As many of you may have seen by now, _TheGovernor_ and I arranged to build some advanced statistics of our War is Coming matches. We set off on a whim to try to discover some interesting pieces of the greatest AoC tournament to ever happen such as: what variables are important in predicting what teams win? Who are the strongest teams at the moment? What teams can be judged to be "most similar"? What would a good statistical model predict for future outcomes of the tournament? After putting in some serious time in the past few weeks, I believe TheGovernor and I are finally ready to deliver some results.. And they came out to be extremely interesting.
First Phase: Data Gathering.
In deciding on a statistical model that can accurately predict anything you may be interested in, you first need to make sure you have really good data that your models can be based off of. That being said, TheGovernor and I first started talking about the variables that we can get, and that might be interesting in predicting the games: 1v1 Ratings, team game ratings for each of the teams was the first obvious piece of data we collected. We decided to take that even further by adding "number of games played in the past month", which required going to each and every player's profile, and counting the number of games they have played for the entire month prior to their WiC game. As a subset of this variable, we also made the variable " least games played" and "most games played" which correspond to the player on that team who has played the most in the past month, and the least in the past month. We then decided that even though 1v1 and TG ratings were very essential, we needed to include something of a "difference between rates" for the teams that were playing each other, which we dubbed "1v1 rate delta" and "TG rate Delta". Finally, we recorded 3 variables that correspond to the games played in each series: a binary code that said "1-0-0-0" if a team won the game, "1-1-0-0" if a team won the game, as well as the previous game, "1-1-1-0" if the team won the game, the previous game, and the previous 2 games in a row, and finally "1-1-1-1" which corresponds to winning the game, winning the previous game, winning the previous 2 games in a row, and winning the previous 3 games in a row.
Phase two: Exploratory Data Analysis
Part 1: Clustering Algorithms
Taking this datamatrix that consists of 11 predictive variables, all attempting to predict "1" or "0" (win or lose), I first began by converting the data into a Variance-Covariance matrix. From there, I decided I wanted to use what is known as a "Complete-Linkage Clustering Algorithm", a simplistic page of how it works can be seen here: http://en.wikipedia.org/wiki/Complete-l ... clustering
In a mathematical sense, this algorithm is fairly simple, and attempts to find patterns on how "similar" each team is from each other. Here is the output of the Algorithm:
How does one go about interpreting this output? The "Height" on the left corresponds to how similar two teams are that are in the same branch. So in this instance, we can see that for the Complete Linkage Algo, it found that Tyrant Warlords and Tyrant Legends were very similar, while VN,PoZ, and PL were all very similar. Likewise, it found that SY-B, PL, poZ, VN, nC, and VNS are all weakly similar (in the same overall branch) while SY, NaBs, L_Clan, Str8, KGB, poJ, zZ, Tyrant L and Tyrant W are all weakly similar as well.
After using the Complete Linkage Clustering Algo, I decided to see what an "Average LInkage Clustering Algo" would produce, and here is the output:
The results from the Average Linkage, despite being fairly similar to the Complete Linkage, had some differences. For instance, now only the Tyrant Teams, SY, NaBs and L_Clan are considered to be weakly similar (same overall branch) while the remaining teams are all on a seperate branch. However, it still lists the two TyRanT teams as being very similar, as well as NaBs with L_Clan, as was indicated by the complete linkage algo. Please keep in mind that these algorithms are not biased by me in ANY WAY, and that interpreting these graphs is fairly arbitrary. I would love to see an open discussion of these images (as well as discussion of all further analysis in this post) from everyone reading this!
Part 2: Discriminant Analysis
Moving away from our simple clustering algorithms, I then created a Linear Discriminant Analysis of our dataset. Once again, for anyone who may be interested in how it works, a simple run through of the basics of it can be seen here: http://en.wikipedia.org/wiki/Linear_dis ... t_analysis
Some of the most significant variables in Graph-Matrix form can be viewed here:
From this we can see how the algorithm is choosing to divide each pair of our predictive variables (such as Average 1v1 Rate, Average TG rate, etc) to find the most optimal method of depicting if a team should win or lose. Some of these come out to be pretty obvious: for example, if you look at the Average TG - Average 1v1 rate pair (first graph down from the top left), we can see that a team is predicted to win "w" if the Average TG rate added to the Average 1v1 Rate is above the threshold going from the cyan color (bottom left) to the green color (top right). The little w's and l's you see in there are real data points that have occurred in WiC so far.
Part 3: Hidden Factor Analysis
In many problems in modern science, in particular the social sciences such as Economics, Financial markets and psychology research, researchers have been interested in quantifying the "hidden factors" that might be driving a data set, but are not quantifiable (there is no explicit data for it). In essence, I utilized something known as the spectral decomposition of the variance-covariance matrix, combined with a "varimax rotation" to find linear combinations of our predictor variables in an attempt to quantify our hidden variables. Due to the binary nature of part of our dataset, I was forced to only consider the variables Average 1v1 Rate, Average TG Rate, Least Games Played in past month, Most games played in past month, Average Games played in past month, 1v1 Rate Delta, and TG Rate Delta. The output is the following:
As one can see, the first three components combine for roughly 81.7% of the total variation in the data, making 3 "hidden" factors seem appropriate for our dataset. The first factor puts a heavy emphasis on the average 1v1/TG rates (.637 and .881, respectively) as well as the 1v1/TG rate deltas(.674,.872), indicating the first hidden factor is simply the rating strength of the team relative to it's opponents team.
The second factor puts a very high emphasis on Average Games Played in the 1 month prior to game, as well as putting a heavy emphasis on the Low Games played in 1 month prior (.778 and .962 respectively). It seems to also add in a small amount of weight to Average team game, average 1v1 rate and TG delta rate. In my opinion, this hidden factor has something to do with the team's teamwork, or how much they are playing.
Finally, the third hidden factor puts a whopping .994 on Highgames, .483 on Average Games, .312 on TG rate Delta, .512 on Average TG rate, and a surprising -.149 on 1v1 rate. This indicates a variable that puts a high emphasis on team game rates, and even more emphasis on the player on the team who is playing the most, while at the same time discounting the 1v1 rate. Therefore, I want to say that this variable is something of a leadership variable, due to the high emphasis on HighGames variable.
*Please keep in mind, that my interpretation of each hidden factor is completely arbitrary, and is open to any and all discussion!
