Kelsey Moser: Weldon Green's troubling G2 Esports training narrative

A perplexing narrative surrounds G2 Esports’ training philosophy in the EU LCS.

The first I heard of it was listening to commentary during the series between H2K-Gaming vs G2 Esports while reviewing games in preparation for Intel Extreme Masters Katowice.

EU LCS caster Mitch “Krepo” Voorspoels described a conversation he had with G2 Esports’ support, Alfonso “mithy” Aguirre Rodríguez, which began with Krepo asking him why he might not trade as aggressively against an opponent’s mistakes in lane. I assume Krepo’s response was a paraphrasing.

“You know what, we could go for a massive lane lead right here, but it involves a little risky play. Instead we’re trying to play a very clean game, focus on objectives, and if we get lane leads, we want to use lane leads to get a team lead.”

Krepo presented this as if it made perfect sense, but I was left with questions. Why the distinction between a lane lead and a team lead? Do G2 not understand how to play around pushing lanes?

The concluding line in this discussion, however, was even more puzzling.

“That’s where the whole theory from that game style comes from. Because on the international stage, you will not be able to generate any lane leads versus these players.”

This assertion is factually inaccurate. Historically speaking, the now de facto best team in the world, SK Telecom T1, often make most of their mistakes in laning phase. When teams have beaten them, it is often due to snowballing lanes and playing well around mid lane pressure with a team’s jungler. One of Lee “Faker” Sanghyeok’s greatest strengths — calculating mid lane outplay potential — is also his greatest weakness if a team can capitalize on his over-extension.

G2 managed to get early game leads against one of the LCK’s historically best early game teams, the 2016 ROX Tigers. If G2 are claiming they shouldn’t rely on their opponents making mistakes, that’s one thing, but if they are trying to claim that they won’t be able to get early leads off punishing mistakes internationally, they haven’t watched their own games at Worlds.

Enter the Mind Game

One of the team’s assistant coaches, Weldon “MindGamesWeldon” Green, published another version of the early game handicap narrative just prior to the first day of the IEM World Championship. The premise of the video was that casters should stop using average game time as an indication of a team’s skill, as 20-40 minute decision-making was the most important part of the game.

I agree with Green’s assertion that short average game time shouldn’t be used as an indication of skill, but rather to contextualize playstyle. From there, he lost me. Green goes too far in stating that game time in losses is more strongly associated with skill. He should know Jin Air Green Wings are not the best team in the history of LoL by any stretch of the imagination.

Beyond that, the more difficult part of the picture Green has painted is the idea that you can break a game down into parts and objectively decide which is the most important one to focus on.

Unlike the narratives the casters propelled (this in itself is troubling, as it implies G2 or those around them lack a sense of consistency in understanding the motivation for their approach), Green seems to be of the opinion that it doesn’t matter if you can generate an early game lead, as the game will come down to a few decisions post-20 minutes. Rather than arguing G2 won’t get lane leads against teams at international competitions, he claims they won’t matter.

Green insists that once habits of “late game compositional play” are so engrained, “you literally cannot lose games anymore.” He also makes a somewhat blanket assertion that “when teams choose to not snowball a game (so for example, when they get a lead, and then they don’t do anything with a lead for like 20 minutes, and they wait till 24 or 25 minutes when Baron is easier to tank), this isn’t actually a flaw.”

I will give Green the benefit of the doubt and assume he’s over-simplifying for convenience. Not all compositions are well-suited to setting up to play around Baron at 24-25 minutes. Some benefit much more from rushing turrets if the enemy team tries to take Baron or forcing teams into objective trades that don’t even require Baron to spawn.

This leads to the somewhat baseless assertion that, without focusing on training decision-making in the first 20 minutes (indeed, he states teams “should focus on the most important facets of the game almost exclusively”), you can rely on accumulating at most a “0 to 3k” gold deficit.

Even prior to the IEM World Championship, flaws in G2’s early game were plain to see. Just pulling up G2’s first game against Origen, the worst team in the league, mistakes are obvious.

