Not too long ago, Riot’s patch notes included a familiar refrain:
Our big story for this patch is to bring some equality back to the jungle…we quickly realized the problem was less about buffing tanky junglers to keep up and more that we lacked a strong item path for utility junglers to pick up
This came from patch 4.11, back in July 2014.
Or how about this one, from October 2013:
The overarching philosophy for our jungle update is to create more options and possibilities for junglers of all kinds
Jungle diversity has been a sore point for quite some time, as no matter what preseason changes Riot tries in November, inevitably three or four junglers rise to dominate competitive play by June. Season Three had Nasus, Evelynn and Zac; Season Four had Lee Sin, Elise, and Kha’Zix; and up until recently, Season Five was Rek’Sai against Nidalee against Lee Sin against Vi.
The resolution to this problem is more complicated than simply widening the shrinking set of viable junglers, because the jungle is by its nature extremely volatile. Unlike laners, who eke out advantages little by little, only rarely getting the chance to make a huge game-changing play, junglers are consistently put in high-risk high-reward situations. Success or failure in a few small instances changes the complexion of the game. For example, a successful gank can lead to a kill, but also a period of map control which allows the jungler to place wards deep in the enemy jungle. These wards, in turn, lead to successful counterjungling or more successful ganks, which can work to further push the game out of control. The jungler has the power to capitalize on an advantage anywhere on the map, as well as to spread that advantage to the rest of the team. Thus, it is extremely dangerous for Riot to allow high-variance champions to be consistently successful in the jungle, since they are far less limited than laners in how completely they can take over the game. Presumably, Riot wants a relatively large class of relatively small-variance jungle champions to be viable enough for strategic diversity, without allowing too many edge-case champions to skew jungle results.
This is the backdrop for the most recent attempt at jungle changes, patch 5.5 and its newly ubiquitous Cinderhulk:
one of our goals with this ‘new’ tanky jungle enchant was to provide an item that not only provides valuable tank stats but also helps junglers with their early to mid game clears
Deep at the core of League of Legends is a triangle of ideal champion interaction that, as much as any other factor, defines the game. Autoattackers such as AD carries outscale and eventually defeat tanks, who survive and disrupt assassins, who themselves kill the carries. There are other factors to the game, but at the fundamental level, AD Carries are essential because they are the only type of champion that can consistently kill tanks. Tanks are powerful sooner than carries, since health is a cheap and accessible build path, and it is easy to multiply health with relevant resistances. To massively oversimplify, a Giants Belt, for 1000 gold, adds 380 health, while a BF sword only adds 50 attack damage for 1550 gold. The difference is that resistances are the only multiplicative stats available to tanks, while AD Carries can buy both critical strike and attack speed to multiply their damage. This means that autoattacks scale quadratically compared to the linear scaling of tankiness. Assassins, on the other hand, generally scale fairly well with flat damage items, allowing them to dominate lower-health targets, but they lack the multipliers to eat through tanks’ resistance. In a perfect world, these three roles act as natural checks on each other, and allow the game to correct itself if one element gets out of whack.
Difficulties arise because LoL is not that ideal vacuum-sealed world in which tanks, assassins and AD carries fight to one anothers’ mutually assured destruction. Even the most clearly defined champion is somewhere on the sliding scale of each of these three attributes. For example, Graves is clearly more tanky than Quinn, who has much more well-expressed assassin attributes than Jinx, even though all three are very obviously marksmen. At the highest level, effective jungle champions have historically had the mobility and spell damage of assassins for early skirmishing power and the disruption power of tanks, in order to remain relevant with less gold through the mid- and lategame. This usually holds true even if a jungler gets significantly ahead, presumably because the jungler practices with that specific role in mind. There have been occasional exceptions, but these are frequently champions who scale with something other than gold, such as Nasus or Feral Flare Master Yi. Thus is born the jungle stagnation that Riot has worked for years to fix.
Enter Cinderhulk, the insanely efficient jungle item that gives tanky junglers like Sejuani and Gragas a smooth transition from their decent early game to an improved midgame. However, not only does it allow these junglers to survive a potential midgame onslaught from the old guard of hybrid assassin-tanks, the bonus health passive also gives them a second form of multiplicative scaling, dangerously wobbling the triple balance at the core of League of Legends. It is thus not a surprise that Cinderhulk has lately made an appearance in lanes as well as the jungle, since tankiness can significantly simplify the game. Executing a creative game-winning strategy can easily become secondary to the simple question of whether or not a team has enough damage to eat through an enormous front line, no matter how well they devised and executed their game plan.