Phase 3: Statistical Modelling
Now here comes the fun (and hard) part of all this analysis. I spent a fair chunk of time attempting to work on several different statistical modelling techniques that could fit to this situation, and eventually came up with a model that seems to work out pretty well! After finding a good model, I broke it into two pieces: one model is used for power rankings, while the second model is used for game-by-game predictions. Instead of posting in-depth results of the models here, you can see the power rankings in the upcoming pieces of this post, conducted by _TheGovernor_!
Team Previews and power rankings:
VNS (vs) TyRanT Warlords
Historic of the Teams in WiC:
VNS's historic
TyRanT Warlord's historic
Prediction
VNS's players
VNS_ACCM
VNS_yellow
VNS_SeOul
VNS_Fish
TyRanT Warlords' players
RiuT
F1Re
BacT
Slam
Commentary:
VNS:
VNS has lost a lot when BacT left them, even if they’re still such a strong team, they can’t compete at the very top level anymore. ACCM, yellow, SeOul and Fish seem to be their best team players composition. They are 1850, in both 1v1 and TG Rating, and they managed to take one game from KGB in the final of their group. That wasn’t enough to take the first place though, and they got one of the worst matches up possible: TyRanT Warlords! Then the've defeated VN (4 – 1) who tried to duel them. They’ll have to increase their level if they want to defeat TyRanT Warlords, but that seems a bit complicated. In case of a defeat, their last hope will be to reach the semi-finals of the loser brackets, which will be rather hard but not impossible: really good and close matches might come!
TyRanT Warlords
TyRanT Warlords is the second TyRanT team. Even if Legends’ players look a bit stronger on the paper, Warlords are still an important threat. Slam and RiuT are probably the best players of the team. Slam is always active, playing a lot every weeks and he seems in a really good shape. RiuT is playing a bit less nowadays, but he hasn't lost his skill at all, regarding to his recent 1v1 against Chris who he defeated 3 – 1. F1Re, seems to be the weaker member of TyRanT, despite he’s playing really well his post-imperial games. The games which are coming will give him a chance to show that he is a true TyRanT warrior! BacT is really trained, he got about 1500 TG games, in the ladder, and is in the top 5. This team could get impressive result, especially since we can consider two duos, on a side RiuT & F1Re, who’re use to play together, and on the other slam and BacT, who played the recent 2v2 tourney together.
Power Rankings:
TyRanT Warlords: 1
VNS A: 12
SY (vs) VN
Historic of the Teams in WiC:
Vikingos Nomadas's historic
SY's historic
VN's players
VN_Rob3rStinson (TaToH)
VN_DarK_KnighT_
VN_MoOniTe_
VN_HEIMDALL
SY's players
Yo__
FengXin (Paladin)
NO_23
james__
Air_Abu_ (nomad games).
Commentary:
VN:
VN_Rob3rStinson (TaToH) is the main player of VN which is also composed by VN_DarK_KnighT_, VN_MoOniTe_, VN_HEIMDALL, 3 players who already have some experience in tournaments, but they didn’t often reach the final stages tho. With the arrival of TaToH who’s carrying the team pretty hard, they managed, rather luckily since they had the same number of points than VNS B to reach the double stage elimination. Then they lost their duel against VNS (4 – 1) which means they must play against SY. There’s probably no way to win this, so they’ll have to comeback through the Loser Brackets. They will have to count on a really good TaToH!
SY
SY, is the #1 Chinese team. Composed by Yo__, Paladin, james__, Air_Abu_ and NO_23 they are the winner of the last tournament game, the Nation Cup 2014. The main strong of SY is not the individual skill of their players, even though they’re pretty impressive, but their coordination, and their decision making. This is definitely a team which is made to play teamgames! They were invincible during the Nation Cup, even though they played against some strong teams like Brazil A and B and also Argentina A. That’s why only a few teams may stop them, the TyRanTs and perhaps the L_Clan or the KGB. They’re also one of the two teams which hadn’t been defeated in RM during the group stage (although they gave 2 games to BuzZ ‘cause they didn’t want to play them).
Power Rankings
SY: 2
VN: 10
Legion Clan (vs) CKF
Historic of the Teams in WiC:
Legion Clan's historic
CKF's historic
Legion Clan's players
dogao
Chris
FeAge
Miguel
CKF's players
[CKF]geoffrey
[CKF]Tim_
[CKF]_PBT_Dee
[CKF]IamOK
Legion Clan:
L_Clan is probably one the strongest team of WiC. Mainly composed by 4 expert players, Dogao, Chris, FeAge, Miguel, they’re training really hard every week-end. Dogao and Chris are probably two of the best TG players of the moment. Currently, Chris hold the #1 place of the TG Ladder, and Dogao and FeAge are in the top 12. FeAge is also a very impressive player, who’ve been improving lately, and he’s definitely a good number 3. As for Miguel, we've seen him doing brilliant games, but also less good ones. He's a good #4, but he's gonna have to be on a really good day if he wants to face the strongest teams of the tournament. Regarding to a lot of experts, Legion Clan is a serious top4 challenger. During their training games, L_Clan has managed to put down many teams, like NaBs and zZ.
CKF:
CKF is probably the second best Chinese team. Everybody is expecting great results from them, mainly due to the comeback of Tim_. However, they’ve been defeated quiet rudely by PoJ in the Group Stages. That put the CKF to the second place and give them a really bad match-up, against L_Clan. It’s a bit surprising that CKF didn’t play with Tim the whole RM games, and even more surprising that Lyx, still didn’t play a game despite he’s one of the greatest player of this team. Geoffrey and IamOK, are two really impressive players who’re both experimented, have a good team play and are really strong. CKF will have to train a bit more, and perhaps to get another team composition if they want to reach the top 8 of the tournament.
Please note, CKF was not added to the model due to model fitting issues. Therefore, they are the only team who do not have a power ranking
Power Rankings
L_Clan: 3
CKF: ?
Nadya's Boys (vs) PL
Historic of the Teams in WiC:
Nadya's Boys:
PL's historic
Nadya's Boys' players:
1st Tier:
_8th_Wonder_
Zuppi
(czech) Lojza_
mentalist_
2nd Tier:
NorthenPike
Spaden
Lavie_Head
PL's players
PL_Remi
PL_Buniek
PL_Dziamdziak
PL_Bity_kablem
Commentary:
Nadya's Boys:
Nadya’s Boys is probably one of the most well-known team of WiC. Due to Pike’s stream and lately Nadya’s stream many players and spectators had the pleasure to see them playing a long time before WiC. They’re definitely the most serious team of the tournament regarding to the number of hours they're using every day to train. It’s pretty rare that a day passed, without any games played from Nadya’s Boys. Another thing that makes the strength of the team is the fact that they got many members. According to Nadya and a few experts, we can basically separate this team into two tiers. First one would be composed by (czech) Lojza_, Zuppi, mentalist_, _8th_wonder_. (czech) Lojza_, recently won the final of The Arabian Conquest against [TyRanT]BacT_, mentalist_ reached the 7th place with Canada in the Nation Cup 2014. He’s probably one of the most impressive players of the moment. He reached 2k one year ago, but he’s now one of the better TG Players of the moment. As for the 2nd tier, we can put in: NorthenPike, Lavie_Head, Spaden. NaBs are mainly training against zZ, L_Clan, and got pretty good results! They also made a really good impression against poZ during the group stage that they defeated 5 – 0. True, poZ aren't really training for the tournament, but this is still an impressive performance!
