[KotD2] Final rounds statistics

FranceTarsiz

Longswordman
Feb 27, 2017
415
1,097
108
27
Seattle
#1
After a weekend packed with KotD action, the tournament is (sadly) over, crowning TheViper who successfully defended his title against last year’s runner-up, Liereyy. The final took 8 games to finish, following roughly the same pattern as last year, with Viper taking the first three games, Liereyy clawing his way back, and Viper eventually closing the series in a game that saw him heavily fortify his base.

Prior to the final, TheViper had dropped his first game of the tournament to the hands of Yo, eventually edging him 4-2 to qualify for his second straight KotD final. Liereyy had to face off against a nemesis of his, Vivi aka Fat_Dragon, who had beaten him in a recent showmatch and in the semifinals of the ECL Europe East 1v1 phase. After losing the first game to the Fat Dragon, Liereyy had put on a convincing performance and won four games in a row to join Viper in the final.

Country statistics
In total, 32 of the best players in the world took part in the second edition of King of the Desert, a 1v1 tournament set only on a special version of the most popular and played map in the game, Arabia. These 32 players represented a total of 15 countries from 4 continents (only missing Africa and Oceania). Asia led the way with 12 players, followed by Europe (11 players), South-America (6 players) and North-America (3 players).

The country with the highest representation was China with 7 players, making up almost 25% of the draw in the first round, improving to a 50% representation in both the quarterfinals and semifinals.

Brazil came second with 4 players, but only put one player in the second round and none in the quarterfinals.

The third place was shared between Vietnam and Finland with 3 players each, but, like Brazil, failing to place any player in the quarterfinals.

Per capita, Finland wins with only 1.83 million people per player in the draw. The smallest country featured in the tournament, Norway, comes second with 2.63 million inhabitants per player. The United States come last with only one player, equaling 326 million people per player, with China coming just next at 198 million inhabitants per player.

The tournament featured 3 same-country matches:

F1Re vs RiuT in the first round
Yo vs Lyx in the quarterfinals
Fat_Dragon vs yinghua in the quarterfinals

Civilizations statistics
The later stages of the tournament featured drafts reusing the same rules as in the first edition of King of the Desert, preventing mirror matches and enabling players to ban civilizations. The quarterfinal round, however, was controversial as the “non repeat through the first three rounds” rule, supposedly making life harder to the higher seeds in the first round, ended up with lopsided matches in the quarterfinals: TaToH and MbL not having access to any of Aztecs, Chinese, Mayans and Slavs, considered the 4 best civilizations on the map, while their opponents still had not used any of these.

Considering the overall tournament and only the civilizations who were played at least 5 times, Incas surpassed Chinese as the civilization with the best winrate, at 75% (6 wins in 8 non-mirror games). Chinese remain a strong second with 70% (14 wins in 20 games). Franks come third with a 64% winrate, winning 9 out of 14 games.

Ethiopians performed the worst with only 1 victory in 5 games (20%), followed by Spanish with just 29% wins (2 out of 7), Indians at 30% (3 out of 10). The Mongols were a very popular civilization, but that popularity did not match their actual performance in games as they got the 4th worst winrate of the regularly played civilizations with 31% and just 4 wins out of 13 games.

Aztecs, Mayans, Slavs and Chinese were considered the top 4 civilizations of the tournament and were indeed the four most-played civilizations with respectively 31, 28, 21 and 20 games featuring at least one of these. They were the first 4 civilizations drafted by the two finalists, were all played and won all their games but one; the Mayans/Aztecs clash won by Liereyy.

Goths, Italians, Koreans, Persians, Portuguese, Teutons and Vietnamese were the only civilizations never played during the event.

Looking at the rounds, I had the idea of separating games played in the first two rounds and games played from the quarterfinals onwards They make out for roughly 2/3 and 1/3 of all the games respectively. The idea behind that being that, from the quarterfinals, match ups are supposed to be much closer than in the first two rounds (with TheViper, top seed, winning against BL4CK, 32nd seed, with Khmer against Aztecs).

Comparing those sets of games, Huns do much better (100% winrate from the quarterfinals and onwards against 25% in the first two rounds), so do Slavs (100% against 35%), Malians (100% against 50%) and Aztecs (80% against 31%). The number of games played is small so it’s hard to draw meaningful numbers, but Japanese went down from 67% winrate to 20% (with 9 and 5 games) and Vikings from 80% to 40% (with 5 and 5 games).

Towers statistics
Some updates on the tower numbers for the whole tournament, and for the last three rounds. As a reminder, here are the factors I tried to look at in every single game to study the amount of tower aggression and tower defense:
  • How early is the trush? Towers early in feudal age are a trush, towers in late feudal age or even castle age and later don’t really qualify as tower aggression as the backbone of one’s strategy
  • How many towers are built – simpler, basically one tower is a “no, it’s not a trush”, more than one is a “yes, it’s a trush”
  • For tower defense, 2+ towers => “yes, defensive towers were built”
With that in mind, there were, in total, 59 games that featured at least some kind of trush, out of a total 122 games, representing 48.4% of the total set of games. The winrate of trush, in case of trush from only one side, was just above average at 55.8%.

The first round heavily featured towers, as 61.4% of the games did (35 out of 57). There were less towers in the second round, but tower aggression was insanely successful during that round, at 92.3% winrate (when one side only trushed, they won 12 out of 13 times, the only loss coming from ACCM trushing… TheViper).