In the LCS, Cinderhulk made its presence immediately felt after its introduction in week 8, as it massively moved nearly every role way up on the tankiness spectrum. I looked through the NA LCS games of weeks 6 and 8 and I quantified the tankiness of champions in each role per game according to the following criteria:
- Tanky champions have high natural tankiness from their kit, with a maximum of 1 offensive or support item.
- Medium champions buy a mixture of offensive and support items, with 2 or more defensive items, or use an offensive build but have significant natural tankiness relative to their class
- Not tanky champions buy almost entirely offensive or support items, with limited natural tankiness from their kit
I allowed myself some flexibility with these definitions, both to account for different build paths, and also to categorize champions based on their class. A not tanky top laner could conceivably still be more tanky than a medium mid laner, so I gave myself some leeway in choosing where each champion fits in the categories above. This means the results below, apart from being a very small sample size, also have a bit of my own interpretation. That being said, they still tell an interesting story, even including error bars.
There was a fairly even mixture of toplaners, but every other role had a predictable lack of variety. Apart from three jungle Nidalees and one lonely Azingy Fiddlesticks, every single jungler built a Warrior enchantment then went tanky. Week 8 brought enormous changes:
The midlane and ADC distribution remained approximately the same. I put Fizz in the tanky mid category since Keane built Frozen Heart and played an unusually tanky role on his team, but he could easily be categorized as medium. Support had a moderate but notable shift, as not tanky champions like Janna were being replaced by medium utility-based champions like Thresh.
Most obvious are the shifts in top lane and jungle. The midrange toplaners like Fizz and Irelia were mostly abandoned in favor of tanks like Maokai and Sion. Additionally, jungle Nidalee almost entirely disappeared, and a number of the hybrid tank junglers gave way to full-tank Nunu, Sejuani, and Gragas.
Overall, teams were doubling down on tankiness, adding tanky top laners and supports to buttress their front line. Tanks scale with each other, since they generally have strong initiation but do not fight well on their own, so the increase in jungle tankiness directly led to increased tankiness in other roles. Teams were not yet buying Cinderhulk on top laners, but it is plain to see that buffing tankiness anywhere leads to increased tankiness everywhere.
Besides the granular team composition statistics, I also wanted to check how much success tankier teams were actually having. To that end, I categorized team compositions in a similar way to champions:
- Tanky teams have multiple tanky champions and at most one not tanky champion. I also generally regressed the support’s category to medium for this definition, unless the support was a primary source of close-range initiation or built significant non-supporting items. Occasionally a utility-based midlaner like Orianna or Lulu was in a composition with three big tanks, so I would situationally bump those compositions up to tanky as well.
- Not tanky teams have at least three not tanky champions and at most one tanky champion. This was also somewhat of a judgment call. I wanted tanky teams to generally only have one really vulnerable target and not tanky teams to have at least 3 or 4.
- Medium teams are everything in between.
- Also, I omitted Team Coast’s results from these charts and from the team composition charts above. In fairness, Coast was ahead of the curve in some strategic elements, but their teamplay was so bad that I worried they would skew the results.
Weeks 3 through 7
|Not Tanky Loses||Medium Loses||Tanky Loses|
|Not Tanky Wins||9||10|
Five weeks is a lot of League of Legends time. There are patches, theory shifts, the inevitable fall of Team Liquid. It is difficult to accurately draw conclusions about the game because early in the season style of play shifts so fast. That being said, the most obvious part of this table is the total lack of tanky team compositions. This is due in part to the lack of tanky junglers, but besides that, teams were going with not tanky champions in nearly every role. As could probably be expected, the most common team compositions were midrange. There were no obvious imbalances in how different team compositions performed against one another.
