From 2014 to 2017: Team SoloMid's mid-jungle problem

On Episode 17 of Narrative Wake, now Team SoloMid General Manager Parth “Parth” Naidu confessed he chose to step down from his position as Head Coach of North America’s top League of Legends team in part because there was a major flaw in TSM’s early game play, and he found he could not fix it in time.

Based on a cursory look at gold data, 2017 was Team SoloMid’s worst early game at a World Championship since Søren “Bjergsen” Bjerg became the centerpiece of the team. In 2014, TSM had an average gold difference of -110 across group stage and its single quarterfinal appearance. In 2015 and 2016, TSM averaged a gold lead at 15 minutes. Yet in 2017, TSM averaged a massive deficit of 1,866 gold at 15 minutes, a reversal of its 2016 lead of 1,867 gold at 15 minutes.

Something must have happened in the most recent World Championship to destabilize TSM’s early game. Looking at how TSM played out its early game extending even as far back as Season 1 paints a separate picture. TSM have almost consistently had similar priorities in terms of setting up its mid laner to have control of the lane and looking to deny the enemy blue buff to make this easier. Where they fall short, however, is in constructing a plan to control the map once they have achieved a mid lane lead. Namely, the jungle and mid lane often appear to have separate objectives and rarely work together to extend the mid lane lead.

To investigate problems in mid and jungle control, I went through TSM’s World Championship games starting from 2014 through 2017 to pay close attention to:

1. Priority when the jungler ganked top, mid, and bottom and Bjergsen’s actions at the same time
2. Bjergsen’s actions when he could push out mid lane and leave either to ward, gank, generate fake pressure, or assist in invades
3. How TSM junglers used or coordinated their behavior after backs relative to mid priority

With this closer look, I found gaps in TSM’s early game even in arguably their most impressive early game World Championship appearance (2016). Highlighting these flaws serves as a starting point for solving TSM’s flaws in 2018.

History and the pre-Bjergsen era

Investigations of TSM’s early game flaws have revealed a few clear predictable traits over the years. At the conclusion of 2016, I investigated the team’s fascination with blue buff denial and its dependence on Søren “Bjergsen” Bjerg having push in mid lane. Statements by Dennis “Svenskeren” Johnsen and others supported the existence of this approach. It’s also easy to see how TSM function off mid lane control domestically.

Evidence of priority on mid for buff control exists as far back as TSM’s first World Championship. Over the course of the event, jungler Brian “TheOddOne” Wyllie began favoring the likes of Nunu, and TSM aimed to start in their opponents’ blue buff area from level one. Many have gone so far as to attribute heavy counterjungling to TheOddOne, but there’s no evidence to support TSM actually devised the strategy.

Observing TheOddOne and Andy “Reginald” Dinh, one can reasonably assume TSM put a lot of emphasis on blue control so Reginald could keep an advantage on his opponents. As was common at the time, Reginald and TheOddOne didn’t often roam together, but if Reginald went bottom for a gank, he donated mid lane minion waves to TheOddOne. Mid lane priority or keeping mid pushed for these kinds of plays wasn’t considered as fundamental in competitive strategy as it has become over the years.

William “Scarra” Li pointed to a particular predictability in how TSM played in Season 2. “TSM was running, like pick whatever champions, these 12-15 champions, we’re going to invade blue buff at seven minutes. Let’s go team… That would be their strategy. They did that strategy for all of Season 2. They were much more than just, but like, you could just bet they would contest your seven-minute blue.”

While Scarra may have exaggerated this trait, Team Dignitas did, to an extent, seem to operate on the assumption that TSM would look for blue buff invades during the Season 2 Regional Final. In the second game, Dignitas grouped mid at around 13 minutes after placing defensive wards to head-off an invade attempt.

Of course, setting up mid and jungle play was significantly less advanced in the first three seasons. Laners felt much more confined to top, mid, and bottom (with the exception of lane swap scenarios), while the jungler or support shared the bulk of the responsibility for map control. Any observations outside noting early roots of how TSM’s strategy evolved wouldn’t be productive.

Gank priority (the “Dyrus treatment”)

Marcus “Dyrus” Hill served as TSM’s top laner for a significant portion of its run until his retirement at the 2015 World Championship. While he played for TSM, commentators learned to refer to how TSM played as the “Dyrus treatment” where the jungler seemed to gank mid and bottom lane almost exclusively, leaving Dyrus isolated. He didn’t necessarily react well to this, giving up kills to the opposing top and jungle.

No matter where one stands on the side of this debate, looking at TSM’s gank priorities over the year can paint an interesting picture.

