Runes and the Ashe Rework

Day one of EULCS is behind us, and, as is the tradition in such things, it is time for the massive overreaction to small samples. Elements is back! Origen is going to worlds! SK is done! Roccat…uh…still sucks! But if we fire up the way-back machine, we can see that just one split ago, Giants kicked off their season with a 2-0 start — immediately before a six-game losing streak and 6-13 final record. The generally hapless Copenhagen Wolves beat eventual third place H2K, and on day 2, the totally hapless Elements beat spring darlings Unicorns of Love. The moral of this story is that early season results, while meaningful in the standings, are often meaningless for actual team analysis, especially in the lawlessly kooky EULCS. But that does not mean that we cannot glean any information from day one results; rather it means that we have to get more granular. Can we guess what picks and styles will emerge in this split, going into playoffs and eventually worlds? Tower pressure from Blue Side Azir was terrifying. Jungle Evelynn made a resurgence in games in which the big two of Gragas and Rek’Sai were banned. And, of course, newly reworked Ashe made a two-game cameo, winning in the capable hands of Fnatic’s new old AD Carry Rekkles and losing in the less capable hands of Giants’ Adryh.

As of Patch 5.9, Ashe got sizable changes to all of her skills besides her Ultimate. According to the patch notes:

We really wanted to solidify Ashe as the utility markswoman of League of Legends

From just a two game sample, it is difficult to evaluate something as nebulous as Ashe’s “utility,” but it is fairly clear that the teams playing Ashe value her damage highly — in both games she was played, Ashe was the only significant source of sustained damage for teams full of mobility and disruption. Part of this is Ashe’s new Ranger’s Focus, giving her a powerful and reliable damage steroid. Another part is Ashe’s new passive, which can lead to some interesting interactions with items, masteries, and, of course, runes.

From the League of Legends Wikia:

Frost Shots
INNATE: Ashe’s basic attacks and abilities apply Frost to enemies damaged, slowing them by 5 / 11 / 17 / 23 / 29 / 35% for 2 seconds.

Ashe’s basic attacks against frosted enemies always critically strike for modified critical damage, but Ashe otherwise cannot perform critical strikes.

TOTAL DAMAGE: 110 + (%Critical Strike Chance × (1 + Bonus Critical Strike Damage)) %AD

Crucially, Ashe’s basic attacks critically strike on all targets, as long as they have been previously Frosted — by basic attacks, Volley, or Enchanted Crystal Arrow. This leads to some interesting results. For example, the magic damage portion of Statikk Shiv will always critically strike for Ashe’s modified critical damage on any Frosted target, adding consistent burst damage. In the offensive Mastery tree, Frenzy is instantly stackable, regardless of Ashe’s critical strike chance, and the now-crucial Reinforced Armor Mastery allows Ashe’s enemies to essentially reduce her damage by 10%. And, of course, there are new runes to investigate. From lolesports, here are the rune pages of the two Ashe users:

Rekkles

3 X +4.5% attack speed Quintessences

9 X +0.95 attack damage Marks

9 X +1 armor Seals

9 X +1.34 magic resist Glyphs

Standard enough. Apparently AD Carries are running two or three Attack Speed quints for tank busting nowadays.

Adryh

2 X +4.5% attack speed Quintessences

1 X +4.26 armor Quintessence

9 X +0.95 attack damage Marks

4 X +8 health Seals

5 X +1 armor Seals

4 X +0.64% attack speed Glyphs

5 X+1.34 magic resist Glyphs

Goofy! Looking back at the Spring Split, it seems that Adryh likes to be a little bit off the wall with his rune choices; for example, he runs Mana Regeneration runes on a lot of his champions. However, there is important commonality with Rekkles — namely the attack speed quintessences and the attack damage marks.

Now let me introduce you to this guy:

GMarks_(2)+ 0.93% critical strike chance

Notorious for its gamechanging inclusion in physical damage rune pages, Critical Strike runes have, for most champions, an interesting tradeoff. They scale multiplicatively, compared to the linear scaling of physical damage or armor penetration, and they do not significantly decrease theoretical damage per second compared to the other runes in the early game. However, for most champions, these runes exchange early game reliability for overall damage, a trade which does not fit the precision required at the highest level of League of Legends. Ashe, on the other hand, can make immediate use of these runes.

