MSI Preview: Irreconcilable Differences - A Statistical Comparison of MSI Teams' Play Styles
When you bring together the top teams from each region, each drawing on their own region's current metagame flavour, there are bound to be some clashes in play style. Different League of Legends teams approach the game in different ways: some teams avoid fights and work to out-rotate their opponents, sieging lanes and killing towers so they can win by starvation. Other teams try to kill everything that moves. Some teams just try to keep the game close until they can land a perfect initiation, mop up a team fight, and convert that into a Baron or an inhibitor.
In this article, we will specifically be looking at MSI participants' play styles: how does each team prioritize objectives; how do they distribute their resources; and to what extent do they play a controlled, objective-focused game versus a bloodthirsty, fight-heavy game? The numbers will give us a clearer idea of what to expect when the regions clash in Tallahassee this weekend.
The statistics in this article are based on each team's regional playoffs, with the IWCI playoffs used for Be?ikta? e-Sports Club. The numbers do not indicate which teams are "better" or "worse"; they are only a comparison of style and approach, and relative success in their own domestic regions.
Interested in comparing individual MSI players? Check out my previous article, which covers the players' laning phases and what to expect in the head-to-heads over the first 10 minutes.
Note 1: Unfortunately, statistics for Edward Gaming are not available because the LPL does not release its match histories publicly. Insights into EDG's play style are based on the excellent work of Kelsey Moser of theScore eSports, including her recent article, MSI Preview: the 5 Stages of Edward Gaming.
Note 2: Due to a Match History error, Game 1 of Be?ikta?'s IWCI Semifinal is not included in these statistics.
The laning phase can sometimes decide the outcome of an entire game. Depending on team composition, a team might be trying to build a big early lead, or they might just be trying to survive until their late-game threats, like Kassadin or Kog'Maw, can start rolling.
Three early game "firsts" can be used to describe each team's approach to the early game. The teams' First Blood rate, First Dragon rate, and First Tower rate are charted below. Through these numbers we'll get an idea of how each team prefers to play their early game, and how successful they are in that approach.
The number that stands out most for TSM is their poor control of the first Dragon. They were excellent at grabbing the first kill and first Tower, but most of the time they allowed the first Dragon to slip. TSM's first Dragon rate was this low all split long, which suggests that they may have intentionally given the Dragon low priority early on. Instead of fighting over the Dragon, TSM took their first Tower kill at an average of 10.8 minutes, the fastest time among the listed teams. Alongside those Towers, TSM also used their fast pushing to accumulate an average 12.0 CS lead at 10 minutes (all team members combined). TSM's priority on pushing and Towers led to an average gold lead of 848 at 10 minutes, and 2,591 at 20 minutes.
SKT took a very different approach from TSM: they still took the first Tower in five of their eight games, but on average their first Tower kill didn't come until 17.3 minutes! Instead, they secured first Dragon three-quarters of the time. Because of their higher Dragon priority and lower Tower priority, along with an unimpressive 50% First Blood rate, SKT had much smaller gold leads: on average, they led by 204 gold at 10 minutes and 1,072 gold at 20 minutes.
Fnatic had excellent early objective control, including an impressive three pre-10-minute tower kills in 10 games. They were able to successfully secure early kills without giving back deaths: at 10 minutes, they averaged 1.0 more kills than their opponents (compare to kill advantages at 10 minutes of 0.5 for TSM and 0.6 for SKT). Because of their early combat focus, Fnatic was behind by 2.1 CS as a team at 10 minutes, on average, but had big XP advantages. A big question for Fnatic is whether their early gank and dive efforts will come off successful, or whether they will blow up in Fnatic's faces against international competition.
For AHQ, strong objective control didn't translate into overall leads, with an average gold lead of only 21 gold at 10 minutes, combined with CS and XP disadvantages. But through very early first Dragons (at 6.4 minutes on average), they were able to secure non-gold advantages that set them up for the mid game, where they turned on the jets. Despite being very close in gold at 10 minutes, AHQ had massive leads at 20 minutes, up by an average of 2,591 gold.
Be?ikta? had pretty poor early game control, on average, giving up First Blood in two-thirds of their playoff games and only taking half of the first Dragons and first Towers. On average, their first Tower kill didn't come until 15.7 minutes, and they only had an average gold lead of 241 at 10 minutes. (In fact, at 20 minutes they were behind in gold by 588, on average!) Against international competition, these weak early games could very well put Be?ikta? too far behind to recover.
According to Kelsey Moser, EDward Gaming tends to have poor Dragon control, but earlier in the season they had good success with early roaming by their Jungler, Support, and sometimes Top Laner. That could lead to CS disadvantages but potential gold advantages if any of the ganks or dives are successful. On post-5.5 patches, though, Moser paints a picture of EDG as a "stall-and-comeback" team, content to give up small advantages early in favour of mid and late game team fighting.
