Transcription
The Elite Classic just concluded this last weekend with the grand finals between Marine Lord and Poppy Paw. If you haven't watched it yet, go check it out and watch out. There might be a few spoilers in this video as well, as we'll be taking a look at the numbers of this tournament and try to break down what the current meta picks say about the civilizations and their strength levels.
Right now, we've got the data on 23 civilizations, win rates, the games that have been played, the bans, and when you actually look at all three together, some really interesting things pop up. So, let's go tier by tier.
If you were to do a tier list based on the win rates from the tournament without taking into account the different maps, the tier list would look something like this. You would have Mongols sitting here with their 85% win rate, and the rest of the civilizations that have 60% win rate and higher would be in S tier. Then in A tier, let's say that's 50% to 60%, that would be Roose, Golden Horde, Byzantines, Delhi, French, and Japan. The most surprising one here would be Roose, that's really had a resurgence, a real comeback with the Trinity builds that people have been going for. We've been seeing everything from two town center Kremlin. We've seen um one hunting cabin opener Kremlin into Trinity where people go pretty fast into Imperial as well. They're using the militia to go for these big timing attacks or to contest relics in two town center versus two town situations. Again, KT over here primarily a water civ and the same with Mongols. So, they're staying pretty high up here even though they should probably be lower. Lancaster also performing way better than it should and it's been a really popular pick in uh both practice and the tournament. And so it's a civilization that's also seeing a lot of play now with two town center and spear rushing.
China is quite low together with Ottomans down here in the sub 50% win rate. 40 to 50% win rate here. China, Ottomans, Abbisid, Macedonians, are civilizations that are usually considered A tier bar of course Lancaster and Abbisid. China and Ottomans often considered top three civilizations in the game by by pro players. Not having a high win rate is really, really surprising. The same with Macedonians and Hayubits. And then in C tier, well, now we're getting to the below 40%. John Do, English, Jin, Order of the Dragon. Not a lot of picks on these civilizations. They haven't been played a lot. These are civilizations that have typically picked more on hybrid maps or in a best of seven, best of nine where you have a lot more saves in your pool and the best civilizations are gone uh by the third or fourth pick. And of course, at the end of it all, 0% win rate that goes straight to F tier. There's no doubt about it. Tuklak Dynasty could not perform this tournament.
Before we get to an actual land tier list, I just want to talk about how it was built. The tournament itself, Elite Classic 4, was played across a mix of maps, both water, hybrid, chaotic, and land maps. And they're not all the same. Some are open land maps like Arabia, Lipony, and Gorge. And some are water maps where you have naval control, docks, and fishing completely change what civilizations are good here. And that's a problem if you're going to rank civs because a civ might be really good on one map and not very good on another type. But I think most people watching this video would agree that we would define a good civ based on how it performs on land, at least for an average player. So I pulled them out. Every game playing it on Cavasan, Four Lakes, and Lakeside is gone. And I also removed mountain clearing since it's a very chaotic land map. Even though it is technically a land map, it plays so differently that it would skewer the data here in my opinion.
Start at the top. The strongest civ in the data isn't one of the originals. It's Sengoku Damio. 30 games played and a win rate of just well under 67%. What makes that stand out is it's not a niche pick. It's both popular and winning. And that combination is the clear signal that we have a civ that is generally strong, not just uh a small sample size with a high win rate. Right behind it is Shushi's Legacy. Uh sitting around 62 win 62% over 21 games. So, the two barbarian civs are leading on land actually. Um, [snorts] Malians technically post the highest number on the board at 80%, but that's off of just 10 games, but it was also banned uh a hefty amount. Uh, I think it was the second most banned civ right after Ottomans.
Now, here's the part people skip and you actually shouldn't. Once you get below the top three, almost everything clusters around a sort of coin flip. Ottomans, Delhi, and French is in the high 50s, um to low 50s. Roose, Byzantines, Abbisid, and Chinese. All within a couple of points of 50% in either direction. It means the map you're playing and the player in the seat are deciding games far more than the civ on the loading screen. For a competitive game, that's about as healthy as the numbers can get.