Origen should have had two winning lane matchups after Shen’s first back. At least initially, Taliyah should have push advantage against Ryze. Origen’s laners didn’t attempt to defend Kim “Wisdom” Taewan’s camps against straightforward invade routes, even when Wisdom matched Kim “Trick” Gangyun, creating an expectation for an invade, or G2’s solo laners backed at the same time of a contest (this happened twice in the first six minutes).

G2 also have habitually drafted lower pressure junglers with weak lane matchups and abused mistakes (contrary to the EU LCS caster narrative) to make it to late game. For example, in their first game against H2K, they forced Fabian “Febiven” Diepstraten to burn a summoner spell before minions spawned. This gave them the luxury of allowing Trick to farm uncontested as Rumble against an Ivern.

Regardless of whether you go off the EU LCS caster narrative or the argument Green proposes, these types of mistakes make it hard to believe G2 can rely on only accumulating a minimal deficit by the 20 minute mark. Indeed, Flash Wolves categorically proved at IEM that they cannot.

In G2’s first encounter with Flash Wolves, Hung “Karsa” Hauhsuan used experience advantage and a stronger early game jungle pick to ensure a pressure lead in mid lane for Huang “Maple” Yitang. With a mistake from G2’s bottom lane at level six, and Karsa appearing top lane, Flash Wolves guaranteed themselves three lanes they could play around before Trick could leave his own jungle on Rengar. As a result, Flash Wolves used this to maintain river control and keep the enemy jungle well warded, growing their lead with contests on neutral objectives and turrets.

Going into this IEM World Championship, I personally discounted Flash Wolves for reasons that align with Weldon Green’s philosophy. In their games against Unicorns of Love, Flash Wolves reflected a poor understanding of how to adapt to play to their composition around mid game, defaulting to grouping mid, even with a 1-3-1 composition.

This is consistent with their play in LMS and the last few international tournaments. The addition of Lu “Betty” Yuhung may have buffed Flash Wolves’ teamfighting, but their ability to close and make lane assignment decisions in the late game remains flawed.

The first game against Flash Wolves and the initial blowout by ROX Tigers against G2 seemed to be enough to make them change their tactic and draft stronger early game lanes in other matches throughout the tournament. Trick placed much more pressure on his lanes, but G2 still struggled to play around the correct lanes to control the jungle.

In Game 3 against ROX, Trick and mithy attempted an invade without mid lane control against a Shen composition. Tigers’ top, mid, and jungle were able to react quickly and punish them. These kinds of mistakes continue to plague G2 throughout IEM and appeared most glaringly in the final (even despite initial early kill pickups), a crushing 2-0 win from Flash Wolves.

Flash Wolves, “Koreans”, and TSM

One could easily dismiss Flash Wolves’ pre-20 minute strengths if they didn’t consistently net them single game wins against some of the top teams in the world historically (like SKT and 2015’s Tigers). Green’s main motivation for his proposed philosophy that strong decision-making in mid and late game is the most important thing to focus in beating what he refers to as “Korean teams.”

There’s a habitual reduction of “Korean teams” to one style of play or a collection of a finite and fixed list of strengths and weaknesses within the western narrative. Part of the reason this has been perpetuated is SK Telecom T1 has overwhelmingly represented LCK at international events, especially in 2016 when they won the IEM World Championship, Mid Season Invitational, and the World Championship in the same year.

SKT and Samsung Galaxy did focus the strengths Green praises in honing mid-to-late game decision-making, but 2016 Tigers were something else. Their strong early game, propelled forward by Han “Peanut” Wangho’s top gank and jungle invade strategy, may align more with the high-risk, coin-flip play of the current ROX Tigers, though with much better execution.

But Flash Wolves have almost consistently bested every top “Korean team” they’ve collided with internationally. They have won 3/4 encounters with SK Telecom T1, and pundits have advocated examining what has made Flash Wolves so powerful.

As demonstrated by the recent IEM, it’s the exact opposite of what Green has suggested. With the highest average gold lead at 15 minutes in the event, Flash Wolves took home an international victory, struggling the most against H2K-Gaming, a European team whose strengths align most closely with Flash Wolves: strong early snowball and macro play, but with poor mid-to-late game decision-making.

Obviously, it’s smart for G2 to train mid to late game decision-making if they threw games at Baron against a top team in their group at Worlds, but it seems strange to discount victories over SKT from teams like Flash Wolves or the hyper aggressive EDward Gaming at 2015’s MSI. And this raises more questions about G2 and Green’s motivations.

