GDI Top 50 -- Suzaku APC and very diverse Air Play

Every month, I analyze the decks of the Top 50 Nod and GDI players in Rivals. This is the October 2019 issue for GDI. You can find the Nod variant here.

• Please find the raw data here, incl. a full list of the decks including nicknames
• Average deck level went up by +0.1 to 13.9
• Deck composition see in the graphics below. For each unit type, the bar shows the distribution of number of units per deck.

Now to the main data, the unit frequency table. How to read the table below:

• "Frequency" is the share of Decks that run this unit or commander. First, the frequency in % over last months is shown. Then, a bar chart visualizes the June frequency, and a +/- number shows the change to the previous month. Lastly, there is a small line chart that shows the trend over the last months.
• "Δ vs average winrate" is how the winrate decks utilizing this unit/commander deviates from the average (e.g. Jumpjet decks had a 4% higher winrate)
• "Δ vs average levels" is how the levels of decks utilizing this unit/commander deviate from average (e.g. decks utilizing the Hammerhead are 0.6 levels lower than the average)

Observations:

• The Suzaku APC deck (Strongarm, Rifle, Missile, Pitbull, APC, Drones, Orca) is the most common deck. If we include the variations Liang vs Strongarm, Talon vs Drones and Mohawk vs Orca, 11 of the the top 50 decks follow this pattern.
• As APC also appears in two MLRS decks, two Predator Tank decks and an MG deck, there's 16 decks with APCs, an increase from 24% to 32%.
• Pitbull usage increased by 10 ppt to 68%, driven partly by the Suzaku APC deck popularity.
• Slingshot usage went down in parallel by 8 points -- which makes sense, as having both Pitbull and Slingshot feels inefficient in most cases.
• Jumpjet usage increased from 16% to 24%, and the JJ decks have a 4 points higher winrate compared to the average. They appear in a pretty diverse range of decks: 4 Orca Bomber decks, 2 MLRS decks, 3 decks with Orca, and a couple others. They are often the choice of high-skill players with relatively low levels such as srpss, r3bb and Maximus
• Sniper usage went up from 18% to 24%, which might partly be a reaction to the popularity of Chem Troopers (see Nod analysis). To see how good Snipers can be in the hands of a skilled players, watch the AliciaDestiny vs ThxForMMR finals from last Friday's Crazy Pimpin' tournament:

• It's nice to see that all GDI air units see significant, but not dominant play. All of the 6 units range between 14% (Hammerhead) and 34% (Droneswarm) frequency.
• Hammerhead usage went up by 8 points. It is either pared with Orca Bomber or Mammoth Tank as part of the "Air Mammoth" deck.
• Jackson lost 12 points of usage to Liang and also to Solomon. 
• Strongarm is seen in 15 decks, 10 of which are Suzaku APC and it's variants.

Player highlights:

• srpss has a strong 80% winrate with L12.9. He fields Rifle, Missile, Jumpjets, Pitbull, Talon, Orca Bomber and Liang.
• KlarJK stands out for 84% winrate on "just" L13.4. He plays a Talon, Mohawk variant of Suzaku APC.
• r3bb and Maximus have 72% / 73% winrates with L13.1 and Jumpjet Troopers.
• Bikerush, AliciaDestiny and Benven2 form a league of their own with winrates in the 90ies. 

Thoughts on balance: I believe people are often too quick to judge balance based on their personal experience or even on reports such like this. Balance heavily depends on maps (which map and whether you know it before the game), levels and skill:

• In tournaments, people play at equal level, which means that Epic units are not underleveled (e.g. you might not be able to play Snipers on ladder because they are underlevel). You often know the map before the game and can adjust your deck, which can, depending on the map, favor specific playstyles such as Tech, Fortress, Snipers etc., which would be hard to play on a random map. 
• If we'd now buff a certain unit to make it more viable on ladder, it might become even stronger on certain specific maps.
• Whilst on ladder, Nod is stronger than GDI, GDI was still seeing a lot of play in the recent tournaments. Reasons for this can include being able to pick a very specific deck for a specific map, or to surprise your opponent. Alarak's main theory is that Nod Laser Drones are so good against Nod, that their prevalence alone made GDI more attractive.
• At the absolute top of ladder, e.g. Top20, certain decent decks become less viable. For example, 13lade recently said he doesn't play Air Mammoth on ladder, as ladder opponents in very high Tiberium know how to deal with it, so his winrate is lower compared to other decks.
• Does this mean that Mammoth is too weak? No, because it's very hard to deal with for most players -- and is still even used in tournaments by the very best players.
• Tech or Scarabs might not be that viable in the Top 100. But buffing them so they see "equal play" in the Top100 would mean that for 95% of the player population, they would become even stronger. In lower leagues, you see 95% Tech -- so in my view, there's even a case to be made to nerf Tech rather than buff it (because it's not critical to have viable Tech for top players, but it can be critical to not have 95% Tech for newer players).
• Even Bikes that see 98% usage in the last Nod analysis are not necessarily broken. They might just be a necessary staple for Nod. In most RTS, there are some units that are built nearly every game. I don't want to say that Bikes are weak (they are certainly strong), just that 98% usage does not imply we're in a horrible state of balance.

I'm not making a specific case for certain buffs or nerfs here. My message is: (a) Balance is delicate, and optimizing for just one of these scenarios (such as Top Tiberium or Tournaments) can have adverse effects overall. (b) I personally find the current balance and diversity of play pretty decent right now.

See the Nod variant of this analysis