From nine Physical Damage marks, Ashe gains 8.55 attack damage, bringing her base total without Masteries to 59.55. Adding a Doran’s Blade to start gives her 66.55 AD, and with her passive, Ashe will deal 73.21 damage to Frosted targets. From nine Critical Strike Chance marks, Ashe gains 8.37% critical strike chance, and, using the fancy math from Ashe’s Frost Shot, every autoattack against a Frosted target will deal 118.37% of Ashe’s AD to an enemy, meaning she deals 68.65 damage to Frosted targets with a Doran’s Blade. In complicated chart form: 

Damage Crit Chance Crit Damage Modifier Unfrosted Autoattack Frosted Autoattack W + Auto Combo
AD Marks Crit Marks AD Marks Crit Marks AD Marks Crit Marks AD Marks Crit Marks AD Marks Crit Marks AD Marks Crit Marks
Base 59.55 51 0 8.37 1 1 59.55 51 65.50 60.37 165.05 151.37
Dorans 66.55 58 0 8.37 1 1 66.55 58 73.20 68.65 179.75 166.65
Infinity Edge 146.55 138 20 28.37 1.5 1.5 146.55 138 205.17 210.52 391.72 388.52
Infinity + Statikk 146.55 138 45 53.37 1.5 1.5 246.55 238 260.12 262.27 624.17 630.33

How to read: Every row represents a different item threshold, and every column is a statistical value based on those items and either AD or Critical Chance runes. The final pair of columns evaluates the damage of a level 1 Volley to apply Frost Shots and then a single autoattack, which is Ashe’s thematic combo. Don’t worry too much about the table; it is just here for the analysis that follows.

First of all, in the early game, physical damage marks are better against both Frosted and unFrosted targets. Ashe will struggle to trade until she finishes her Infinity Edge. Secondly, it is equally clear that once Ashe finishes her Infinity Edge, all autoattack trades will favor the Ashe with critical chance runes. As soon as Ashe gets her Statikk Shiv, Ashe’s Volley + Autoattack Combo is significantly better with critical chance runes. Building more physical damage will widen this gulf, but paradoxically, additional critical chance actually hits a point of diminishing returns, as we notice from the decreased Frost Shots damage between Infinity Edge and Infinity + Statikk. If Ashe is running critical strike runes, she needs to build more physical damage before more critical chance after Statikk Shiv to maximize her total damage.

On the whole, it is unclear that critical strike chance marks are better than physical damage marks, even though they are in this case just as reliable. On the one hand, Ashe’s lategame is significantly improved with critical chance runes, provided she never allows herself to surpass 100% critical strike chance, but on the other hand, it takes a significant item threshold before the critical strike runes equalize with physical damage. Her Volley will always be worse, and she will have a slightly more difficult time last hitting, which can both lose gold and make her laning phase more precarious. Also, physical damage runes play nicer with certain Masteries, especially Warlord. Overall, physical damage marks are probably still preferable, but I know for a fact I will be running a critical chance rune page on Ashe in Solo Queue very soon. If I like it, maybe I will even try some glyphs or quintessences, efficiency be damned!

As a final note, I want to mention critical strike chance’s black-sheep cousin, the critical strike damage mark. In my perfect world, some bizarre mix of critical chance and critical damage is optimal, giving Ashe some early strength combined with late game dominance. Unfortunately, each critical damage rune weakens Ashe’s early game even further, and they do not become relevant until well over 50% critical strike chance. The tradeoff is almost certainly not worth it, especially because we already handicap Ashe’s early game with the critical chance runes. If critical strike chance marks are for Solo Queue, maybe critical damage are for Normal Games. Or maybe just use Quintessences…

Anyway, enjoy LCS today, and maybe something else will catch my fancy from today and I’ll have another update tomorrow! But let’s not dream too much.

Tanks, Junglers, and Tanky Junglers — Understanding the Cinderhulk Changes

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.

Week 6

Tanky Medium Not Tanky
Top 6 6 4
Jungle 12 4
Mid 2 14
ADC 16
Support 2 2 12

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:

Week 8

Tanky Medium Not Tanky
Top 10 1 5
Jungle 9 6 1
Mid 1 2 13
ADC 16
Support 2 5 9

Graphically:

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
Medium Wins 7 14
Tanky Wins

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
Medium Wins 1 19 2
Tanky Wins 2 5 2

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.

