forums again until the person can show that he is not a troll.
1 Like
Thatās just going off of ladder trends. In higher leagues Protoss has been increasing over the last few years and most significantly after the latest patch. I donāt know exactly how much because I donāt have all the numbers but the trends are pretty clear.
The uneven matches balance out (race vs race) if there are enough matches played. This is a serious case of being unable to understand stats.
oh my god you are just clueless. Please just stopā¦if you canāt phrase a logically coherent argument, just stopā¦you are embarrassing yourself at this point. Trying to bait people into your nonsense questions to troll them doesnāt work if the poster you are doing it to has any semblance of competence.
Batz isnt using mere tallies (Batzās analysis is what this is about)
Thankfully, Batz does this
reasonable, in my humble opinion
Mere increase =/= OP
If we start with the premise that protoss is underpowred a few years ago (this is just a hypothetical), an upward trend would not necessarily mean itās OP
Yeah, Iām saying in my opinion Protoss is OP, but the data shows Protoss has gotten stronger.
Yeah but I donāt see these numbers. Thatās why I asked for them a couple of times. The first post has a youtube link and this thread is insanely long so I donāt know where to see them.
YEP!
Donāt know why aligulac doesnāt filter plats and donāt take players of similar ELO. There is also measure based on evidence from previous month vs prediction of results. If Terran had last month ELO -100 and now +100, means there is +200 difference and Terrans were performing about (54.5%) better then expected! Problem i see with this, it takes as base only 1 previous month, which isnāt enough time to get unbiased results. E.g. after a major patch, they should take whole period before, until some significant patch in past! Supposing that is long enough time!
Maybe ignore patches of low significance to get some base numbers, supposing there is long enough time, to get unbiased results.
http://aligulac.com/misc/balance/Also there is very interesting correlation to cyclone nerf patch. You can notice from May 2017 Protoss winrates were total shi t, like 44% (in PvT) and they they skytocketed to over 53% in Nov 2017 (cyclone nerf patch). Reactored cyclones could threat protoss third and kill it, if protoss wasnāt careful he could lose greedy third, so he couldnāt expand greedily every game. As i said in PvT only problem ever was, was the economy, except like couple thingsā¦ Also from Nov 2017 PvT win rates are heavily in favor of Protoss! Protoss lead like 27:9 on win rate periods above 50% , while protoss ones go to like 55% even and Terran ones max to something like 51% maybe with like 1-2 exceptions.
There is similar correlation when cyclone patch was first introduced in Nov 2016. Protoss PvT win rate dropped to record 41% which was after cyclone patch! It wasnāt good couple months before (donāt know reason there) but this was new kind of low!
While at release of LOTV Protoss had 3 periods with over 50% win rate, then 2 Terran, then 4 tos. And after that PvT dropped for 3 months to very low 46-44, lowest then it ever was at that time. It is long ago, donāt remember exact reason for that, if there is obvious one! After cyclone patch instantly next month to 41% While 41% after this history still seems unlikely. It was probably caused by cyclone buff patch and some issues at that time for PRotoss, maybe maps?
Also you can see TvZ win rate going up and down, up and down. So the Zerg win rate, but it didnāt stay heavily in favor of one race 3 years. Unlikely PvT win rates since nov 2017 (cyclone nerf patch). Because what are the ods in like in PvT, that once race will 3 years have so much better win rates consistently, if it is not OP!
And since release of LOTV, PvZ was heavily Zerg favored, then it got better (probably adept buff) but PvZ was still zerg favored 48% win rates only for tos. Which also correlates to the time, when zerg was like 40% GM in 2016 and PvZ was heavily Zerg favored.
Always the cry and troll threads with you. Blah blah blah nerf everything that isnāt Zerg. Is there nothing else in your life?
1 Like
LMAO, Protoss dominating 3 years clearly on aligulac and 2 years consecutively has highest GM population. Instead of flaming someone, why not give arguments to prove your point of view. Clearly you are using logical fallacy to dismiss someoneās argument. Stop AD Hominem, just because he is Zerg, or even if he posted to cry about Zerg. It is completely irrelevant to the argument, whether or not tos is OP. You iq 30ā¦
http://aligulac.com/periods/