Jump to the Golden Horde. That's the single most revealing data point. It's actually the fact that it's the most picked civ in the entire data set, data set chosen in roughly a quarter of all games. It is by a clear margin the community's default. And yet its win rate is just 52%. So uh yeah, basically a coin flip there. At the other end, uh a few civs are generally struggling on the land right now. Japanese and Knights Templar are the weakest with a real sample behind them. Both sitting around 36%. Holy Roman Empires and Macedonian Dynasty aren't far ahead or stuck in the high high 30s to low 40s. If you're picking purely to win on open maps in this patch, these are the names of the names the data is quietly steering you away from.
Win rate is tangled up with who's playing the civ. A top-tier player on a mid-tier civ still beats a weaker opponent. And these numbers describe outcomes, not a raw civ power in a vacuum. Uh, and second, the sample size matters a lot. The 57% win rate from Ottomans is just seven games. So that doesn't really tell us as much as it does for, for example, Sengoku with a 67% win rate. Uh, Mongols, English, and a few others barely got picked at all, which is why that I've left them out of the ranking instead of just putting them in there. and modeling the waters a bit. So, uh they've been putting this sort of a question mark uh with less than five picks for land maps, although they were picked a lot more for water, which would probably explain.
Before we jump back into the game, a quick message. If you're serious about improving at Age of Empires 4, I offer one-on-one coaching where we work on everything from strategy and decision-making to mechanics and build orders, no matter what rank you're at. And if you want even more content to help get better at AoE4, check out my members channel. Here you'll get access to exclusive guides, recorded coaching sessions, strategy discussions, and behind-the-scenes that isn't available anywhere else. So go check out the links are in the description. Now, let's get back into the game.
I think the real story in all of this data and tier listing and everything is that the meta right now at the pro level is especially map and matchup dependent. So [snorts] that means that when we talk about maps, a map has an incredible influence on how a civilization performs and that's also shown in the data. And then the same thing with the matchup, every matchup is either sort of neutral, really good for one civ, and really bad for another. On Hill and Dale, a civ like House of Lancaster did four and zero. Uh on King of the Hill, it was three and two. On Arabia, it went two and six. It is not a weak civ. It's a closed map civ. and uh it does really well because currently the meta for uh Lancaster is very two town center dominated. It's the same story for Macedonian Dynasty. It's zero and five on dry Arabia, but it's winning on coastal cliffs and hill and dale. It enjoys having deers close to it, uh which you do on hill and dale and coastal cliffs. The boar is usually really close to the town center, so there's a lot of food that can go underneath that golden horn tower. The Sengoku might be the top civ. It's extremely powerful on on open maps. It's four and zero on dry Arabia. It's seven and two on Lipony, but it actually drops to one and three on Coastal Cliffs, even though it should be fine there, but there's just not enough games to say. Uh, so it's kind of it's kind of fine. Uh, French is uh kind of the same, four and zero on Coastal Cliffs, 72 on Dry Arabia. Uh, but it kind of loses more on Gorge and Bald. And the map that's quietly sorting the field is Dry Arabia. uh the most played and most open map. It's brutal on the slower closed map civs like Japanese which went zero and five, Macedonian that went zero and five, House of and Lancaster that went two and six, but it rewards French, Sengoku and Golden Horde. So, uh if you want the takeaway here, it is that the more open the the map is, um the more the safe choice punishes you.
So, here's when it actually gets really interesting. If you look at this graph, you can see how many times during a set or a match or whatever you want to call it, a best of five, best of seven, you can see how many times that civilization was banned and Ottomans was banned 100 times during this tournament. So it averages out at 1.64 bans per best of five per match, etc. Malians follow suit right below it. And Byzantines is the third most banned civilization. So of those three, those are the ones that people fear the most. What's interesting though is that Byzantines were played 16 times but it was it has only won 50% of its games whereas Malians won 80%. And 57% of the win rate of Ottomans doesn't truly justify the 164 uh ban rate for Ottomans. So it's pretty interesting to see that there are also other civilizations like for example Sengoku which is banned zero times and it has a 67% win rate across 30 land games. So everyone's banning Ottomans and it's coming out with 50% 57% win rate, but Sengoku is being let through and it's winning 23 of all matches. So that those are some crazy stats, but again, every time we talk about these stats, there's always this caveat of some pro players being really good with some civs, others playing them because they're supposed to be good and then losing, right? So it's it's an extremely complicated math problem. So I'm not going to give any conclusions definitively here other than what the data itself uh allow us to observe, which is people fear these civilizations but they don't always win.