The number of trush games significantly decreased in the later rounds of the tournament, with 4 trush games in the quarterfinals (26.7% of the games) and only 1 in the semifinals (9.1%).

There was a total of 14 games (11.5%) where both sides trushed.

Players statistics
One of the interesting differences in style between TheViper and Liereyy is the defensive, macro style of the one opposed to the more aggressive approach of the other. This is reflected by the average length of their games: Viper’s average game lasts 46:12 while Liereyy’s average game lasts 36:21. There is little difference between the average time of a victory or a defeat for Viper, but Liereyy wins even faster, at 34:34 in average (average loss at 41:06 game time).

TheViper is the player who spent the longest time playing during the tournament at 17:42:38 total game time – helped in that with the fact his series against Liereyy and Yo were the longest two of the tournament. Despite playing 3 less games than Liereyy, Yo takes second at 14:13:14 game time. Liereyy comes 3rd at 13:19:40 spent playing. Joint 3rd of the tournament, Vivi only shows at the 6th place of players having spent the most time playing, behind his countrymen Lyx and yinghua.

BL4CK is the player with the longest games on average, his games lasting 57:54 on average. F1Re is the player with the shortest games at 26:17 average game time.

Viper is also the player who built defensive towers in the most games, building them in 8 games (35% of his games). Yinghua also made a lot of defensive towers, building them 7 times, which represented 54% of his games.

To each his own: Viper opened 10 games with scouts, more than anyone else in the tournament, while Liereyy’s strategy of choice was men-at-arms, which he did 14 times, more than anyone else.

Trivia stats
122 games were played during the whole tournament, with the lower seed emerging victorious on 40 occasions (32.8% of the games). In total, these games took 85 hours, 22 minutes and 54 seconds to complete.

The longest series of the tournament was the final between TheViper and Liereyy, lasting 6:14:27 game time and 8 games. It was longer than the semifinal between TheViper and Yo by more than an hour and a half.

Men-at-arms were the most popular opening, played 124 times. Scouts were second, played 86 times. The third most popular opening, Drush FC, was only attempted 12 times. The non-mirror winrate of the two most popular strategies evened out at 52% for men-at-arms and 48% for scouts.

A total of 31314 units were killed by the players during all their games, with an average 257 units per game. The longest game, during the series between yinghua and LaaaaaN, caused the death of 2761 units alone. On 23 occasions (18.9%), the winner of the game had a negative K-D. On five occasions, due to friendly fire and monk conversions, both players ended with negative K-Ds.

Closure
Thanks for all who’ve read this and the previous (very) long posts I made on statistics and write-ups for the tournament. I had a lot of fun watching the games and trying to find out some, as much as possible, meaningful stats for you guys. I hope this was insightful or, at least, interesting to read. I am sure there are more statistics that can be crunched from the numbers and details I marked down from watching the games, so I have attached the Excel file I used to monitor the tournament. You will find the raw data (in the “Games” tab) that you can use to compute any statistic you feel like I missed, in addition to the integrality of the indicators I’ve made.

Obviously, from a data perspective, the subset of games played in a tournament such as King of the Desert, where players are handling as many as 31 different civilizations, is still very small and thus very hard to infer anything from. I’ve always felt the in-game statistics were a bit lacking in a sense of explaining why a player wins or loses (since they reflect achievements rather than methods used to attain the final result), but I feel like with the recent tool @_unicorn developed for the analysis of recs, we could move one step further. Next time!

Anyway, the usual thanks to Memb and Chrazini for the organization of the event, it was hands down one of the most professional Age of Empires tournament I had the occasion to follow. Thanks to all the generous supporters who helped make this event a reality with a sizable prizepool and to all the casters who covered the games. The record viewer numbers of yesterday only bodes good things for the future and I’m excited to see what comes next!
 

Attachments

FranceTarsiz

Longswordman
Feb 27, 2017
415
1,097
108
27
Seattle
#8
How did you got those infos? That is really well done, very interesting read.
I had to do most of it by hand.

I watched half of the games of the first round and then downloaded the recs of the ones I had not seen. For every game I paid attention to opening strategy and the choice to make towers or not, and then wrote down the other stats with the post-game achievements.

Once all the raw data was collected, calculating the stats was just a matter of playing with Excel.

Thanks for the comments!
 
Jan 30, 2015
127
63
43
Minnesota
Voobly
AlphaAndOmega
View profile
Ladder
RM - 1v1
Rating
1807
Wins
244
Losses
203
Streak
-3
#11
Very illuminating. I guess I should stop doing archers in 1v1s haha. I am shooting myself in the foot by not starting with men at arms or scouts! Also should use towers more but I already knew that, just don't like towers games as much.
Top Civilizations were kind of obvious :P. But the game length stats are sick! People are saying in NAC that Viper always plays for late game and it is borne out in these stats! Liereyy being aggressive too (it's surprising to me).
 
Likes: OLADUSHEK

Time

Your time
G M T
Your zone

Upcoming Events

ECL SE Asia 1v1 MbL vs Zuppi
Tuesday 14:00 (GMT +01:00)
ECL SE Asia 1v1 MbL vs Zuppi

Age Of Empires On Twitch

There are in total 31 streamers online
Click here for details
Age of Empires II 361 viewers
Age of Empires II 25 viewers
Age of Empires II 23 viewers
Age of Empires II 19 viewers
Age of Empires III: The Asian Dynasties 17 viewers
Top