PL:
The Polish team can be considered something of an underdog of the tournament. Not usually a conversation starter, PL managed to defeat all but SY in their group, having struggled vs. them in their matchup by going 4-1. Despite having low expectations for this team, they pride themselves in teamwork due to having teamed together for a long period of time (including in Nations Cup tournament).
Power Ranking:
NaBs: 4
PL: 15
TyRanT Legends (vs) Str8
Historic of the Teams in WiC
TyRanT 's historic:
Str8's historic:
TyRanT Legends' players
DauT
TheViper
JorDan_23
kkab
Str8's players
Emin (FoxValentine)
Jugband
Artras
(czech) Skull
Commentary:
TyRanT Legends are favorite beyond any doubts in this match-up. They've lost only game in the group stage, against Cz. Back to that time, some of the TyRanT players looked a bit rusty. DauT played many games since this match and trained pretty hard, during the last month. Indeed, he played about 70 1v1 against really good players like Spaden (3-0), Monoz (3-0), _Ra_ (7-4), FeAge (2-1), Edie (1-1), TaToH (5-2), Miguel (2-3), Yo__ (6-6), TheViper (0-5) and IamOK (4-0). Thanks to these games, he managed to come back to the top 10 of the 1v1 ladder. Back to the overall team, we still have no idea who NeferTiTi, the last player TyRanT Legends is. However, he did not play the most important match of TyRanT in the group stage, against Nc. The games were pretty impressive and TyRanT won 5-0, which means that DauT, TheViper, JorDan_23 and kkab will most likely play this stage!
On the other hand, Str8 are a talented team (2k average TG and 1v1) composed mainly by Emin, Jugband, Artras, (czech) Skull and Sletten. It's almost the same team which played in the Nation Cup 2014, where they managed to reach the knock out stage. They've been defeated then by the SY (4-0) but they fought pretty well which means TyRanT won't easily win. Emin is getting better and better and he's definitely reaching an impressive level. He managed to take some 1v1 games against TheViper, and Chris a few months ago. Talking about competitions, he reaches the 4th place in the 1v1 tournament, the Arabian Conquest, after being defeated 3-4 against (czech) Lojza_ the winner of the tournament! In WiC, Str8's players have done a really good job in the group stage by winning 5-0 every match up they had, except the one against zZ, where they've been defeated 3-1.
Power Rankings
TyRanT Legends: 5
STr8: 9
*Please note that TyRanT Legends does not fit our statistical models very well. This is in large part due to many of their games are played in secret non-rated games, as well as kkab having a 1600 1v1 rate, and only 1800's TG rate, with very few games played (rated). Therefore, the model has a hard time correctly fitting (predicting) the outcomes of TyRanT Legends games.
Dreamers (vs) nC
Historic of the Teams in WiC:
zZ's historic
nC's historic
Prediction
zZ's players
Nicov
Afrodita
Gary
PaTiN
Fedex
nC's players
Skittle
Andorin
Tsu_NilPferD
JohNNieW_aLc_eR (Arma)
Commentary:
zZ:
Dreamers team has kinda the same composition than in SY Nation Cup 2014 where they reached the 3rd position, after really tough matches against the Brazil A which have finished second. That’s why nobody should underestimate this team. Nicov is probably the most important player of the team at the moment. He has definitely improved pretty well those last months and years, and he won some important 1v1 against famous players few days ago like Yo__ (5-3), F1Re (7-1), mentalist (3-0)... zZ is also really strong thanks to their hard training. Indeed most of the players have played more than 75 games this last month and most of them are team game, played with other members of zZ. They’re mostly playing against NaBs and L_Clan.
nC:
nC is probably one of the weakest team in this double elimination stage. The same team reached with some difficulty the knock out stage of the Nation Cup, but that was with JorDan_23. The nC is mainly composed by Skittle, Andorin, Tsu_NilPferD, JohNNieW_aLc_eR (Arma), strong players who got about 1k9 TG Rating. That said, nC had some troubles to be qualified during the group stage. They've been for only 1 point, and they must now face the Dreamers (zZ) who’re looking stronger than them. Arma and Skittle will have to carry the team if they want to go any further in the competition. But in order to do that they’ll have to train a bit more than a few games per months...
Power Rankings
Zz: 6
nC: 13
KGB (vs) SY B
Historic of the Teams in WiC:
KGB's historic
SY B:
KGB:
KGB is really interesting team, formed very recently. Alive, GoKu, St4rk, Ra, Hasan, and Whack are the main players. So basically we got 2 members of Brazil A, one member of Brazil B, one member of Canada and finally, two members of USA A, 4 teams that have got (very) good results during the Nation Cup. Now, who is gonna play the final Stage? Even if we of course can’t know that, it seems that the third Brazilians have been really active during the Group Stage, so they might be preferred. Hasan looks like a good number 4, even though _Ra_ has played a lot those last months, and became really strong. On the paper KGB looks really strong, but they’ll need to train way more in order to defeat the best teams of the tournament. If the first match against SY B will be difficult, it will be nothing in comparison of the second match against the Legion Clan, that’ll be a good indicator of the strength of KGB.
SY B
Despite it is supposed to be a B team, SY B are actually really strong. Their only defeat was against TyRanT Warlords, vs who they managed to take a game. They’ve dominated quiet easily the Titanic team, and reached the second place of their group. The main players of the team are Yu____, pity_memory, xiaoxiong, and Ds___. One of the best advantages for SY B is probably that there are rather unknown; they’re mostly training in Voobly China and can create some surprises, especially on Nomad Games, where they look pretty strong!
