Identifying strategies of eSports players at various proficiency: the case of League of Legends.

Identifying Strategies of Esports Players at

Various Proficiency: The case of League of

Legends

Thesis Presented by Lawrence Edmondson

to the Department of Art + Design

In Partial Fulfillment of the Requirements for the Degree of Master of Science in Game Science

and Design in the Graduate School of the College of Arts, Media and Design

Northeastern University

Boston, Massachusetts

May, 2021

Abstract

Developing effective strategies in esports is paramount for top tier playing. However,

despite being core to gaming proficiency, strategy development for competitive gaming has been

under represented in esports research. Here, I look at strategy development in esports by

considering the specific case of League of Legends (LoL). Specifically, I identify strategies used

by LoL players at various degrees of proficiency (i.e., from top tier to low tier) by analyzing 20

YouTube videos of professional LoL tournaments and regular ladder games. Through content

and thematic analysis, I identify multiple strategies in LoL that include (1) team composition

(hero selection for each team), (2) summoner spells used, (3) champion abilities, and

systematically analyze how they are used by competitive gamers, as well as the techniques used

to deploy these strategies (e.g., wave clearing or split pushing). Results from the analysis show

that players tend to play more aggressive and look to make more offensive plays versus playing

defensive. I conclude by discussing what implications these results have in advancing the

scientific understanding of phenomena related to esports.

Introduction

Electronic sports (i.e., esports) originated in 1980 in the form of arcade video gaming

(Tang, 2020). Because of the growth and evolution of the internet, which in turn has contributed

to the growth and evolution of gaming, esports have become key to gaming culture. In 2016, the

industry accumulated more than 30 billion in revenue, and was expected to bring in 1.5 billion

dollars by 2020 (Wallner, 2019, pp. 1–3). Starting from 2024, esports will be part of the

Olympics. Their notoriety, increasing media exposure, and annual revenue, are beginning to

increase the esports overall consideration by both the media, industry, academia, and elevating

them to the rank of “actual” sports (for a discussion, see Kane et al., 2017).

Hence, we can consider esports like any other professional sport, where developing

strategy and game-planning are key to effective performances and increase chances of winning.

However, despite the increasing interest of academic research around esports (for a review, see

Reitman et al., 2019) the way in which players develop strategies in esports is yet undervalued

and lacks systematic analyses. Here, I consider the case of League of Legends (LoL), a

multiplayer online battle arena game (MOBA) created by Riot Games back in 2009, and explore

how players of different ranking and proficiency develop strategy in their games (see Tang,

2020; Harteveld et al., 2019).

Specifically, I analyze 20 YouTube videos from professional players playing tournament

games of LOL to high ranking ladder games in the solo queue (the normal ladder where

everyone plays) and extrapolate game play strategies by systematically identifying individual

instances of LOL strategy archetypes (e.g., wave stacking, split pushing). This thesis advances

the scientific understanding of how players develop strategies in esports (here LoL) and its

results will be useful to both esports researchers and trainers. I conclude by discussing what these

results mean to esports and possible applications/directions in future work.

Background

League of Legends (aka LoL)

Figure 1.

League of Legends (LoL) is a multiplayer online battle arena game (MOBA) created by Riot

Games in 2009 (Figure 1). The game scope is simple: two teams of five compete against each

other with the goal of destroying the opposing team’s nexus (the heart of your team's base. When

the nexus is destroyed, the game is over). Since its creation, there have been a total of

approximately 64 million people playing the game worldwide and making it one of the most

successful online video games ever (Ramler et al., 2017). The most recent international

tournament of LoL has seen more than one billion hours watched in total (LoLEsportsStaff,

2020), proving the high popularity of this game in esports. LoL has a large competitive scene.

The higher the rank, the more skilled the players will be, and the more refined their strategies

and tactics will be. Recent academic work around LoL has investigated problematic discourses

of toxicity or emotion regulations of players (Kou et al., 2020a; 2020b). Here, I continue

research efforts around esports and LoL by exploring how players develop strategy and tactics in

this game at competitive levels. Next, I briefly overview the game mechanics of LoL.

LoL Game Mechanics

Figure 2.