Percentage of ganks within the first 15 minutes to each lane by year

A gank is counted if the jungler appears in the lane with the apparent intent to secure a kill, force a back, a summoner spell, or otherwise regain the lane advantage for his laner. That doesn’t, however, include the jungler coming to lane without the intention of taking the enemy laner by surprise. Only ganks that occur within the first 15 minutes of a game are counted.

Sample sizes are small (10 games in 2014, seven in 2017, and six games otherwise), but a small shift in priority seems to have occurred between 2015 and 2016 when Dyrus left the top lane role, and TSM’s “super team” built around Bjergsen and Yiliang “Doublelift” Peng took the reigns. Mid also received less ganks, as TSM seemed to rely more upon giving Bjergsen the counterpick and allowing him to get lane control. Every year, TSM only blind picked their mid lane champion in two of their games.

This trend continues further when only examining the first gank the jungler made in a game.

Percentage of first ganks of the game to each lane by year

From 2015 to 2016 and 2017, top lane received the first gank of the game in two thirds of the games in which a lane received a gank in the first fifteen minutes (in Game One for TSM against Flash Wolves of 2017, Svenskeren didn’t gank within the first fifteen minutes), practically switching priority with mid lane.

Part of this shift should be attributed to enforced standard lanes in the 2016 and 2017 World Championships. With lane swaps in meta, mid lane often stood out as the only lane to gank early if a team got tempo advantage. In standard lane meta, mid more often served the role of clearing mid or roaming to side lanes with the jungler for a gank.

TSM continued to counterpick its mid lane champion picks to gain mid priority. In 29 total World Championship games starting in 2014, TSM only blind picked a mid lane champion in eight games total: two each year. Despite the drop in priority to early ganks in mid lane, there was no appreciable difference in Bjergsen’s ability to maintain lane priority.

I measured lane priority by the percentage of time both mids are in lane, and Bjergsen is pushed over the halfway point within the first fifteen minutes of the game. In 2014, Bjergsen had lane priority for most of the first fifteen minutes in 50% (5/10) of games played, 50% (3/6) in 2015, 66.67% (4/6) in 2016, and 57.14% (4/7) in 2017. As such, Bjergsen keeping the lane pushed forward and enabling the rest of the map continued despite him receiving fewer first or early ganks in 2016 and 2017.

As such, it’s reasonable to claim that TSM relies on Bjergsen to enable the jungler to have pressure on the map and not the other way around. Observing what Bjergsen does while the jungler ganks can also shed more light on the situation. As I observed by looking at TSM’s 2017 Spring Playoffs matches, Bjergsen’s mid lane wave status was correlated with the success rate of Svenskeren’s first gank.

While I didn’t measure gank success in this investigation, it’s easy to theorize ways in which Bjergsen having mid priority can impact a gank. By pushing out the wave, Bjergsen restricts his opponent’s ability to roam, can create more opportunities to have better vision coverage, might force the enemy side laners to play less aggressively for fear of his interference, and could even pull the enemy jungler’s attention, lessening the likelihood of an even match.

Mid lane status at time of jungle gank (percentage of total jungle ganks) by year

*One duo gank in 2015 was achieved by Bjergsen Teleporting top lane ~2017 only had 10 total ganks in the first 15 minutes, relative to 36 in 2014 and 18 each in 2015 and 2016

A duo gank is only counted if Bjergsen leaves lane before his jungler appears in top or bottom lane for the gank. Bjergsen has priority at the time of a gank if the mid wave beyond the halfway mark in mid lane. If he is backing, roaming to another part of the map, or warding during the gank, he either is “resetting” or “out of lane.” If mid lane has negative priority, the enemy mid laner is able to push over the halfway line along with the wave. “Mid Gank” simply means that mid lane is being ganked.

Because of the small sample size, not a great deal can be gleaned from these observations except that coordination with the mid laner by the jungler in side lane ganks didn’t change with increased side lane gank priority. The most coordination in ganks that a jungler had with mid lane occurred in 2014 and 2015 when Amazing and Santorin ganked for Bjergsen. Svenskeren and Bjergsen didn’t often roam together, and Svenskeren did not note the position of mid lane in his ganks. Compared to teams like Misfits that often involved their mid laner in gank setups, it seemed that TSM’s mid-jungle duo lacked planning and coordination.

While this may not seem significant, securing priority in mid for map plays in side lanes can have a significant game impact. Especially in 2015, when mid lane often took Teleport, having priority mid usually made it very easy for the mid laner to defend his jungler in case of a countergank. If a mid laner has to leave lane without priority, his opponent can either freely pressure the turret, leave lane to match without falling behind, or create fake pressure (leave lane with no destination in mind, but disappearing into the fog of war to force opponents to back off).

