Please release a preliminary balance batch

I can’t play this game in this state. Winrates vary from 10 to 90% depending on civ and division. What the hell is this? At least separate ranked match and quick matches cooldowns. With some civs its completely reasonable to dodge due to staggering skews in winrate; even if one plays their best it doesn’t matter. It can’t offset the winrate unless the match making totally messed up and accidently allocated you to someone with 500 less elo.

Not to mention the amount of players that appears that have played 20 games in their life and won 19 of them… If we are to believe their profile.

I’m not sure what hyperbole you are going off about, but the worst offending civs currently have a 56% winrate, which while high and should be adjusted is not hotfix levels of absurd. The only real exception is Byzantines, but even just over a few days, their winrate is noticably increasing as people learn to play the extremely difficult civilizaiton.

I refer you back to the 2nd sentence of my post which you seem to have only read halfway.

You said winrates vary from 10% to 90% depending on division. The only places where winrates get above around 56-58% are where the amount of games played is too small to be statistically relevant.

The game balance shouldn’t be based around stats where 1 person stubbing their toe causes the winrate to drop by 10%

Ehm, if you play X vs Y for some iterations, say about 100 games, and the winrate is 25/75 then that is statistically significant to draw a conclusion… And it didn’t take long to find a match up where that applies…

100 games is not a significant amount of games, especially for some civs. Even accounting across all ranks combined, most matchups do not have a statistically significant amount of games (there’s even a disclaimer on the AOE4 webpage saying low sample size posting this would be unwise). Besides certain matchups will always be swingy towards one civ or another which isn’t necesarily a balnce issue.

I have no clue how that it not significant “especially for some civs”. It makes zero sense to me. How would the significance substantially be different for one civ over the other? If we throw team A vs team B in a statistics program with 100 games it will return P > 0.001. Unless we assume that someone, despite motivation for the effort, went above and beyond to push specific match ups in opposite direction across various divisions but I don’t think that is likely.

Scientific journals have studies with a myriad of variables, including moderator variables, that use 100 samples. I honestly feel like that as soon as someone sees statistics they just Pavlov to “small sample size” and think that is an argument of substance.

Which civ has a winrate of 75%+ with at least 100 match, and which league are you?

Byzantines 26% vs Joan in conq (156 games). But don’t pin me on rounding. Just raising the point many civs have a very skewed winrate. That balances is to be desired after introducing is expected. But to this order of magnitude is quite insane. Some specific matchups are just not worth playing. Not if you want to win, and lets be honest: if you play very well and the enemy just wins due to current balancing it isn’t much fun either.

1 Like

100 is a very small sample size and just a small margin of error can have drastic differences. Heck if Beasty was playing jean the first week and ran into like 10 Byzantine players, he probably singlehandedly brought down that winrate stat by 10%. Sample sizes small like 100 are even more innefective at high elos since the top handful of conqueror players are signficantly better at the game than other conqueror players, even if only seperated by a couple ranks, to the point that if you are looking at sample sizes around 100 and Beasty or ML played 25 of those games, those civs, no matter who they are, will have drastically skewed winrates.

Byzantines in particular had an atrocious winrate stat that has been quickly climbing because the civ is so complicated, to the point that the games played in the first week vs the 2nd week has a huge difference in winrate.

You can even see this easily in winrate stats by looking at quick match. Comparing the winrate of 1200-1299 Rus to 1300-1399 Rus, their winrate drops by over 10%, but then if you look at the 1100-1199 games, the winrate increases by another 10%, but giong down to 1000-1099 games it drops again by 10% winrate, just from a small difference in elo despite samples in every elo being well over 100. This very small elo difference apparently turns a garbage civ into a top tier civ and back up and down again, which stat do you even base it off when determining how good they are?

There is so much in this post that doesn’t make sense I have no clue where to begin. Feel free to address the core of my post and argument instead.

Long story short 100-500 games is an extremely small sample size. We can see this by looking at some civ stats where winrates of the civs fluctuate up and down and back up again by more than 10% differences in winrate at a time just by marginal changes (like 100 points of elo).

Posting any stats with sample sizes not at the minimum in the thousands is meaningless and can be basically ignored. Civs currently are mostly balanced with some minor changes needed on a couple that are over or under performing (but nothing drastic or so bad as to need a hotfix)

Heck if we look at your example of Jeanne vs Byzantines (the largest skewed matchup by winrate atm). Jeanne is apparently terrible against the Byzantines in low elo based on stats amusingly. It’s easy to cherry pick high performance and low performance stats at such low sample size.

So you took the absolute outlier in around 200 matchup and made your entire argument based on that 1 outlier ?
Where is the 10% and 90% ?

The DLC was just released like 2-3 weeks ago, patch will come.

i should also point out that on the patch notes 2 weeks ago, they said there was going to be a late November/early December balance patch. It’s not like a patch isn’t about to drop in the next week or 2 already.

1 Like

Jeanne and Ayuubids are OP as hell, don’t try to hide this saying is a low sample size.

Check Jeannne stats and Ayuubids stats.

There you have a sample near 20k for each civ, and win rates in most of the cases don’t drop from 58-64% that’s insane.

I would recommend anyone who wanna rank just use these OP civs or don’t play ranked until they nerf hard them.

2 Likes

I refer you back to the second sentence of my reply which it seems you did not read.

So anything with less than 1000 is not significant, ok, got it. Love to see how 99,99% of all gamma studies are utter trash despite their publication in scientific journals and practical implications. Honestly… at which university did you take your statistic courses? I’ll be sure to not send my kids there, lol.

100 is an acceptable sample size on things with limited variables, but something like an online game with tons of variables from experience to players is a completely different story. Heck even just population surveys, 100 people is the bare minimum acceptable amount of responses to get usable data, and even that is considered usually to be far to low to act off of.

100 is an accepted sample size for things where you do everything possible to limit variables in the experience to a very small amount.

1 Like

jeanne and ayyubid are indeed too strong but this post is purposely exaggerated. They have 60% winrate not 90%.

Im sure they will be nerfed in next patch so there is no need to complain, they are planning balance patch already. This thread doesn’t bring anything constructive.

Some civs have near 60% winratios in Diamond/Conq 1v1 and even 70%/80% against specific civs.
In some parts of Diamond/conq some civs even have 65% PLUS winratio.

That stupid horse of the new french has to be reworked and massively weakened.
It completely breaks the game right now in any elo from platinum and above.

A massive balance-patch is more than overdue…
The devs got enough feedback from pro-players pre-release and nothing has been done about the balancing in 2 weeks POST release of this ridiculous dlc.
Something has to happen ASAP.

Also cliffside is a completely broken map, Malians are unplayable on it, as the goldmines are bugging and pitmines can’t be built on them.
The few ones, where a pitmine can be built, the tiles arround the goldmine are either blocked with cliffs or trees, so in many of them just 3-5 houses can be built to boost the pitmine’s gold income.

Come on Devs, come on! you can do better… hopefully… ?!?