Power Rankings:
KGB: 7
SY-B: 14
PoJ (vs) poZ (hong kong)
Historic of the Teams in WiC:
Power of Japan's historic
Poison of Zone's historic
Prediction
PoJ's players
[PoJ]_Tokiwa_
[PoJ]_Chocobo_
[PoJ]dodonpa
[PoJ]_Escape
poZ's players
poZ_Tammy
(hong kong) poZ_Cloud
(hong kong) poZ_ratS
poZ_Tobi
Commentary:
PoJ:
Power of Japan is a really old team formed by expert TG players. [PoJ]_Tokiwa_, [PoJ]_Chocobo_, [PoJ]dodonpa , [PoJ]_Escape have managed to defeat CKF, 4 – 1, and thanks to that result they have taken the first place of their group. Often underestimate, PoJ’s players have an incredible team play, and had good results in the past tournaments: they’ve taken 5th place in TMW and almost reached the final stage of the Nation Cup 2014 after being defeated 4 – 5 by Canada A. A serious contender for getting the qualification thanks to the loser bracket!
(hong kong)poZ:
poZ is the (hong kong) Hong Kong China Team formed principally by (hong kong) poZ_Tammy, (hong kong) poZ_Cloud, (hong kong) poZ_ratS, and poZ_tobi. A Historical team which has been present for many tournaments, they have really good team play. However, so far they've got really disappointing results in WiC, including a defeat of 5 – 0 to the hands of Nadya’s Boys. Considering that horrible defeat, putting them up against a team such as PoJ will be tough, given that PoJ is considered to be a close match to the NaBs. This will probably be one of the closest matches of the week! If poZ’s players want to win it, they will have to start training much harder and more often!
Power Rankings:
PoJ: 8
poZ: 11
This concludes our current post. Stay tuned for upcoming game predictions, as well as other possible ideas! If anyone would like to include some positive feedback for the statistical analysis that we can include later, please let us know! Also, feel free to discuss any and all parts of our analysis in the above thread. And as always, thank you to the sponsor and everyone else involved in this great tournament 8-)
First Phase: Data Gathering.
In deciding on a statistical model that can accurately predict anything you may be interested in, you first need to make sure you have really good data that your models can be based off of. That being said, TheGovernor and I first started talking about the variables that we can get, and that might be interesting in predicting the games: 1v1 Ratings, team game ratings for each of the teams was the first obvious piece of data we collected. We decided to take that even further by adding "number of games played in the past month", which required going to each and every player's profile, and counting the number of games they have played for the entire month prior to their WiC game. As a subset of this variable, we also made the variable " least games played" and "most games played" which correspond to the player on that team who has played the most in the past month, and the least in the past month. We then decided that even though 1v1 and TG ratings were very essential, we needed to include something of a "difference between rates" for the teams that were playing each other, which we dubbed "1v1 rate delta" and "TG rate Delta". Finally, we recorded 3 variables that correspond to the games played in each series: a binary code that said "1-0-0-0" if a team won the game, "1-1-0-0" if a team won the game, as well as the previous game, "1-1-1-0" if the team won the game, the previous game, and the previous 2 games in a row, and finally "1-1-1-1" which corresponds to winning the game, winning the previous game, winning the previous 2 games in a row, and winning the previous 3 games in a row.
Phase two: Exploratory Data Analysis
Part 1: Clustering Algorithms
Taking this datamatrix that consists of 11 predictive variables, all attempting to predict "1" or "0" (win or lose), I first began by converting the data into a Variance-Covariance matrix. From there, I decided I wanted to use what is known as a "Complete-Linkage Clustering Algorithm", a simplistic page of how it works can be seen here: http://en.wikipedia.org/wiki/Complete-l ... clustering
In a mathematical sense, this algorithm is fairly simple, and attempts to find patterns on how "similar" each team is from each other. Here is the output of the Algorithm:
How does one go about interpreting this output? The "Height" on the left corresponds to how similar two teams are that are in the same branch. So in this instance, we can see that for the Complete Linkage Algo, it found that Tyrant Warlords and Tyrant Legends were very similar, while VN,PoZ, and PL were all very similar. Likewise, it found that SY-B, PL, poZ, VN, nC, and VNS are all weakly similar (in the same overall branch) while SY, NaBs, L_Clan, Str8, KGB, poJ, zZ, Tyrant L and Tyrant W are all weakly similar as well.
After using the Complete Linkage Clustering Algo, I decided to see what an "Average LInkage Clustering Algo" would produce, and here is the output:
The results from the Average Linkage, despite being fairly similar to the Complete Linkage, had some differences. For instance, now only the Tyrant Teams, SY, NaBs and L_Clan are considered to be weakly similar (same overall branch) while the remaining teams are all on a seperate branch. However, it still lists the two TyRanT teams as being very similar, as well as NaBs with L_Clan, as was indicated by the complete linkage algo. Please keep in mind that these algorithms are not biased by me in ANY WAY, and that interpreting these graphs is fairly arbitrary. I would love to see an open discussion of these images (as well as discussion of all further analysis in this post) from everyone reading this!
Part 2: Discriminant Analysis
Moving away from our simple clustering algorithms, I then created a Linear Discriminant Analysis of our dataset. Once again, for anyone who may be interested in how it works, a simple run through of the basics of it can be seen here: http://en.wikipedia.org/wiki/Linear_dis ... t_analysis
Some of the most significant variables in Graph-Matrix form can be viewed here:
From this we can see how the algorithm is choosing to divide each pair of our predictive variables (such as Average 1v1 Rate, Average TG rate, etc) to find the most optimal method of depicting if a team should win or lose. Some of these come out to be pretty obvious: for example, if you look at the Average TG - Average 1v1 rate pair (first graph down from the top left), we can see that a team is predicted to win "w" if the Average TG rate added to the Average 1v1 Rate is above the threshold going from the cyan color (bottom left) to the green color (top right). The little w's and l's you see in there are real data points that have occurred in WiC so far.
Part 3: Hidden Factor Analysis
In many problems in modern science, in particular the social sciences such as Economics, Financial markets and psychology research, researchers have been interested in quantifying the "hidden factors" that might be driving a data set, but are not quantifiable (there is no explicit data for it). In essence, I utilized something known as the spectral decomposition of the variance-covariance matrix, combined with a "varimax rotation" to find linear combinations of our predictor variables in an attempt to quantify our hidden variables. Due to the binary nature of part of our dataset, I was forced to only consider the variables Average 1v1 Rate, Average TG Rate, Least Games Played in past month, Most games played in past month, Average Games played in past month, 1v1 Rate Delta, and TG Rate Delta. The output is the following:
As one can see, the first three components combine for roughly 81.7% of the total variation in the data, making 3 "hidden" factors seem appropriate for our dataset. The first factor puts a heavy emphasis on the average 1v1/TG rates (.637 and .881, respectively) as well as the 1v1/TG rate deltas(.674,.872), indicating the first hidden factor is simply the rating strength of the team relative to it's opponents team.