If a gank begins in top lane, for example, and Bjergsen has priority, he can always join a resulting skirmish before his opponent, who will have to push the wave to bounce it. Otherwise, when Bjergsen returns to lane, he immediately regains control and can pressure the lead. This is also why many high level teams often coordinate ganks to side lanes around their mid laner getting an earlier back. The enemy mid laner backing after the team’s mid laner will trigger the team’s jungler to back and immediately go to a side lane because he knows his mid laner will have priority.

Bjergsen’s actions upon leaving lane

After establishing the importance of mid lane priority to actions that occur across the map, it’s also important to look at what Bjergsen does when he has wave control and can leave lane.

Actions by year when Bjergsen leaves lane with mid priority in the first fifteen minutes of a game (Percentage of total)

*The ganks in 2015 includes one Teleport gank

These observational data are based on World Championship games for Team SoloMid. These data do not include instances where Bjergsen leaves lane to take his own blue buff. “Ward” constitutes warding near mid lane either on jungle entrances, in pushes on the side of the lane, or in river or pixel brush near mid. “Gank” constitutes instances where Bjergsen leaves lane either with jungler or on his own to gank a side lane before the gank is initiated. “Defend team” signifies that a skirmish or gank is occurring in jungle or the side lane, and Bjergsen leaves his lane after it begins to defend his teammates. “Dragon” indicates Bjergsen has left lane to help his team take dragon. “Fake pressure” signifies instances where Bjergsen leaves lane to disappear into the fog of war, but performs no other action. “Invade buff” signifies that Bjergsen is either warding or invading to take enemy buff with or without his jungler. “Camp clear” signifies when Bjergsen leaves lane to clear a jungle camp. “Swap lane” occurred only once when Bjergsen left mid entirely to side lane within the first fifteen minutes of a game in 2014.

The most interesting thing to note here is that the year where Team SoloMid managed the largest average early game lead (1,867 gold at fifteen minutes in 2016) was also the only year where Bjergsen’s warding on priority comprised less than 60 percent of his action. Even in this case, however, most of Bjergsen’s behavior was reactionary; 21.82 of his actions when he left lane were simply to defend Svenskeren on a jungle invade or if he got caught in river.

With the exception of 2015 where this only happened once, TSM junglers died twice per year in the river between mid lane and dragon or around the blue side raptors area. While Svenskeren was criticized heavily for these deaths in both the first game against Team WE and the first game against Misfits in 2017, getting caught in this area isn’t uncommon for TSM’s junglers. “Defend” actions generally occur because the jungler will take an initiative on a play either against mid priority or without considering the position of members of the enemy team (in 2017, it was specifically to try to bait out Shen’s ultimate).

The fact that the jungler commonly dies near mid lane further indicates a lack of coordination between mid and jungle. If most of Bjergsen’s plays with his jungler are reactionary “defense” plays, then that indicates either a lack of pre-planning and communication on map movements between the two roles. The jungler needs to either wait for mid to have priority or mid can take the initiative to call to contest river vision with the jungler.

Actions on the jungler’s backs

While it’s difficult to compare Bjergsen’s actions directly to his opposing mid laners’ because of the change in flow of the minions, champion matchups, and red and blue side differences, seeing what TSM junglers do relative to their opponents on backs can be a fruitful exercise. When a jungler backs and returns to the map, he may have a brief window with an item advantage if the opposing jungler has yet to back or if his mid laner has priority where he can extend an advantage. He can do this by placing offensive wards to extend vision control or looking to make an invade or another type of play on the map.

Simply clearing a camp in his own jungle can waste the jungler’s back advantage. Of course, sometimes clearing one’s own camp is the best course of action if it isn’t safe to venture into river after the back because the jungler is behind in itemization or his lanes cannot assist him. It’s important to note whether jungle backs with mid and mid has priority when the jungler returns to the map.

Looking specifically at what the jungler does when his mid laner has priority during his first action after a back also compensates for the imprecise measure of majority mid priority. In matchups, mid control can often change on an item spike or misplay so that a mid laner can spend just over half of the first fifteen minutes with wave control. Like gank coordination, action of junglers on backs can give a more specific look at how mid priority impacts TSM jungler behavior.

In 2014, 2015, and 2016, TSM either had mid lane priority or the mid wave was resetting when the jungler returned to the map in more instances than the opposing team. As a result, one might predict that TSM jungler actions on backs should have been more aggressive than the opponents to take advantage of refreshing wards or an item advantage. In 2017, the opposing mid laner had priority for a larger proportion of TSM’s backs (27.59%) then Bjergsen had priority on the enemy jungler’s backs (25.00%), but it’s a difference of only one back instance.