 

Bonus

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.

Hai Lam, A Great Player, but…

Cloud 9 announced this week the departure of longtime midlaner Hai “Hai” Lam from their active roster, ending weeks of speculation about the status of the Cloud 9 mid lane. Teammates, opponents, and personalities rushed to laud him on Twitter, while rampant speculation about his replacement abounded on Reddit. A number of outstanding analytic retrospectives are also making the rounds, highlighting his brilliance as a tactician, yet full of the implicit “…but” that will likely always be a large part of Hai’s legacy.

“Hai was a great leader, but…”

“Hai captained the most dominant team in NA history, but…”

“Hai’s elite shotcalling won C9 nearly unwinnable games against the top teams in the world, but…”

After the “but…” comes “he wasn’t that great of a player,” or “he couldn’t match up with the best mid laners.” Inevitably, studies on Hai’s legacy highlight his value as leader and especially shotcaller of Cloud 9, and they leave out his actual play. Perhaps this is a bit unfair; for most of his career Hai at least went even with the titans of his day, first Reginald and mandatorycloud, then later Bjergsen, Shiphtur, and XiaoWeiXiao. Though lacking, by intentional design, in the kinds of unbelievable solo plays that have come to synonymize great midlane play, Hai executed a different kind of dominance. He used the pressure generated by the threat of his superlative jungler to fuel his aggressive play in lane, which in turn allowed that jungler the space to impose Cloud 9’s map-wide stranglehold. Early in Cloud 9’s LCS run, this often manifested in a subtle advantage; Hai maintained lane parity against his opposing laner despite rarely getting blue buff or significant early game jungle help. Occasionally though, Hai was able to turn this small advantage into spectacular results — for example in Hai’s brilliant performance during 2013’s Battle of the Atlantic. In a rematch of C9’s 2-1 World Championship defeat at the hands of Fnatic, Hai leveraged Cloud 9’s early midlane pressure into a surgical evisceration of the European titleholder with brutally efficient assassin play. In the process, he stymied superstar midlaner xPeke’s attempts to counterpick him with Renekton against Kassadin and to outcarry him with Orianna against Kha’Zix. It was a terrifying look at what could happen when the entire Cloud 9 apparatus focused on their selfless midlaner.

As Hai and Cloud 9 evolved, Hai took his unique brand of carrying by deferral to an extreme. While his rival mids consistently led their teams in gold and kills, Hai took a different path, rarely pressuring nor permitting pressure in lane, picking Soraka and Lulu while his opponents chose Ahri and Nidalee. Although his team enjoyed nearly the same level of success, weaknesses showed in Hai’s individual game, continuing into subsequent splits. Despite his shift to more supportive champions, Hai’s KDA suffered, beginning a string of consecutive splits in which he was last on his team by that metric. Hai also occasionally struggled in lane, losing farm and dying to ganks at a much higher rate than ever before. Crucially, it was becoming no longer safe to leave Hai completely alone in the midlane, nor to funnel his resources to teammates. And, though his signature Zed remained strong, it eventually became a crutch, without which he sometimes imploded. In Cloud 9’s 2014 Summer Split final against TSM, Hai infamously recorded a 10 KDA ratio in his two Zed games, and a combined KDA ratio of less than 1 in the other three, drawing a game 5 Zed ban which may have decided the series. And though his meteoric climb up the Korean ladder before Season 4 worlds gave fans hope, his questionable performance at the actual event led to even more questions about his role on the team.

