The International 2018: After the Group Stage, Before the Main Event

The Group Stage of The International 2018 is over, with eight teams entering the Upper Bracket of the Main Event, eight dropping into the Lower Bracket, and two going home unheralded.

On July 2nd, I predicted the top eight teams. How did they do?

Of my top eight predictions (Liquid, LGD, Virtus Pro, Secret, VGJ.Storm, Vici, TNC, and Optic), six are in the Upper Bracket. The failures: my sixth pick, Vici Gaming, who, in spite of their recent decline still seemed a formidable opponent, and my 7th pick, TNC Predator.

The International 8's story though, isn't about failures, but the surprise upsets. Two teams superseded any reasonable expectation. The newly remade Evil Geniuses and OG strove for better seedings than any reasonable person could have assumed. Both teams, after the ESL Birmingham Major, realized they had no chance at The International Aegis with their tepid results. Neither was going to earn enough DPC points for a direct invite. But their roster changing reactions were astonishing and devastating. EG dropped two players (Fear and Misery), and poached OG's stars S4 and original co-founder of OG, Fly. OG responded by recruiting well known players Ana and Topson to fill their empty slots.
With only one more event before the Open Qualifiers, Evil Geniuses tried to demonstrate a new proficiency at the China Supermajor, but they earned only a 9th-12th place finish, knocked out by aspiring North American superpower, Optic Gaming. OG, already invited to the Supermajor, withdrew to focus their attention on the Qualifiers. Despite these unexpected alterations, both teams distinguished themselves in the Open and Regional Qualifiers, where OG went 14-2 in Europe, securing themselves the only slot, and EG narrowly avoided elimination, earning the 2nd North American invite with a 8-3 result over Immortals.

After a poor performance at the Supermajor, and a self disqualification, the question remained. How would these newly reconstituted squads perform against the best teams in the world? In the initial predictions I placed them 12 (OG) and 15 (EG), but both with a caveat. If the teams, stocked with star power, could synergize, they might be unstoppable. The results of the Group Stage prove both teams have coalesced well, offering a chance at a championship.
My new power rankings (change since TI8 Qualifiers)

Top Tier:

1) Team Liquid    (0)
2) LGD                (0)
3) Virtus Pro        (0)
4) Evil Geniuses (+11)

Upper-Mid Tier:

5) VGJ.Storm    (0)
6) Team Secret  (-1)
7) Vici Gaming  (-1)
8) OG                (+4)
9) Optic Gaming (-1)

Lower-Mid Tier:

10) TNC Predator (-3)
11) Fnatic              (-2)
12) Newbee           (-1)

Bottom Tier:

13) WinStrike       (-3)
14) Mineski          (-1)
15) VGJ.Thunder (+1)
16) Serenity         (+2)

Observations:

Team Liquid is in the top Tier, but they're performance this year has elevated them to a tier of their own. Not only did they lose only three games during the Group Stage, they have demonstrated for the entire year their excellence and consistency. Of the thirteen tournaments they participated in this season, they placed 4th or higher in all but three. Other teams rise and fall, but Liquid remains above them all.

One might be surprised at a few changes, or lack of. Even though Vici recorded only a 7-9 result in the Group Stage, they only drop one place in the rankings. Their loses to TNC, Optic, and Storm hurt, but their victory over the indomitable Virtus Pro steadied their strength. VGJ.Storm, in spite of winning their group with a strong 12-4 record, remained in 5th place, and behind Virtus Pro. Fact, two of Storm's four loses were to the Russian squad. And Optic dropped a spot even though they made top eight in the Group Stage. But they barely snuck into the Upper Bracket, needing a triple tiebreaker with Secret and Newbee, while both OG and EG vaulted to the top of the list.

All the other results seem self explanatory, though as a Universe fan, disappointing.

Below you'll find my Bracket for the Main event, in which all decisions were made by my power ranking. Unfortunately, sometimes my ranking says two teams are essentially equal, in which case I am forced to execute an executive decision.
The most difficult decision was the starting Upper Bracket game between Virtus Pro and LGD (And the Lower Bracket Round 5, where I have them meeting again). Also agonizing, was the LGD and VGJ.Storm game in the lower bracket. According to my model, LGD is significantly stronger, but Storm has been surging. I decided to follow the model. And finally, VP vs EG in the lower bracket final. Both are about the same strength at this moment, with VP only a sliver ahead. I think EG's recent performance might prophesy success (until they lose to Liquid in the final).

On a unrelated topic (but still Dota), the Fantasy Draft bothered me last year and though I should abandon the topic, I can't let it lie. For the Fantasy Draft, each day, the viewer picks 2 core players, 1 offlaner, and 2 supports. Depending on a number of criteria, each player earns fantasy points for each game. The problem is, players earn more points, the more games they play. A player in two games, will earn roughly double the value of a player in 1 game. During the group stage, teams play different numbers of games each day. Some teams play 2 the first day, while others play 3, or 1. This limits the viewer in their fantasy draft, not based on how good a player, but by the schedule. It severely restricts the players the viewer can choose.
The problem is eliminated on the Main Stage.  A player is awarded points for their best 2 games in a best of 3, or their best 3 games in the Grand Final best of 5.

Two lesser problems remain. The first day, there are 4 best of ones. All these players are not viable options because they only play one game. And the last day of the main event includes the lower bracket final and the grand final. Because one team will be playing in both series, they will play a minimum of 5 games (a best of three, and a best of five), while the team already at the Grand Final will only play the best of five.  The rules are unclear if these players only earn their best 3 games, or their 2 best of 3 and 3 best of 5.  If the later then the team at the Grand Finals can not possibly play more games than the lower bracket winner. Therefore a viewer can earn the most points by picking players from the team they think will win the lower bracket, regardless of their performance at the grand final. 

Ultimately, the fantasy draft is less about picking the best players, but picking the teams which will play the most games. If that's what Valve want's the system to be, that's fine, but if they want it to be about the best players, I have a solution.

Each day, the players earn their points, but they are divided by the number of games played. The viewer only scores the average value of their players for the day. It makes all the teams viable, everyday (except those not playing at all). It would allow many more options, and increase the viewer variability.
One last thought on the Fantasy Game. As I said earlier, Dota enables viewers to pick 2 cores, an offlane, and 2 supports. Unfortunately, anyone picking a position 4 support will not earn enough points. This is because, position 5 supports, place vision wards, and wards are worth too many points. Therefore, there is no reason to pick position 4 players, who rarely ward. One fifth of players are completely nonviable the entire tournament. Placing a ward should be worth less points, but that's not the only change I recommend. Placing a ward requires no skill. Placing it well does. Instead of receiving points for placing a ward, I have a few recommendations.

Points are awarded:
For a ward that is not dewarded, and lasts the entire time.
For how long a ward lasts, before it is dewarded.
For how long enemies are in vision of the ward.
For how many enemies are killed in vision of the ward.
Not for how many wards are placed, but how many are dewarded by a player.

As a person with zero programming skills, except for my time with Human Resource Machine, I don't know how difficult it would be to implement any of these options, but I believe any of them would be preferable to the current system.

To concluded, I'm interested about any reader's bracket. I've looked at Gosu.AI's and theirs shares some similarities with mine. Next Monday article will conclude the 2017/2018 Dota Season. After that, Awkward Mixture will begin an adventure into Darkness.

Recent:

Relevant:

Comments