When teams frequently throw games around the swinging Baron objective, that speaks to its power. As Baron doesn’t spawn until 20 minutes, it makes sense that G2 would focus on that phase of the game. But the problem is that how a team performs in the first 20 minutes will heavily influence Baron conditions.

Maintaining river and jungle control throughout the early game should give a team vision control and a stronger ability to play around the objective. This comes through securing strong pushing lanes and using a team’s pushing side of the map to set up and defend vision. These concepts don’t start after Baron spawns, but well before. If you have strong early control, that puts more control in your hands for the late game. The onus is on you to make a mistake and let the enemy team into the game; with poor early game decision-making, you rely on the enemy team to make a mistake instead.

Last year, Riot Games drastically altered the way EU LCS teams in particular were playing the early game. As the region with the highest frequency of lane swaps, one could watch EU LCS for greater nuances in how they were played. Yet the removal of lane swaps made many top EU LCS teams focus on learning a new way to play out the early game. For a team that focuses on early game, it’s easy to see how a radical patch might undo all their hard work.

This kind of change can happen to any stage of the game. As already mentioned, Baron currently has too much power. Riot Games has made attempts to balance Baron’s power by introducing other scaling objectives like Elemental and Elder Dragons, but so far this hasn’t achieved the desired result. This suggests that Riot could attempt to alter Baron again, which might transform certain given aspects of late game decision-making and require new training.

Following the removal of early game lane swaps, League of Legends strategy became much more about playing around strong lanes for jungle invades. This game concept has long been a core part of how select teams have played. That hasn’t changed at all, and it’s hard to imagine it will unless Riot makes a League of Legends 2.

Beyond these considerations, Green’s main motivation for tunneling on mid-to-late game came out in his discussion of Team SoloMid’s failures at the 2016 World Championship. When Green discussed TSM’s failure despite their dominant run in the NA LCS, he brings up a training philosophy centered around abusing and pressuring the early game.

“This was because it was specifically trained … One training mechanism that I implemented was essentially like, if we aren’t in a play at any point in the game, we’re doing it wrong … What that results in is a really proactive and overpowering 0 to 20 minutes against teams that cannot play defense — and even against teams that can play defense, it’s a lot of pressure.”

It’s misleading to bring up Team SoloMid as an example of a team with strong early game that wasn’t training mid or late game technique and lost as a result, because Team SoloMid didn’t lose games at the World Championship as a result of their mid or late game play. With the exception of one loss to Royal Never Give Up, TSM were behind in gold at the 10 minute mark in all of the World Championship losses, and in that game, they had a modest lead of 553 and never achieved a gold lead over 2.3k. In addition, one of TSM’s three wins featured them falling behind 561 gold at 10 minutes and amassing a deficit of 6.4k by 32 minutes before they could orchestrate a comeback. In their other two wins, TSM effectively cultivated a lead and snowballed it to victory.

Green later points to the specific game against Royal Never Give Up where he states they had a lead the entire time, but Cho “Mata” Sehyeong was able to execute engages and win off three teamfights. This seems disingenuous given that RNG’s engage plays were spaced throughout the game and based on somewhat easily exploitable pathing from Dennis “Svenskeren” Johnsen with the first pick occurring before four minutes into the game. RNG only lagged slightly behind in mid game objective trades, which is far from the “snowball” narrative Green tries to paint.

Team SoloMid over-training the early game and losing later wasn’t the problem internationally. I’d instead argue, based on observing their games, TSM over-tunneled on the bottom side of the map and made fundamental mistakes in early macro play on red side because they were not able to control the top side river.

Their strategy relied a lot on denying the blue buff from the enemy mid laner, and if they could not control the top side river from red side, they would not be able invade the enemy blue buff properly. In this manner, Søren “Bjergsen” Bjerg couldn’t get the same advantages in lane he was used to. Arguably, in their Week 2 loss to Samsung, it was failure to execute this strategy that prevented TSM from having a strong laning phase and snowballing enough against SSG to win mid game.