Yet despite his individual struggles, this was paradoxically the most brilliant and influential period of Hai’s career. Partly because of Hai’s diminished condition, Cloud 9 could no longer simply brute force their way to victory on the back of superior teamplay aided by a world-class jungler and AD Carry. In their early splits, in the rare case when Cloud 9 fell behind, they could usually expect to capitalize on missteps from their opponent to turn the game around. Certainly this reflects an outstanding grasp of strategy and decision-making from their leader, and their opponents were almost totally unable to cope with their forward-thinking approach to objective control, but in the early days Cloud 9 never had to make difficult compromises. Rarely did Cloud 9 ever have to trade away objectives or map control, because they usually had an ironclad grip on the map. As their opponents began to catch Cloud 9 strategically, and surpassed some of them in individual skill, Hai’s tactical mastery showed brightest. Cloud 9 became a team of unstoppable split pushing, opportunistic barons, favorable objective trades, and nearly perfect map-wide coordination of resources. In Summer 2013, Cloud 9 came back to beat Vulcun from a huge gold deficit by brute force; they punished positioning errors and outplayed their opponent in lategame teamfights. One year later, in Summer Split 2014, Cloud 9 came back from a similar deficit to defeat Counter Logic Gaming, but this time they won based on manipulating their enemy into dubious objective trades, avoiding or disengaging teamfights altogether. Although CLG won most fights and outperformed Cloud 9’s players individually, Cloud 9 never let CLG exchange their resource lead for an objective lead, and C9 eventually won based on superior lategame movement throughout the map. That split, as Hai’s skill level diminished and their opponents improved, Cloud 9 lost more games than in both their previous splits combined, but their mastery of League of Legends grew substantially, catalyzed by the need to find new ways to stay ahead of their rivals.

Cloud 9’s performance in their two World Championship appearances reflected their change in style. In 2013, a regionally dominant Cloud 9 failed to advance past quarterfinals, victim to Fnatic’s strong individual players and concerted commitment to jungle control. Cloud 9 came into the matchup expecting to simply outclass Fnatic as they had done to all of their North American opponents, only to fall embarrassingly, with Hai in particular ominously failing to make an impact. By 2014, a diminished Cloud 9 advanced from group stages against more difficult opponents by finding their opponents’ pressure points and exploiting weaknesses with coordinated precision. It was a testament to Hai’s prodigious skill as a leader, just as it was an indictment of Hai’s skill as a player. Cloud 9’s breathtaking comeback attempt against Najin White Shield was a masterpiece of coordinated pressure, and there cannot be a more quintessentially “Hai” game than Game 2 of the quarterfinals between Cloud 9 and Samsung Blue. Cloud 9 lost in kills 5 to 29 and was utterly outclassed every time they fought. Yet against a team considered at the time perhaps the strongest in the world, Hai repeatedly pulled the Samsung team out of position, bringing his team back into striking distance in a hopeless game, despite his horrific 1/10/1 score. The commentary, crowd, and clash of styles in this game combine symphonically into an amazing exhibition of League of Legends. In the middle of it all stands Hai, outplayed and outclassed, yet still nearly managing to snatch victory from the jaws of defeat, only a few pixels away from stealing an inhibitor and keeping his team’s resuscitated hope alive.

In Spring 2015, Cloud 9 failed to finish in at least a tie for first place for the first time since they joined the LCS. In the first match of the year, Hai was astonishingly solo killed in lane by TSM’s Bjergsen. This is perhaps fitting, as Hai probably suffers more than anyone else by comparison to TSM’s superstar midlaner. The centerpiece of their respective teams, Hai and Bjergsen have squared off in the midlane of every LCS final since Bjergsen joined TSM in Spring 2014, and Cloud 9’s diminishing results have mirrored Hai’s loss of midlane control. In Spring 2014, Hai did not die a single time in the finals, as his team annihilated TSM 3-0. By Spring 2015, merely two splits later, Hai only managed an execrable 1.13 KDA ratio, worst on his team, as Cloud 9 barely managed to win a single game against the TSM juggernaut. His decaying wrists and deteriorating team environment surely played a part, but whatever the reason, Hai could clearly no longer compete with the Summer MVP, and the mid lane mismatch helped decide the split in favor of Cloud 9’s ascendent rivals.

In the end, Hai should not be remembered as the deficient midlaner he became, wracked by injury and lack of confidence, nor as the virtuoso conductor of unchallenged LCS champion, fifth best player on both his team and in his region. Rather Hai’s legacy is to be the inevitable victim of his own stratospheric success. Hai rarely overcame his opponents with dazzling individual mastery, but instead he wielded his team like a scalpel, carving at his opponent’s weak points until they collapsed under their own weight. Once his opponents discovered they could not beat him with skill alone, they too began to search for weak points to exploit. Cloud 9, Hai at the head, birthed the evolutionary process that allowed the best North American teams and players to pick themselves up by their own bootstraps, eventually leaving Hai behind. So yes, Hai was a bit deficient, and his underwhelming individual play became Cloud 9’s biggest weakness, but…

No North American LCS player has ever been more central to his team and region’s stupendous strategic growth, and Hai can absolutely retire as one of the greatest ever.