The second factor puts a very high emphasis on Average Games Played in the 1 month prior to game, as well as putting a heavy emphasis on the Low Games played in 1 month prior (.778 and .962 respectively). It seems to also add in a small amount of weight to Average team game, average 1v1 rate and TG delta rate. In my opinion, this hidden factor has something to do with the team's teamwork, or how much they are playing.
Finally, the third hidden factor puts a whopping .994 on Highgames, .483 on Average Games, .312 on TG rate Delta, .512 on Average TG rate, and a surprising -.149 on 1v1 rate. This indicates a variable that puts a high emphasis on team game rates, and even more emphasis on the player on the team who is playing the most, while at the same time discounting the 1v1 rate. Therefore, I want to say that this variable is something of a leadership variable, due to the high emphasis on HighGames variable.
*Please keep in mind, that my interpretation of each hidden factor is completely arbitrary, and is open to any and all discussion!
Phase 3: Statistical Modelling
Now here comes the fun (and hard) part of all this analysis. I spent a fair chunk of time attempting to work on several different statistical modelling techniques that could fit to this situation, and eventually came up with a model that seems to work out pretty well! After finding a good model, I broke it into two pieces: one model is used for power rankings, while the second model is used for game-by-game predictions. Instead of posting in-depth results of the models here, you can see the power rankings in the upcoming pieces of this post, conducted by _TheGovernor_!
Team Previews and power rankings:
VNS (vs) TyRanT Warlords
Historic of the Teams in WiC:
VNS's historic
Group Stage (Group 6)
VNS 5 - 0 NoStrat
VNS 5 - 0 Arg A
VNS 4 - 1 FOX Gaming
KGB 4 - 1 VNS
RM Stats: 12 Wins - 4 Defeats
DM Stats: 3 Wins - 1 Defeat
VNS 5 - 0 NoStrat
VNS 5 - 0 Arg A
VNS 4 - 1 FOX Gaming
KGB 4 - 1 VNS
RM Stats: 12 Wins - 4 Defeats
DM Stats: 3 Wins - 1 Defeat
TyRanT Warlord's historic
Group Stage (Group 2)
TyRanT Warlords 5 - 0 RoaMer_
TyRanT Warlords 4 - 1 SY B
TyRanT Warlords 5 - 0 Kotowari
TyRanT Warlords 5 - 0 Titanic
RM Stats: 15 Wins - 1 Defeat
DM Stats: 4 Wins - 0 Defeat
TyRanT Warlords 5 - 0 RoaMer_
TyRanT Warlords 4 - 1 SY B
TyRanT Warlords 5 - 0 Kotowari
TyRanT Warlords 5 - 0 Titanic
RM Stats: 15 Wins - 1 Defeat
DM Stats: 4 Wins - 0 Defeat
Prediction
VNS's players
VNS_ACCM
VNS_yellow
VNS_SeOul
VNS_Fish
TyRanT Warlords' players
RiuT
F1Re
BacT
Slam
Commentary:
VNS:
VNS has lost a lot when BacT left them, even if they’re still such a strong team, they can’t compete at the very top level anymore. ACCM, yellow, SeOul and Fish seem to be their best team players composition. They are 1850, in both 1v1 and TG Rating, and they managed to take one game from KGB in the final of their group. That wasn’t enough to take the first place though, and they got one of the worst matches up possible: TyRanT Warlords! Then the've defeated VN (4 – 1) who tried to duel them. They’ll have to increase their level if they want to defeat TyRanT Warlords, but that seems a bit complicated. In case of a defeat, their last hope will be to reach the semi-finals of the loser brackets, which will be rather hard but not impossible: really good and close matches might come!
TyRanT Warlords
TyRanT Warlords is the second TyRanT team. Even if Legends’ players look a bit stronger on the paper, Warlords are still an important threat. Slam and RiuT are probably the best players of the team. Slam is always active, playing a lot every weeks and he seems in a really good shape. RiuT is playing a bit less nowadays, but he hasn't lost his skill at all, regarding to his recent 1v1 against Chris who he defeated 3 – 1. F1Re, seems to be the weaker member of TyRanT, despite he’s playing really well his post-imperial games. The games which are coming will give him a chance to show that he is a true TyRanT warrior! BacT is really trained, he got about 1500 TG games, in the ladder, and is in the top 5. This team could get impressive result, especially since we can consider two duos, on a side RiuT & F1Re, who’re use to play together, and on the other slam and BacT, who played the recent 2v2 tourney together.
Power Rankings:
TyRanT Warlords: 1
VNS A: 12
SY (vs) VN
Historic of the Teams in WiC:
Vikingos Nomadas's historic
Group Stage (Group 4)
VN 5 - 0 Loyal Knights of the Cross
VN 3 - 2 VNS B
VN 4 - 1 The Kebab
Legion Clan 4 - 1 VN
RM Stats: 10 - 6
DM Stats: 3 - 1
VN 5 - 0 Loyal Knights of the Cross
VN 3 - 2 VNS B
VN 4 - 1 The Kebab
Legion Clan 4 - 1 VN
RM Stats: 10 - 6
DM Stats: 3 - 1
SY's historic
Group Stage (Group 3)
SY 5 - 0 Don't lame us pls
SY 3 - 2 Happy Fun BuzZ Family (2 AW).
SY 5 - 0 Supermen
SY 4 - 1 PL
RM Stats: 14 Wins - 0 Defeat - 2 Admin Win
DM Stats: 3 Wins - 1 Defeat.
SY 5 - 0 Don't lame us pls
SY 3 - 2 Happy Fun BuzZ Family (2 AW).
SY 5 - 0 Supermen
SY 4 - 1 PL
RM Stats: 14 Wins - 0 Defeat - 2 Admin Win
DM Stats: 3 Wins - 1 Defeat.
VN's players
VN_Rob3rStinson (TaToH)
VN_DarK_KnighT_
VN_MoOniTe_
VN_HEIMDALL
SY's players
Yo__
FengXin (Paladin)
NO_23
james__
Air_Abu_ (nomad games).
Commentary:
VN:
VN_Rob3rStinson (TaToH) is the main player of VN which is also composed by VN_DarK_KnighT_, VN_MoOniTe_, VN_HEIMDALL, 3 players who already have some experience in tournaments, but they didn’t often reach the final stages tho. With the arrival of TaToH who’s carrying the team pretty hard, they managed, rather luckily since they had the same number of points than VNS B to reach the double stage elimination. Then they lost their duel against VNS (4 – 1) which means they must play against SY. There’s probably no way to win this, so they’ll have to comeback through the Loser Brackets. They will have to count on a really good TaToH!