Despite opposing mid laners having priority less often when their junglers returned to the map after a recall or a death, enemy junglers had a higher proportion of invade instances in all four years. With the exception of 2017, TSM junglers were consistently more likely to just clear a camp on respawn instead of looking to extend a map advantage in some other way. Even in 2017, the majority of non-camp clearing actions were defensive wards (perhaps a symptom of less early game control overall).

More aggressive actions from the enemy jungler might not be a positive thing if his mid laner has priority less often when he returns to the map, as it means he could get caught out more often. Yet, with the exception of 2016, TSM’s jungler died more often in the first fifteen minutes of games in total than enemy junglers, and only one extra death separated enemy junglers from TSM junglers in 2016.

Even so, it’s important to see how the proportion of the junglers actions change on returning to the map while the mid laner has priority. Looking only at instances where mid lane has priority when the jungler performs his first action after backing, one would anticipate a higher rate of invades for both TSM and enemy junglers.

Opposing junglers still generally a wider variety of actions after backing when mid lane had priority than TSM’s junglers did. It’s important to note that, in 2016, 28.57 percent of first actions on back with mid priority took place in the enemy jungle, but this is the only year where TSM’s junglers performed more invasive actions after backing when mid lane has priority. In 2014 and 2015, Bjergsen having priority didn’t seem to have any impact the the types of actions the jungler made after backing.

While actions of a jungler after backing is a specific concept, the results indicate that TSM’s use of mid priority to further map control in all four years was relatively limited.

Conclusion and looking forward to 2018

Observations on jungler and mid lane coordination in ganks, Bjergsen’s actions when he has lane control, and jungler actions on backs from TSM’s World Championship games in 2014, 2015, 2016, and 2017 supports that TSM had worse coordination overall between mid and jungle than their competition in these games. The jungler didn’t seem to react differently depending upon whether mid had priority, and mid didn’t often move with the jungler to get vision.

Yiliang “Doublelift” Peng on Narrative Wake at least indicated part of the Bjergsen’s thought process on Narrative Wake when he played from behind.

“If [Bjergsen’s] behind,” Doublelift said, “he’s like ‘Guys, I can’t do anything. I’m going to sit at my tower, and he’s going to roam on every wave.’ … I was like ‘There’s no way he can roam on every wave,’ and he’s like ‘No. He can roam on every single wave.’”

Bjergsen is mindful that keeping mid control restricts mid from roaming, getting better backs creates more opportunity for the team, and that it’s important to note what can be done by himself and the enemy mid on “every wave.” But TSM have yet to maximize these factors on the international stage, and most of the criticism has blown back on Bjergsen and his junglers as individuals rather than their coordination and overall team play which should be under fire.

In 2018, TSM have opted to sign yet another jungler known for his aggression. Mike “MikeYeung” Yeung follows a similar pattern to Maurice “Amazing” Stückenschneider, Santorin, and Svenskeren. He displayed a high level of jungle aggression on previous teams. Following this same pattern might indicate that TSM are making the mistakes of assuming junglers have to take sole responsibility for controlling the map off mid pressure instead of this being a team responsibility. Yet MikeYeung’s high level of communication may make him more moldable.

MikeYeung himself displayed similar problems of invading against priority on Phoenix1. He invaded against priority, and when I asked him about his decision-making, and he advocated the best pay to play is “without fear or making a mistake.” But a foundation in communication could lay groundwork for the mid-jungle core on TSM to improve.

"Before every single game,” MikeYeung told me at Rift Rivals, “I like to talk to myself, or I like to think to myself, or have the coaching staff come talk with me in the room with all the players. I like to think about how I want to play today.”

When TSM junglers have had a tendency to die near blue sides raptor entrance, even when Bjergsen has control of the lane, communication and planning has seemed absent. Talking about what mid and jungle can do together on every wave may change the situation.

In addition, planning jungler actions on backs will prove more important. More work can be done to see what TSM junglers do when they back before their opponents, how often backs are coordinated with laners and how that impacts vision control. For much of 2014 and 2015 especially, TSM’s junglers often immediately went for their own raptors after backing, making them easy to track. Emphasizing river vision on priority over clearing camps will serve the added bonus of making TSM’s junglers harder to read.

A historical look at TSM’s World Championship performances indicated that some of the 2017 “early game problems” have existed for much longer, even if TSM averaged gold leads in previous World Championship Group Stages. But identifying them and setting up a communication plan for improvement means that 2018’s problems don’t have to be as predictable.

---

Photo Credit: Riot Games