GDI in March 2019 – what do the top players play?

Inspired by a couple of similar exercises in the Rivals subreddit in recent months, I analyzed the Top 25 GDI players (by Medal count) and what decks they play. I took the deck they played the most this season, and when there were two with similar amount, I took the ones with higher winrate.

To add some sophistication, I took into account win rates and unit levels. With these points, I created two sub-segments of 10 players:

1. "Low Level 10": The 10 players with low unit levels in the top 20 (L10-11.5). They still have an impressive 71% win rate at this high level. This group includes players such as HayprO, Lex[Xen], VictorBansemer and Nick, who all are in the Top 25 with decks lower than L11.
2. "Strong 10": With a compound of winrate and unit levels, I scored deck strength. This e.g. puts Lex[Xen] in the top tier together with Bikerush and Alicia, as he manages his 76% winrate with a L10 deck.

Let's look at the data below. The % show the amount of decks the unit is in.

For people who play in the highest leagues, this will not be surprising:

• Tech is played very little, and half of Tech usage is down to the Kodiak, commonly used for Jackson-boosted Base rushes.
• Pitbull, Shock Troopers and Jumpjet Troopers (JJ in the table above) are played the most. 12 of 25 decks contain all three of these units.
• GDI air has become very viable with the additions of Mohawk and Orca Bomber (which in turn also make Hammerhead more attractive)
• Rhinos don't see much play, likely due to a mix of Dogs still being viable and GDI decks containing more Air (so there are fewer decks with 3 factory units). 
• Despite voices complaining that the Predator Tank would be too strong, it does not see that much use. I expect to say way more Scorpion use for Nod. This shows that you cannot compare units just on their 1:1 (which the Predator wins), but you need to see them in the context of the full range of units and roles these units play.
• The new Orca Bomber already sees quite some usage. If it's indeed viable at top level, I expect it to be very strong in lower leagues, where double harvester is more common and where players likely will find it harder to counter the Orca Bomber with ground in emergency situations.

If we look at the how "Strong" als "Low Level" decks deviate:

• Kodiak seems more viable if you don't have a unit level disadvantage. This makes sense, as the Kodiak adds option value of base destruction for sacrificing a unit slot. Low level decks are reliant on being able to hard counter opponent's units, hence cannot afford as well to sacrifice a basic unit. 
• This doesn't mean that high-level decks can play Kodiak just due to their levels. It might well be that the Kodiak option is powerful enough that it reduces these players' variance – 4 of the 5 Kodiak decks sport incredible winrates of 95%-96%. Of course, this is also down to unit levels (all Kodiak decks have L12+), and the skill of the players.
• Missile Troopers are more prevalent in the Low Level decks. Possibly because they are cost-efficient hard counters to a number of key units (e.g. Tanks, Bikes, Pitbulls, Banshees, Mohawks) I expect this to be the same for Nod.

Observations that are not directly visible in the table:

• 23 of 25 decks have 10-20 Tib openers. The exceptions are bhman with his infamous expensive Juggernaut deck, and Alicia, who is one of the few people who like the Rhino opener.
• Only the 4 high-winrate Kodiak decks chose to not use any Heliport unit. From the other 21 decks, 19 have Talon or Hammerhead (the two other using Orca solo).
• MLRS has the highest winrate at 96%, but this is likely because it's used by Bikerush and Skelturix and not so much because of the unit.
• The two Drone players, SOFTSERVE and Lex[Xen] seem to know what they are doing: They catapult the Drone to a 79% winrate, the highest average for a Heliport unit.

Number of units by building in the average deck:

The award for the most unusual deck goes to Lex[Xen], who sports a deck with Missile, JJ, Drone, HH, Borca and Mammoth Tank. Especially his 76% winrate with a deck that is barely L10 is impressive:

If you want to look at the raw data & the specific decks of the top 25 players, check this spreadsheet.

Stay tuned for the Nod version of this, and let me know if you have any follow-up questions on the above, or wishes for the Nod analysis.

Impact of Unit Levels on Winrate

Unsurprisingly, there is a correlation between higher unit levels and higher winrate. But within a difference of a level, the variance is very high, which means that skill plays a huge role as long as the level difference is not huge.

A caveat on this: Winrate is not a perfect metric. For example, if, highly skilled players play their L10 deck in the Champions event with a L10 cap, they can easily get 90%+ winrate there even if they have has a 65-70% winrate on ladder. Thus, already the mix of games between ladder and Champions can skew winrates significantly.

X-axis: Winrate / Y-axis: Unit Levels (rounded to .5s for simplicity);

Red is a polynomial trendline for illustration purposes.