Other problems included a failure to rotate their bottom lane mid with an advantage, struggles to play around top side, and an inability to handle bottom lane turret dives. Many of these problems are associated with the early game, and if TSM really did have a “well trained, proactive play-making machine,” it wouldn’t have cost them.

If Green is really overreacting to say that early game training doesn’t matter based on performances TSM at Worlds, he’s not only neglecting a lot of data from other teams, but ignoring much larger problems in TSM’s play.

Play to your strengths

Rather than gut-react in the opposite direction and advise teams focus early game strategy and mimic Flash Wolves if they want to defeat top “Korean teams”, I suggest this philosophy is just as dangerous. I would instead suggest teams spend time training all types of decision-making. There’s no way even a more mid game focused team like SKT goes extended periods without practicing early game strategy. But if a team has to budget time, they should focus on their strengths.

That’s exactly what SKT, Samsung Galaxy, and ROX Tigers seem to have done on stage in 2016, and it’s precisely what made them the best. What Green has left out of the narrative is that their strengths varied drastically (ROX were an early game top side team, SKT a mid-focused mid-to-late game team, and Samsung an adaptable, reactive grouping team).

Historically, G2’s strengths definitely are not SKT’s. They shouldn’t attempt to blindly copy them.

In 2015 Spring, I wrote an article praising G2’s somewhat advanced understanding (within the context of the EU LCS) of playing around pressure generated in mid lane to control the jungle, and then another on how mid lane and jungle control dictated their victory in playoffs. Over time, this power has declined, and jungle and mid feel disconnected. It got to the point where, when I interviewed mithy prior to Worlds in 2016, he felt EU as a region was behind in playing around their jungler.

Another of G2’s strengths hails from their bottom lane. Jesper “Zven” Svenningsen and mithy have invested so much time into mastering bottom lane matchups that they almost constantly managed to generate lane leads on Origen and G2. mithy can talk at length about how to punish enemy supports in a 1v1. They have historically avoided leashing for their jungler so they could ensure push priority at level 1, but this too has changed this season.

Whether or not it’s because G2 have made early game less of a priority (perhaps they haven’t at all, and Green’s narrative is simply for the cameras), Zven and mithy’s laning has appeared to decline. Zven has made simple mistakes, like positioning too close to a wall against a double snare bottom lane, and mithy has over-extended in trades repeatedly. While G2 have performed well domestically in the 2v2, they have still failed to perform up to their standard, and their flaws came out in full force against Flash Wolves’ improved duo.

But if G2 have focused less on developing these early game strengths, surely their intense mid and late game training has yielded benefits in other areas. So far, I haven’t seen positive results, especially in lane assignments. G2 have consistently kept their duo lane on top side after first turret falls in the bottom lane too long instead of rotating into the mid lane. This exposes them to top river flanks, and at times also fails to utilize other potential compositional advantages.

For example, in Game 2 in the IEM Grand Final against Flash Wolves, Luka “PerkZ” Perković’s Jayce remained confined to the short lane against Syndra well past the 20 minute mark. Rather than take advantage of Jayce’s superior ability to side lane and send G2’s duo to the mid lane to force Syndra to rotate to a less comfortable position, G2 merely left Jayce in a zero sum farm game against Maple. This ultimately benefitted the Flash Wolves when they baited G2 into a frankly idiotic contest over Cloud Drake that allowed the LMS team to pick up Baron at 21 minutes in.

G2 need to train mid to late game decision-making if they are making these choices, but so far they seem to be doing so at the expense of their strengths. That’s an oversight, and their games at IEM, especially when G2 shifted to a more early game focus in draft, made that stand out.

If a team’s natural strengths lie in the early game, it’s even more important that they should continue to hone them. Green advocates building fundamentals, but fundamentals are present in every phase of the game. Only focusing on mid and late game for an extended period of time instead of alternating the types of compositions G2 practice in scrims in much shorter periods (perhaps a week or two per cycle) will cause early fundamentals to lapse.

Congratulations to G2 on their Top 2 finish at an international event. It pains me to see the team lose some of its greatest strengths without payoff. If anything, I hope the experience has taught them that the most important part of the game is the part where you find your best advantage. Your condition for victory is whatever you make it.

Data taken from OraclesElixir.com.

You can follow Kelsey Moser on Twitter @karonmoser.