SY
SY, is the #1 Chinese team. Composed by Yo__, Paladin, james__, Air_Abu_ and NO_23 they are the winner of the last tournament game, the Nation Cup 2014. The main strong of SY is not the individual skill of their players, even though they’re pretty impressive, but their coordination, and their decision making. This is definitely a team which is made to play teamgames! They were invincible during the Nation Cup, even though they played against some strong teams like Brazil A and B and also Argentina A. That’s why only a few teams may stop them, the TyRanTs and perhaps the L_Clan or the KGB. They’re also one of the two teams which hadn’t been defeated in RM during the group stage (although they gave 2 games to BuzZ ‘cause they didn’t want to play them).
Power Rankings
SY: 2
VN: 10
Legion Clan (vs) CKF
Historic of the Teams in WiC:
Legion Clan's historic
Group Stage (Group 4)
L_Clan 5 - 0 TheKebab
L_Clan 4 - 1 VNS B
L_Clan 5 - 0 Loyal Knights of the Cross
L_Clan 4 - 1 VN
RM Stats: 16 Wins - 2 Defeats
DM Stats: 4 Wins - 0 Defeat
L_Clan 5 - 0 TheKebab
L_Clan 4 - 1 VNS B
L_Clan 5 - 0 Loyal Knights of the Cross
L_Clan 4 - 1 VN
RM Stats: 16 Wins - 2 Defeats
DM Stats: 4 Wins - 0 Defeat
CKF's historic
Group Stage (Group 8)
CKF 5 - 0 Master Clan League
CKF 5 - 0 Arg B
CKF 5 - 0 White Wizards (czech)
PoJ 4 - 1 CKF
RM Stats: 13 Wins - 3 Defeats
DM Stats: 3 Wins - 1 Defeat
CKF 5 - 0 Master Clan League
CKF 5 - 0 Arg B
CKF 5 - 0 White Wizards (czech)
PoJ 4 - 1 CKF
RM Stats: 13 Wins - 3 Defeats
DM Stats: 3 Wins - 1 Defeat
Legion Clan's players
dogao
Chris
FeAge
Miguel
CKF's players
[CKF]geoffrey
[CKF]Tim_
[CKF]_PBT_Dee
[CKF]IamOK
Legion Clan:
L_Clan is probably one the strongest team of WiC. Mainly composed by 4 expert players, Dogao, Chris, FeAge, Miguel, they’re training really hard every week-end. Dogao and Chris are probably two of the best TG players of the moment. Currently, Chris hold the #1 place of the TG Ladder, and Dogao and FeAge are in the top 12. FeAge is also a very impressive player, who’ve been improving lately, and he’s definitely a good number 3. As for Miguel, we've seen him doing brilliant games, but also less good ones. He's a good #4, but he's gonna have to be on a really good day if he wants to face the strongest teams of the tournament. Regarding to a lot of experts, Legion Clan is a serious top4 challenger. During their training games, L_Clan has managed to put down many teams, like NaBs and zZ.
CKF:
CKF is probably the second best Chinese team. Everybody is expecting great results from them, mainly due to the comeback of Tim_. However, they’ve been defeated quiet rudely by PoJ in the Group Stages. That put the CKF to the second place and give them a really bad match-up, against L_Clan. It’s a bit surprising that CKF didn’t play with Tim the whole RM games, and even more surprising that Lyx, still didn’t play a game despite he’s one of the greatest player of this team. Geoffrey and IamOK, are two really impressive players who’re both experimented, have a good team play and are really strong. CKF will have to train a bit more, and perhaps to get another team composition if they want to reach the top 8 of the tournament.
Please note, CKF was not added to the model due to model fitting issues. Therefore, they are the only team who do not have a power ranking
Power Rankings
L_Clan: 3
CKF: ?
Nadya's Boys (vs) PL
Historic of the Teams in WiC:
Nadya's Boys:
Group Stage (Group 7)
Nadyas Boys 5 - 0 Cosmopolitan Super Kings
Nadyas Boys 5 - 0 Naughty rubbishes
Nadyas Boys 5 - 0 TILT
Nadyas Boys 5 - 0 Poison of Zone (hong kong)
RM Stats: 16 Wins - 0 Defeat.
DM Stats: 4 Wins - 0 Defeat.
Nadyas Boys 5 - 0 Cosmopolitan Super Kings
Nadyas Boys 5 - 0 Naughty rubbishes
Nadyas Boys 5 - 0 TILT
Nadyas Boys 5 - 0 Poison of Zone (hong kong)
RM Stats: 16 Wins - 0 Defeat.
DM Stats: 4 Wins - 0 Defeat.
PL's historic
Group Stage (Group 2)
PL 4 - 1 Don't lame us pls
PL 4 - 1 Supermen
PL 5 - 0 BuzZ
SY 4 - 1 PL
PL 4 - 1 Don't lame us pls
PL 4 - 1 Supermen
PL 5 - 0 BuzZ
SY 4 - 1 PL
Nadya's Boys' players:
1st Tier:
_8th_Wonder_
Zuppi
(czech) Lojza_
mentalist_
2nd Tier:
NorthenPike
Spaden
Lavie_Head
PL's players
PL_Remi
PL_Buniek
PL_Dziamdziak
PL_Bity_kablem
Commentary:
Nadya's Boys:
Nadya’s Boys is probably one of the most well-known team of WiC. Due to Pike’s stream and lately Nadya’s stream many players and spectators had the pleasure to see them playing a long time before WiC. They’re definitely the most serious team of the tournament regarding to the number of hours they're using every day to train. It’s pretty rare that a day passed, without any games played from Nadya’s Boys. Another thing that makes the strength of the team is the fact that they got many members. According to Nadya and a few experts, we can basically separate this team into two tiers. First one would be composed by (czech) Lojza_, Zuppi, mentalist_, _8th_wonder_. (czech) Lojza_, recently won the final of The Arabian Conquest against [TyRanT]BacT_, mentalist_ reached the 7th place with Canada in the Nation Cup 2014. He’s probably one of the most impressive players of the moment. He reached 2k one year ago, but he’s now one of the better TG Players of the moment. As for the 2nd tier, we can put in: NorthenPike, Lavie_Head, Spaden. NaBs are mainly training against zZ, L_Clan, and got pretty good results! They also made a really good impression against poZ during the group stage that they defeated 5 – 0. True, poZ aren't really training for the tournament, but this is still an impressive performance!
