Looking at the game statistics in that fantastic spreadsheet the bro’s have created I’ve realized that this game is totally broke and who is steering it.
I play brit’s at the 1250 level usually. At that level Brits are fairly balanced in their results as are they at the lower. Swedes are busted op with better than average results across the board.
However, you look at the 1450 level the Brits are op and swedes strong but not too oppressive.
So we get a pup and Brits are heavily nerfed and swedes slightly modified on torps. Brits are essentially losing a vil. Obviously the mods are not considering the mid and lower levels of play. These changes have to be driven from the top level of play.
Here’s the problem. Most of the players are at the mid a low levels. Brits may become balanced at the top but will be at huge disadvantages where everyone is at, below 1450.
The game is great and I’ll keep playing but I never really understood that I was in an entirely different game than I thought. I’ve been on this board spewing what was essentially correct where I was at. Arguing with players saying Brits were broke and op. Me blasting the swede lame. Both were right.
There is no way to reconcile any of this,ever. We may as well stop talking. The statistic drive the changes and they are taken from the top. None of this debate really matters.
As someone who helps them i cant tell if this satire, purposeful misinformation by someone trying to discredit their work, or an honest mistake and misunderstanding of statsistical processes. At this time british only have enough data to make some conclusions to about 95% CI at 1100+. 1250 and above there’s not enough data. While brititsh win rate is currently up, the chance of it being due to something besides civ strength aka error(skill differnce, meta change new players trying brits new strats etc.) I beleive in time it will be able to control for error but not quite yet.
And certainly no one from the project has made any level of sweeping statements such as yours nor will they as there is not enough data. Besides stats inform conversation not remove or trump conversation, and with time this can help test peoples theories not remove them entirely. Analysis ive done shows brits had great internal balance and consistent win rates accross elo for example, not brits 100% op or fair civ or skilled civ etc.
People in general struggle to undersatand stats and proper useage so please feel free to ask questions and ill even walk you through some stuff or you cam ask the main guys directly, but lets temper statements to be factual for both aoe3 and the stats used to help assess aoe3 as they get more refined. Maybe this post was a bit cheeky , but ive seen alot of people recently use these stats in ways they aren’t intended, some purposefully most ofc just a bit off. Hope that helps!
On the lower level of elo this will have zero effect I’m talking 1100 and below because those players are not losing games because they lack a vill, often tc will be idle for minutes at a time. This change is to target 1100+. I’d also suggest that it might not only be down to brits win rate but also the fact they’re the most playd civ.
There is not enough data to state the current rate isnt due to error. Not on win rates, too much variation to get it past even .8CI. Everytime I run analysis the low statistical power plus higher variation of intra matchup win rates messes with the alpha and rejects nulls.
Now eyeballing brits do have some consistent data and its likely over time we will see actually decrease in win rate at higher levels (but patch will mess this) while still retaining its overall lack of truly bad matchups as data comes in. But to say conversation is pointless when the data iirc…60% upper elo brit win rate is not due to error aka chance or other factors is not a super solid foundation yet.
I really enjoy the conversation this has sparked and believe in time the data will be such we can chart things from win rates matchups thry even charting win time to see when civs powerspike like there is so much. But if we jump the gun we only discredit it in the future.
Its better than nothing right now and does a good job of stating a trend. But thats about it. We wil get there
Sorry for tone but ive seen alot of people jump way overboard with this already and think its good to publicly state the limitations and strengths because people and stats mix like oil and gasoline on fire. Apologizes, but as the data grows i think we need to better state the underpinnings, methods, results, and conclusions more directly before saying “haude is perfect civ at its winrate per the grid” even though its failed to pass regression chi sqaure and r2 on why its wins rates are what they are"
Don’t look at win rate. It is useless, every one who consistently play one civ will reach 50% winrate and stabilise there. You gotta look at the same player, using different civs, what elo can they achieve with different civs when winrate settles at 50%. Players play haude and stuck at 1100 and play Brit easily 1400, and their winrates are both 50%.
I don’t think that’s right. If your civ is op it levels out at a higher percentage. When I look at win/loss percentages for players that only play swedes they always have a 20-30 game win advantage. I play mostly Brits and I’m dead on 50% win/loss.
There is more to it than you suppose. The matches are generated about at your skill level but as you get more lopsided matches than not you end up with slight edges over time because all things are not equal.