As MSI unfolds, watch closely for teams' early objective prioritization. Expect TSM to put high value on knocking Towers down early, while SKT and AHQ set themselves up to control the Dragon. EDG may be content to just farm into the mid game, no contesting Dragons too heavily. Fnatic may look to skirmish, gank, and dive, while Be?ikta? will be doing their best to simply hang on. Each team's respective early strategy will play a big role in how successful they are overall. If TSM can get their Tower kill gold flowing, they will be ready for team fighting in the mid game. If Fnatic can use their aggression to secure big early leads, they may be able to steamroll the mid game through sheer item advantages. If SKT can control the Dragon buff, their above-average early game will quickly turn into a Dragon-buff-enhanced ticking time bomb of an advantage. AHQ and EDG, meanwhile, will be counting on just keeping things close, allowing their excellent coordination in team fights to propel them to wins in spite of potential gold deficits and Dragon buff disadvantages.
The Combined Kills per Minute statistic shows how bloodthirsty a team is: the more aggressive they play, the more they will rack up both kills and deaths. In the chart below, Combined Kills per Minute is broken out by kills and deaths (or rather, opponent kills, which helps us out by excluding those annoying Sion jungle suicides!).
It shouldn't be too surprising that the teams from the LMS and the IWCI did the most fighting: it has been a general trend in professional League of Legends that the more developed the scene becomes, the more controlled the teams become.
AHQ's and Be?ikta?'s 0.90 and 0.86 combined kills per minute stand pretty clearly apart from the other teams. Fnatic takes a middle ground in this aggression measurement, while SKT and TSM fought much less often.
If we express these kill rates as ratios, to add a little more flavour, we see that AHQ (93% more kills per minute than their opponents) and TSM (90% more kills per minute than their opponents) had the most efficient fighting, on average. SKT had 68% more kills per minute than their opponents, while Fnatic had 44% more and Be?ikta? only had 19% more.
Without having actual numbers for EDG, it's difficult to know where they would land on this statistic, but with China's reputation for constant aggression on one hand and EDG's recent penchant for Baron control and tower sieging on the other, we might expect them to land somewhere in the middle.
AHQ was very aggressive, but also very efficient in their aggression, constantly starting fights and coming out ahead, judging by their kills per minute and opponent kills per minute. Based on these numbers, the LMS champions dominated the fighting in their region's playoffs, snowballing their mid game leads into more and more team fight wins. Be?ikta?, on the other hand, fought often, but often took back as many kills as they dealt out, barely coming out ahead. Their IWCI victory wasn't as dominant as what AHQ achieved, and small victories might not be enough for Be?ikta? to pull off upsets against the other regions' representatives. During MSI, we can expect AHQ and Be?ikta? to work really hard to find fights throughout the game, but given the level of competition they're facing, those fights may either not materialize due to strong disengage or better positioning by their opponents, or may result in more losses than wins, especially for the Wild Card qualifiers.
Look for Fnatic to also fight early and often, while TSM, SKT, and possibly EDG take a less bloody road. The battles between the aggressor teams and the more controlled teams could largely come down to whether the engages and pick attempts can be successful against the disengage toolkits and potentially superior vision control.
No matter how much or how little each team fights, combat is an integral part of League of Legends, and carries, in all their flavours, are at the core of combat. LoL team compositions can be set up any number of ways, with varying numbers of tanks and carries and in-betweens. Looking at each team's damage share, or how their damage to champions is distributed across their team, helps us to see what styles each team favours. Who are the team's "high value units" (to borrow a phrase from rSquared's naval-based LoL strategy article), and how much value is placed on each of them? Gold share adds to this by showing how teams distribute their resources, and which carries they prioritize. We will also support these numbers with actual damage per minute numbers for Mid Laners and ADCs.
Among the five teams we have data for, Team SoloMid is by far the most two-carry-based team. TSM's Top and Jungle only combined for an average of 28.9% of the team's damage output, a tiny proportion, especially when compared to Fnatic's huge 42.8% combined damage contribution from Huni and Reignover.
Fnatic received the lowest damage share of these teams from their Mid Laner and AD Carry, and from the damage per minute numbers we can see that it wasn't just because Huni and Reignover did more damage: Febiven and Steeelback had fairly low damage output in general. Febiven and Steeelback also had relatively small gold shares for a Mid Laner and ADC.
Unsurprisingly, TSM relied mostly on Bjergsen for their damage output: his 34.1% damage share is the highest among MSI Mid Laners, even without receiving an especially high share of team gold, and his damager per minute numbers were second only to SKT's Faker. Bjergsen's damage averages are definitely skewed upwards by his AP Kog'Maw game, where he did 53.3% of the team's damage on the back of 1,286 damage per minute. AP Kog'Maw does a lot of poke damage! WildTurtle also had a high damage share, but his damage per minute numbers were unremarkable.