As said above, the scope of LoL is for two teams to compete against each other and

destroy their respective nexus’. There are two teams consisting of five heroes that are all

controlled by human players. Each hero is different from another and possesses their own

“special” abilities, stats, and lore. Both teams play their game inside a map, which is square and

split diagonally in the middle into two sides in a symmetric fashion; each team owns one side.

This map, which is also known as the “summoner’s rift” (Figure 2), has three lanes: (1) mid, (2)

top, and (3) bottom. Artificial intelligence bots, aka “creeps”, spawn at each team’s nexus and

walk down each lane until reaching the other team’s nexus and (eventually) destroy it. If they

come along an enemy force on the way, they will engage in combat. The goal of the players is to

help push their team into the other base and destroy the nexus with the help of their own creeps.

Related work

Researchers have explored patterns of combat, metagame strategies, tactical awareness vs

operational skills, and team proficiency. Patterns in combat were presented in a data-driven

method which led to successful game outcomes (Yang et al., 2019). Lee & Ramler use a python

script to identify several non-meta strategies that show a consistent advantage over the meta. Xia

et al. uses evaluation indicators for testing gameplay to conclude that tactical awareness affects

the multiplayer killing indicator and the game result more so than operational skills. Team

composition was broken down by Kim et al. and we find out that player proficiency increases

team performance more than team congruency. I add to previous work by supplying data on how

players conceive different strategies as well as supplying data on the strategies that are being

conceived. My thesis is a mix of all of this research. I cover the metagame, combat tactics, and

team composition.

Methodology

Participants

League of legends has 9 ranks: iron, bronze, silver, gold, platinum, diamond, master,

grandmaster, and challenger. For my study I analyzed 20 games that only consisted of players

residing in the top 4 ranks which were diamond, master, grandmaster, and challenger (in order

from lowest to highest). Pro LoL players are technically challenger rank, but professional players

are normally required to be in the top 50 of challenger or higher for a long period of time. In

2019, there were roughly one thousand pro players. This meant that roughly 1 out of every

100,000 players would go pro. That is less than a 1% chance (Stuart, 2021).

All of the games I used can be found on YouTube. I analyzed games from Vapora Dark,

the Challenger replays youtube channel, and games from the 2020 worlds. Vapora Dark is a high

ranking player who has consistently achieved masters rank since 2015. Challenger Replays is a

youtube channel that provides League of Legends replays from players ranked Master through

challenger. The 2020 worlds was the championship series where teams from all of the

professional leagues across the world come together to see who is the best team in the world.

Design

I will be using the 4 levels of war, as recognized by the military, as the foundation of my

thesis. There are 4 levels of war: tactical, operational, strategic, and political. For the sake of

confusing name mix-ups, I am going to refer to the strategic level of war as the logistic level of

war. Each of these levels has their own important function in terms of winning a war; ranging

from the actual decision making and weapons used on the battlefield all the way to relationships

between allies and psychological warfare. I translated those 4 levels of war into a system aligned

with LoL.

Tactical strategy refers to success in battle due to engagements through decision and

actions that create advantages. Tactical strategy includes weapon systems used, potential combat

power, and the achievement of goals on an operational level. In League of Legends this translates

to how well you can operate and control your champion in-game on the battlefield.

Operational strategy is the middle ground found between action and thought that focuses

on design and organization through campaigns (a plan). Operational strategy is more about

designing a plan to conquer a larger objective. In league of Legends this refers to a focus on team

cooperation and chemistry. Operational planning can consist of plays such as team rotations,

wave management, establishing or denying vision, and lane pressure.

Logistical strategy refers to plans that aim for obtaining objectives and coming up with

game plans before the battle starts. It can be confused with operational strategy because it is

closely aligned with campaigns, but logistical strategy is on a higher level. Logistical strategy is

the entire game plan of how you plan on winning the war. This translates to League of Legends

in the sense of champion itemization, pre-game preparation (runes and masteries), team

composition, etc.

Political strategy refers to your relationship with allies, societal issues, national will, etc.

this translates to the mental exploitation/ psychological warfare in League of Legends. Anything

that has a correlation to mental warfare falls under the political level: camping lanes (usually

performed by jungler), communication with the opposing team in-game (or before), giving kills

to your teammates to help them snowball, etc.