PL:
The Polish team can be considered something of an underdog of the tournament. Not usually a conversation starter, PL managed to defeat all but SY in their group, having struggled vs. them in their matchup by going 4-1. Despite having low expectations for this team, they pride themselves in teamwork due to having teamed together for a long period of time (including in Nations Cup tournament).
Power Ranking:
NaBs: 4
PL: 15
Historic of the Teams in WiC
TyRanT 's historic:
Group Stage (Group 1)
TyRanT Legends 4 - 1 Czech4ever (czech)
TyRanT Legends 5 - 0 Gorlami
TyRanT Legends 5 - 0 THE ORIGINAL TACO TEAM
TyRanT Legends 5 - 0 nC & friends & arma
RM Stats: 15 Wins - 1 Defeats.
DM Stats: 4 Wins - 0 Defeat.
TyRanT Legends 4 - 1 Czech4ever (czech)
TyRanT Legends 5 - 0 Gorlami
TyRanT Legends 5 - 0 THE ORIGINAL TACO TEAM
TyRanT Legends 5 - 0 nC & friends & arma
RM Stats: 15 Wins - 1 Defeats.
DM Stats: 4 Wins - 0 Defeat.
Str8's historic:
Group Stage (Group 5)
STr8 5 - 0 North Crew
STr8 5 - 0 That's Amore
STr8 5 - 0 Oversea Chinese Clan
Dreamers 3 - 1 STr8 (Last DM game hasn't been played).
RM Stats: 13 Wins - 3 Defeats
DM Stats: 3 Wins - 0 Defeat.
STr8 5 - 0 North Crew
STr8 5 - 0 That's Amore
STr8 5 - 0 Oversea Chinese Clan
Dreamers 3 - 1 STr8 (Last DM game hasn't been played).
RM Stats: 13 Wins - 3 Defeats
DM Stats: 3 Wins - 0 Defeat.
TyRanT Legends' players
DauT
TheViper
JorDan_23
kkab
Str8's players
Emin (FoxValentine)
Jugband
Artras
(czech) Skull
Commentary:
TyRanT Legends are favorite beyond any doubts in this match-up. They've lost only game in the group stage, against Cz. Back to that time, some of the TyRanT players looked a bit rusty. DauT played many games since this match and trained pretty hard, during the last month. Indeed, he played about 70 1v1 against really good players like Spaden (3-0), Monoz (3-0), _Ra_ (7-4), FeAge (2-1), Edie (1-1), TaToH (5-2), Miguel (2-3), Yo__ (6-6), TheViper (0-5) and IamOK (4-0). Thanks to these games, he managed to come back to the top 10 of the 1v1 ladder. Back to the overall team, we still have no idea who NeferTiTi, the last player TyRanT Legends is. However, he did not play the most important match of TyRanT in the group stage, against Nc. The games were pretty impressive and TyRanT won 5-0, which means that DauT, TheViper, JorDan_23 and kkab will most likely play this stage!
On the other hand, Str8 are a talented team (2k average TG and 1v1) composed mainly by Emin, Jugband, Artras, (czech) Skull and Sletten. It's almost the same team which played in the Nation Cup 2014, where they managed to reach the knock out stage. They've been defeated then by the SY (4-0) but they fought pretty well which means TyRanT won't easily win. Emin is getting better and better and he's definitely reaching an impressive level. He managed to take some 1v1 games against TheViper, and Chris a few months ago. Talking about competitions, he reaches the 4th place in the 1v1 tournament, the Arabian Conquest, after being defeated 3-4 against (czech) Lojza_ the winner of the tournament! In WiC, Str8's players have done a really good job in the group stage by winning 5-0 every match up they had, except the one against zZ, where they've been defeated 3-1.
Power Rankings
TyRanT Legends: 5
STr8: 9
*Please note that TyRanT Legends does not fit our statistical models very well. This is in large part due to many of their games are played in secret non-rated games, as well as kkab having a 1600 1v1 rate, and only 1800's TG rate, with very few games played (rated). Therefore, the model has a hard time correctly fitting (predicting) the outcomes of TyRanT Legends games.
Dreamers (vs) nC
Historic of the Teams in WiC:
zZ's historic
Group Stage (Group 5)
Dreamers 5 - 0 That's Armore (czech)
Dreamers 5 - 0 Team Boobs
Dreamers 4 - 1 Oversea Chinese Clan
Dreamers 3 - 1 Str8
RM Stats: 14 Wins - 2 Defeats.
DM Stats: 3 Wins - 0 Defeat.
Dreamers 5 - 0 That's Armore (czech)
Dreamers 5 - 0 Team Boobs
Dreamers 4 - 1 Oversea Chinese Clan
Dreamers 3 - 1 Str8
RM Stats: 14 Wins - 2 Defeats.
DM Stats: 3 Wins - 0 Defeat.
nC's historic
Group Stage (Group 1)
nC 4 - 1 Czech4ever (czech)
nC 5 - 0 Gorlami
nC 4 - 1 The Original Taco Team
TyRanT Legends 0 - 5 nC .
RM Stats: 10 Wins - 6 Defeats
DM Stats: 3 Wins - 1 Defeat.
nC 4 - 1 Czech4ever (czech)
nC 5 - 0 Gorlami
nC 4 - 1 The Original Taco Team
TyRanT Legends 0 - 5 nC .
RM Stats: 10 Wins - 6 Defeats
DM Stats: 3 Wins - 1 Defeat.
Prediction
zZ's players
Nicov
Afrodita
Gary
PaTiN
Fedex
nC's players
Skittle
Andorin
Tsu_NilPferD
JohNNieW_aLc_eR (Arma)
Commentary:
zZ:
Dreamers team has kinda the same composition than in SY Nation Cup 2014 where they reached the 3rd position, after really tough matches against the Brazil A which have finished second. That’s why nobody should underestimate this team. Nicov is probably the most important player of the team at the moment. He has definitely improved pretty well those last months and years, and he won some important 1v1 against famous players few days ago like Yo__ (5-3), F1Re (7-1), mentalist (3-0)... zZ is also really strong thanks to their hard training. Indeed most of the players have played more than 75 games this last month and most of them are team game, played with other members of zZ. They’re mostly playing against NaBs and L_Clan.
nC:
nC is probably one of the weakest team in this double elimination stage. The same team reached with some difficulty the knock out stage of the Nation Cup, but that was with JorDan_23. The nC is mainly composed by Skittle, Andorin, Tsu_NilPferD, JohNNieW_aLc_eR (Arma), strong players who got about 1k9 TG Rating. That said, nC had some troubles to be qualified during the group stage. They've been for only 1 point, and they must now face the Dreamers (zZ) who’re looking stronger than them. Arma and Skittle will have to carry the team if they want to go any further in the competition. But in order to do that they’ll have to train a bit more than a few games per months...