Week 8 through Playoffs
|Not Tanky Loses||Medium Loses||Tanky Loses|
|Not Tanky Wins||4||6|
This data set is far from perfect. Even with 41 games, it is still a relatively small sample. Including the playoffs means that a few teams are disproportionately well-represented. Some of the games could have had pre-planned team compositions, constructed to give their opponents a different look in a long series. However, the overall trend is fairly clear. Most of the not tanky team compositions are turning into medium compositions, and the newly present tanky team compositions are having a lot of success against everything. Mid range team compositions are struggling, unable to lock down the high damage and mobility of not tanky teams, and unable to bust through the powerful exterior of tanky ones. I suspect that a large part of this apparent struggle is attributable to the more well-prepared teams having highly specialized team compositions ready against the less committal midrange compositions, but it is hard to ignore the big numbers in the “medium loses” column above.
Apart from the quantitative changes listed above, there was a very qualitative difference among the team compositions that I noticed while going through the games. Whereas before week 7, I would often have to choose whether to categorize a team as not tanky or medium, many teams from week 8 on would be on the border of medium and tanky. On the whole, I tried to err on the side of medium, and only categorized two teams differently from each other if there was an obvious qualitative difference in tankiness between them. Thus, though it does not appear in the charts above, even the overall tankiness of medium shifted upward after 5.5.
Tankiness in League of Legends is stability. Tanky champions struggle to kill and to be killed, so tankiness is an important attribute for a team that values consistency. The teams that recognize this are the well-prepared teams that are most likely to be successful anyway, so it is hard to put total faith in numbers like the win rates above. That being said, Cinderhulk and its accompanying playstyle are likely here to stay, and it will be interesting to watch how teams punish or take advantage of this going forward.
I wanted to examine the two games in which one team played a not tanky style and the other played a tanky style. Both of the not tanky sides were played by TSM in the Spring playoffs, and TSM lost each one. However, in both games they had a chance of victory, and the games themselves offered an instructive view into how to fight with and against tanks.
TSM vs. TIP Spring Semifinals Game 1
TSM had an interesting choice for their last pick. TIP committed to a tanky composition with their first picks of Nautilus and Nunu, to which TSM responded with a pair of AD carries. Then TIP doubled down with another tank and Sivir, a carry who grants significant initiation power to TIP’s tanks at the expense of damage. Gragas was open for TSM, which would have allowed them to have some semblance of a front line to stand ground against a low-damage TIP team. Instead they picked Nidalee, who gave TSM a huge advantage in early game initiative and some midgame poke, but also fully committed them to a mobility-based guerilla fighting style. Then, when TSM lost their initiative due to strong early game play from TIP, TSM spent the rest of the game giving up objectives and trying to avoid TIP’s powerful advantage in a stand-up fight. Even so, it was nearly impossible for TIP to push TSM’s base and to win the game, even with five dragons and baron. Throughout the game, TSM showed an ability to significantly harm the tanks of TIP, but they were rarely able to actually finish them off. Although the game showed some promise for the tankless style, it also showed exactly how brutal multiple strong tanks can be in the midgame if they get ahead.
TSM vs. C9 Spring Finals game 1
This time, rather than multiple sources of moderate damage, TSM picked a high mobility composition designed to support their hypercarry Jinx, who is one of the premier tank-killing AD carries. They also opted for the Gragas pick this time over Nidalee, giving them a tank to rally around if necessary in a stand up fight. Cloud 9, on the other hand, picked a Sivir composition including a tanky top, support, and jungle, with off-tank Urgot in the mid lane. Once again, the stage was set for a team based on damage and mobility to fight a team based on tankiness and initiation. This time, TSM was able to use their early initiative to amass a moderate early-game lead in gold and dragons, and it looked like Cloud 9 was going to have a difficult time chasing down a TSM team capable of picking them apart while kiting away. However, a pair of excellent Sejuani ults, including a miraculous 5-man catch after a baron bait, allowed C9 to close the gap then take over the game. TSM played well for most of the game, maintaining space between their carries and C9’s frightening tanks, but TSM’s two mistakes were enough for C9 to chase them down and take the game. Overall, TSM gave a more promising and refined performance this game than the previous one against TIP, but C9 was still able to parlay TSM’s small positioning errors into a victory. This game was not quite a proof of concept, but at the very least it showed promise for high damage, low tankiness compositions against Cinderhulk tanks. However, the fact that this is the only game TSM lost in the final, and previous game one was the only game they lost in the semifinals against TIP, suggests that for now, it is very difficult to defeat heavy tankiness with heavy damage.