The most ADC-reliant team at MSI, based on these numbers, is SK Telecom T1, especially when playing with Faker in the lineup. Bang didn't have mind-blowing damage per minute numbers, but his damage share was noticeably higher than other ADCs. We have to be careful with these numbers, though, since SKT only played eight playoff games in total, four each with Easyhoon and Faker. The big drop in Jungler damage share between the games with Easyhoon vs. the games with Faker is probably due to champion selection: Easyhoon played alongside Gragas once, Nunu once, and Sejuani twice; while Faker played alongside Nunu once and Rek'Sai three times. Gragas and Sejuani are two of the highest damage-per-minute jungle champions, while Rek'Sai does a relatively low amount of damage, especially as games get longer.
AHQ distributed their damage sources pretty evenly, relying more on their ADC than their Mid for sustained damage, though this is in part due to Westdoor's usage of champions like Cho'Gath, Fizz, and Zed who do low damage per minute, but big burst damage. AHQ didn't put much gold into their Top and Jungle.
Be?ikta? got pretty big damage contributions from their Top Laner and Jungler, leading to a pretty even damage distribution for them, as well, paired with fairly even gold distribution.
According to Kelsey Moser, EDG prioritizes their ADC, Deft, very heavily as their main carry. Mid Laners Pawn and U are powerful secondary threats, but when it comes to the crucial mid and late game team fights EDG's peels and shields are reserved for Deft. Based on Moser's observations, we might expect EDG to have smaller damage shares in Top and Jungle, though not to the extent of TSM.
TSM will probably continue to focus their carry responsibilities onto Bjergsen and WildTurtle, while Fnatic relies on Huni and Reignover to shoulder more of the burden. SKT's damage distribution will depend somewhat on whether Easyhoon or Faker is playing, with higher priority on Faker to carry than on Easyhoon. With either Mid Laner, Bang will be an important piece of SKT's success as one of the most prioritized ADCs at MSI.
Without actual numbers for EDG, and especially for Pawn, U, and Deft, we can't make the same claims about how they prioritize their damage output and gold. But based on expert opinions, we should look for Deft to shoulder more of the carry load, while Pawn and U provide secondary carry potential and Koro1 and ClearLove do a more or less average proportion of team damage for their roles.
Each MSI team has its own approach to the early game, how often they fight, and how they build their team compositions. Here are the TL;DRs on how each team might approach their games, judging by their statistics, and what that might mean against their MSI opponents.
TSM will be looking to play a fast-push, Tower-priority early game to gain a mid game advantage that they can systematically snowball into medium-aggression, objective-heavy victories. Expect that style to be challenged by early aggression from Fnatic, mid game aggression from AHQ, and late game execution from EDG. The fight for Dragon control against SKT could result in some very interesting tactical tradeoffs between the two teams.
Will Fnatic be able to get Huni and Reignover rolling, or will they be forced to rely on Febiven and Steeelback to shoulder a heavier load as damage dealers? Fnatic's aggressive early game will be incredibly important, and if they aren't able to secure big enough early leads they could find themselves in big trouble against the mid and late game execution of TSM, SKT, AHQ, and EDG.
SK Telecom T1
SKT should be able to improve on their 50% First Blood rate from the playoffs when playing against international opponents instead of stronger domestic competition in the LCK. Will that be enough of an improvement for them to carry more momentum into the mid game? If Bengi continues to play a lot of Rek'Sai, SKT's early game should trend towards more aggression, especially when Faker is in the lineup. That aggression could serve SKT well against AHQ and EDG, helping to build the early leads they'll need to overcome those teams' mid game team fights, and could lead to an interesting clash against Fnatic, especially if SKT is able to use their more balanced approach to fight frequency to deny AHQ and Fnatic the kind of frantic pace they're used to.
AHQ e-Sports Club
AHQ's big question is how close they'll be able to keep games early on. If they can approach the mid game with only a small deficit, they should find themselves in pretty good shape, especially if Westdoor has been able to pull off any successful roams out of the Mid lane. But if those roams get shut down, and if teams with better vision control and disengage abilities are able to deny AHQ the fights they want, the LMS representative will quickly find themselves on the outside of the tournament looking in.
Be?ikta? e-Sports Club
IWCI representatives are always massive underdogs in tournaments of this calibre, and if Be?ikta? wants to pull off any upsets they'll have to shore up their early game and perhaps temper some of their aggression, looking for smarter fights. If they continue to throw themselves at their opponents, expecting to win by taking marginally more kills than deaths, they're quickly going to find that their opponents will adapt, take one decisive team fight victory, and turn that into a major objective advantage and a win.
EDward Gaming has to be careful not to take their opponents lightly, especially in the early phases of games. Their tendency to stall into the mid and late game and win through team fighting could be exploited by relentless early aggression on the part of Fnatic or early Dragon control by AHQ. But EDG has spent an entire split dealing with LPL teams who tried to apply exactly that approach to beating them: on the MSI stage, how big of an early lead will their opponents need before it's too much for EDG to overcome?
As always, statistics can be interpreted in many different ways and are based on very complicated context. Please share your own interpretation of these numbers! Discussion is always a good thing. And watch OraclesElixir.com for MSI team and player statistics as the event unfolds.