Hi fellow AoE lovers,
The perennial question about best and worst civs has always surfaced whenever we talk about the game. The answer almost always 'Depends' on the map, on the player, on the settings amongst other variables.
For some time, I have wondered what data says about which are the top civilisations, and whether or not, it can even be established on the basis of data. This analysis (Civ Balance - Reddit) seems to suggest Franks, Incas, Aztecs up there while Khmer, Vietnamese, Sara as the bottom three.
I did similar analysis on the win rate of top 20 players on the Voobly 1v1 ladder for different civilisations. (Data taken on 28/10/2018). Analysis is presented here - AoE_CivData_TableauDashboard
Objectives of Experiment: Please note that the objective is not to get the precise answers to the best civ question, but just to bring in more data in the analysis aspect of the game, and have fun with it. Hence, I have refrained from making any conclusions in this post. The reader can go through the analysis and play with the data using filters for different civilisation and players, to come up with their own conclusions on who plays which civ the most, who wins with which civ the most etc.
Total games considered for analysis = 2903
Arabia = 1616 (55.6%)
Non-Arabia = 1287 (Includes custom arabia maps like ECL_Arabia etc)
Hence, though map is not considered for analysis, it is somewhat an Arabia based analysis.
Disclaimer and Observations:
Before starting with the analysis, I acknowledged that the analysis would be far from perfect due to the following reasons -
Feel free to give me feedback. And please add your salient observations in the comments.
Also, would love to hear inputs on various other aspects of the game where data can be used for analysis.
The perennial question about best and worst civs has always surfaced whenever we talk about the game. The answer almost always 'Depends' on the map, on the player, on the settings amongst other variables.
For some time, I have wondered what data says about which are the top civilisations, and whether or not, it can even be established on the basis of data. This analysis (Civ Balance - Reddit) seems to suggest Franks, Incas, Aztecs up there while Khmer, Vietnamese, Sara as the bottom three.
I did similar analysis on the win rate of top 20 players on the Voobly 1v1 ladder for different civilisations. (Data taken on 28/10/2018). Analysis is presented here - AoE_CivData_TableauDashboard
Objectives of Experiment: Please note that the objective is not to get the precise answers to the best civ question, but just to bring in more data in the analysis aspect of the game, and have fun with it. Hence, I have refrained from making any conclusions in this post. The reader can go through the analysis and play with the data using filters for different civilisation and players, to come up with their own conclusions on who plays which civ the most, who wins with which civ the most etc.
Total games considered for analysis = 2903
Arabia = 1616 (55.6%)
Non-Arabia = 1287 (Includes custom arabia maps like ECL_Arabia etc)
Hence, though map is not considered for analysis, it is somewhat an Arabia based analysis.
Disclaimer and Observations:
Before starting with the analysis, I acknowledged that the analysis would be far from perfect due to the following reasons -
- Many top players don't play seriously in rated games.
- Analysis might be skewed as top 20 players can win with any civilisation when playing against other players.
- Some civilisations such as Khmer and Portuguese invite unorthodox play which affects their win rate.
- Analysis might be skewed as some civilisations get played a lot, while some others are not played leading to not enough data to suggest something conclusively.
- No civilisation was grossly underrepresented and except Huns, no civilisation was overrepresented, thus confirming that the sample of games chosen is indeed random civilisation.
- The win rates of all top players was obviously above 50%, yet the variations are significant.
- (Most importantly) The win rates of civilisations were, till some extent, as expected.
- There was no conclusive link between the voobly ranking and the win rates for players, which was an expected result.
Feel free to give me feedback. And please add your salient observations in the comments.
Also, would love to hear inputs on various other aspects of the game where data can be used for analysis.