So if you were to do a final land tier list to sort of understand the current uh strength level or perceived strength level of a lot of civilizations taking into account how many times a civ was not played in a set uh but instead how many times it was banned. We clearly now see that Ottomans were with the 100 bans the most banned civilization and then with 50% 57% win rate not being the highest still this 1.64 uh bans per match just completely um squanders like swamps the the competition there. And then Malians coming in with the second most banned and then 80% win rate. Byzantines the third most banned but only a 50% win rate. And then Sengoku with no bans but 67% win rate puts it right up there. This is the surprise. This is not a surprise. At least not to me. It might be to some people. That's why I'm making this video. But yeah, this is a surprise to me after all this uh data mining. Shushi's Legacy was banned quite a bit as well uh of the civs that weren't heavily banned. And um it comes in at 62% win rate. This is a really solid civilization right now. And and I would even go as far as saying on a personal opinion, this is a little bit of a broken civ. Delhi alternate didn't get banned, but has a 56% win rate. That's really solid. French got banned a little bit. 53% win rate. Now, these bans might also be player specific. If you go up against Lucifer and Vortex, there's a chance you're going to pick this civ first or you're going to ban it first. I see some people do that at least. Golden Horde banned very little. Um played a lot. 52% win rate. Uh, overall solid default civ to have in your draft um or to play if you're not drafting. Roose 53% win rate, a lot higher than people were ranking it on the tier lists before the the tournament. Abbisid at 50%. China with a pretty decent ban rate but underperforming on the win rate due to some players not uh knowing how to play the civ and other players carrying the win rate. So, uh, yeah, that's that's also going to explain many of the other wins and losses since this is overall still a very low sample, um, of games. It's what we have right now. Macedonians banned a decent bit, very underperforming. Lancaster 44% win rate, picked a decent bit, and that is where the civilization tier list also can be a little bit awkward because people pick civilizations in a draft. They might pick for best of five, I think it's seven civilizations and two of those civilizations need to be water civs if it's for a map pool with five uh four land maps and one water map. Right?
If you look at Knights Templar, there are 36 36% win rate on land. They're still banned a bit. That's not because people fear Knights Templar on land. It's because Knights Templar has a 90% win rate on water only. So if you exclude land maps, Knights Templar is super powerful there. Mongols have an 83% win rate on water with a 10 and two. Uh Lancaster has a 71% win rate on water, 67% win rate for Japanese on water. And uh yeah, Order of the Dragon has barely been picked on uh on land, but it was picked a lot on water, especially Lakeside. It was considered one of the best civs. So yeah, we there's a danger here of calling this the meta. Um, but yeah, again, it's hard to make this video cuz there's so many nuances. And that's going to be it for this video as well. I hope you guys have enjoyed it. It's my first attempt at trying to use um some intelligent AI. I use Clot for this video. And uh I think this is a good use of AI to try to understand these large data sets because I would never be able to do this without it. But I also will fully accept that there might be some um some mistakes in in these data that the way I'm representing them, my understanding of it. I know that there are probably going to be some mistakes here and you guys can call them out in the comment section. Um that's totally fair. Um that will only help me get better at it. Um but yeah, as I said, um I tried to cross-check as much as possible, um when I thought it was really important to do so. And uh I didn't run into issues actually. So yeah, I'm I'm pretty confident this is this is how it is. So yeah, hope you guys have enjoyed this video and uh don't forget to like and subscribe and uh leave a comment. It helps the algorithm and I'll see you guys in the next video. Peace. That's it.