Power Rankings
Zz: 6
nC: 13
KGB (vs) SY B
Historic of the Teams in WiC:
KGB's historic
Group Stage (Group 6)
KGB 5 - 0 Arg
KGB 5 - 0 Fox Gaming
KGB 5 - 0 No Strat
KGB 4 - 1 VNS
RM Stats: 15 Wins - 1 Defeat
DM Stats: 4 Wins - 0 Defeat
KGB 5 - 0 Arg
KGB 5 - 0 Fox Gaming
KGB 5 - 0 No Strat
KGB 4 - 1 VNS
RM Stats: 15 Wins - 1 Defeat
DM Stats: 4 Wins - 0 Defeat
SY B:
Group Stage (Group 8)
SY B 5 - 0 Kotowari
TyRanT Warlords 4 - 1 SY B
SY B 4 - 1 Titanic
SY B 5 - 0 RoAmEr_
KGB's players
Alive
GoKu
St4rk
Ra
Hasan
SY-B Players
pity_memory
xiaoxiong
_VG_Lang
Cigarettes__
Yu___
RM Stats: 12 Wins - 3 Defeats
DM Stats: 2 Wins - 2 Defeats
SY B 5 - 0 Kotowari
TyRanT Warlords 4 - 1 SY B
SY B 4 - 1 Titanic
SY B 5 - 0 RoAmEr_
KGB's players
Alive
GoKu
St4rk
Ra
Hasan
SY-B Players
pity_memory
xiaoxiong
_VG_Lang
Cigarettes__
Yu___
RM Stats: 12 Wins - 3 Defeats
DM Stats: 2 Wins - 2 Defeats
KGB:
KGB is really interesting team, formed very recently. Alive, GoKu, St4rk, Ra, Hasan, and Whack are the main players. So basically we got 2 members of Brazil A, one member of Brazil B, one member of Canada and finally, two members of USA A, 4 teams that have got (very) good results during the Nation Cup. Now, who is gonna play the final Stage? Even if we of course can’t know that, it seems that the third Brazilians have been really active during the Group Stage, so they might be preferred. Hasan looks like a good number 4, even though _Ra_ has played a lot those last months, and became really strong. On the paper KGB looks really strong, but they’ll need to train way more in order to defeat the best teams of the tournament. If the first match against SY B will be difficult, it will be nothing in comparison of the second match against the Legion Clan, that’ll be a good indicator of the strength of KGB.
SY B
Despite it is supposed to be a B team, SY B are actually really strong. Their only defeat was against TyRanT Warlords, vs who they managed to take a game. They’ve dominated quiet easily the Titanic team, and reached the second place of their group. The main players of the team are Yu____, pity_memory, xiaoxiong, and Ds___. One of the best advantages for SY B is probably that there are rather unknown; they’re mostly training in Voobly China and can create some surprises, especially on Nomad Games, where they look pretty strong!
Power Rankings:
KGB: 7
SY-B: 14
PoJ (vs) poZ (hong kong)
Historic of the Teams in WiC:
Power of Japan's historic
Group Stage (Group 8)
PoJ 4 - 1 White Wizards (czech)
PoJ 5 - 0 Master Clan League
PoJ 5 - 0 Arg B
PoJ 4 - 1 Chinese Kung Fu
RM Stats: 14 - 2
DM Stats: 4 - 0
PoJ 4 - 1 White Wizards (czech)
PoJ 5 - 0 Master Clan League
PoJ 5 - 0 Arg B
PoJ 4 - 1 Chinese Kung Fu
RM Stats: 14 - 2
DM Stats: 4 - 0
Poison of Zone's historic
Group Stage (Group 7)
(hong kong) poZ 4 - 1 TILT
(hong kong) poZ 5 - 0 Naughty Rubbishes
(hong kong) poZ 5 - 0 Cosmopolitans Super Knights
Nadya's Boys 4 - 1 poZ (hong kong)
RM Stats: 13 - 3
DM Stats: 2 - 2
(hong kong) poZ 4 - 1 TILT
(hong kong) poZ 5 - 0 Naughty Rubbishes
(hong kong) poZ 5 - 0 Cosmopolitans Super Knights
Nadya's Boys 4 - 1 poZ (hong kong)
RM Stats: 13 - 3
DM Stats: 2 - 2
Prediction
PoJ's players
[PoJ]_Tokiwa_
[PoJ]_Chocobo_
[PoJ]dodonpa
[PoJ]_Escape
poZ's players
poZ_Tammy
(hong kong) poZ_Cloud
(hong kong) poZ_ratS
poZ_Tobi
Commentary:
PoJ:
Power of Japan is a really old team formed by expert TG players. [PoJ]_Tokiwa_, [PoJ]_Chocobo_, [PoJ]dodonpa , [PoJ]_Escape have managed to defeat CKF, 4 – 1, and thanks to that result they have taken the first place of their group. Often underestimate, PoJ’s players have an incredible team play, and had good results in the past tournaments: they’ve taken 5th place in TMW and almost reached the final stage of the Nation Cup 2014 after being defeated 4 – 5 by Canada A. A serious contender for getting the qualification thanks to the loser bracket!
(hong kong)poZ:
poZ is the (hong kong) Hong Kong China Team formed principally by (hong kong) poZ_Tammy, (hong kong) poZ_Cloud, (hong kong) poZ_ratS, and poZ_tobi. A Historical team which has been present for many tournaments, they have really good team play. However, so far they've got really disappointing results in WiC, including a defeat of 5 – 0 to the hands of Nadya’s Boys. Considering that horrible defeat, putting them up against a team such as PoJ will be tough, given that PoJ is considered to be a close match to the NaBs. This will probably be one of the closest matches of the week! If poZ’s players want to win it, they will have to start training much harder and more often!
Power Rankings:
PoJ: 8
poZ: 11
This concludes our current post. Stay tuned for upcoming game predictions, as well as other possible ideas! If anyone would like to include some positive feedback for the statistical analysis that we can include later, please let us know! Also, feel free to discuss any and all parts of our analysis in the above thread. And as always, thank you to the sponsor and everyone else involved in this great tournament 8-)