Procedure

For my research I will be doing content and thematic analysis (Neuendorf, K. A., &

Kumar, A. (2015). Content analysis. The international encyclopedia of political communication,

1-10.). I will be observing League of Legends games on YouTube and analyzing them for

strategies and tactics. I will create a coding system using the 4 levels of war that recognizes these

strategies and tactics. Then I will be using Cohen’s Kappa to assess the reliability of my scoring.

My coding scheme can be found here.

One limitation I faced was the type of analysis I was doing. Thematic analysis can take a

long time especially in these games where you have to pay attention to every summoner spell

and ultimate used so that you don't miss any. It was also hard to document everything every

player did ex: dropping vision wards. Professional games have a roaming camera view that

automatically moves to action activity. This made those games easier to score because the

camera was not locked. Non-professional games have a camera view that locks on to one player

for the whole game, so it's hard to see what other players are doing. Some champions also have

ultimates that do not act like normal ultimates making them hard to code because those ultimates

can skew the scoring system.

Scoring

Games were scored using the 4 types of strategy: tactical, operational, logistic, and

political. Inside of these categories there are multiple different taxonomies that were scored:

● Summoner spells (barrier, clarity, cleanse, exhaust, flash, ghost, heal, hexflash, ignite,

smite/chilling smite/challenging smite, & teleport): a point was scored for each of these

summoner spells whenever they were activated.

● Defensive ultimate ability: A defensive ultimate would be scored when a player uses

their ultimate while they are retreating, in order to retreat, or to stop the opposing team

from doing something.

● Offensive ultimate ability: An offensive ultimate ability would be scored when a player

uses their ultimate in order to engage an enemy.

● Hybrid ultimate ability: A hybrid ultimate ability would be scored when a player uses

an ultimate outside of battle entirely. An example of this could be if a player uses an

ability to scout for vision or using an ability against a creep (battling is strictly meant for

PvP fights).

● Retreating: Retreating would be scored when a player attempts to flee from a fight.

● Initiate combat: Initiating combat would be scored when one player attacks another with

a 5 second grace period in between each initiation.

● Disable opponent: Disabling an opponent would be scored whenever a player is crowd

controlled (CC’d). When a player is CC’d it will appear over their head and say either:

stunned, airborne, blind, cripple, disarm, disrupt, drowsy, sleep, charmed, flee, taunt,

ground, kinematics, root, silence, polymorph, slow, stasis, stun, suspension, or

suppression.

● Camping lanes: Camping lanes would be scored when any player is unseen by a player

who is in one of the three lanes, and attacks them from the jungle or any route that is not

from the lane directly in front of them.

● Split pushing: Split pushing would be scored when a player moves to a weaker or

undefended lane and applies pressure by pushing the lane in an attempt to gain a tower

advantage while the rest of his team is all together. This is normally executed in mid-late

game.

● Wave stacking: the process of letting your creep wave build up by eliminating a number

of creeps from the opposing creep wave so that your creep wave ultimately does more

damage per second. Eventually your creep wave will begin to stack and push the

opposing creep wave back and back until the creep waves are balanced. Wave stacking

would be scored when a player kills more opposing creeps than how many are on his side

(this also applies when a player kills an entire enemy creep wave).

● Vision Advantage: Placing wards on your side of the map for visibility. Wards make it so

that you can see places on the map that might serve as gank lanes for you or your enemy.

Wards grant visibility and visibility wins games. Ward advantage would be scored when a

player uses an oracle lens and sweeps the area destroying any enemy wards that might be

nearby acquiring vision advantage.

● Team composition: This was scored by tallying one mark for each champion archetype

playing in the game. There are thirteen archetypes: enchanter, catcher, juggernaut, diver,

burst, battlemage, artillery, marksman, assassin, skirmisher, vanguard, warden, and

specialist.

● Mediation: Mediation would be scored when any emotes are displayed or any

unsportsman-like conduct takes place.

Results

Cohens Kappa results

Figure 3.

I used the 4 levels of war as a foundation to recognize different strategies and tactics in

LoL. An inter-rater reliability analysis was performed using Cohen’s Kappa statistic to determine

consistency among raters. The inter-rater reliability for the raters on the Strategy spreadsheet was

found to be Kappa = 0.89 (p < 0.01). This means that both the rater and I came out with very

similar results in our scoring totals. My research has shown that initiating combat, using

champion ultimates offensively, and flash were the three most scored categories. This shows that

players are inherently aggressive and are constantly looking to make plays on the rift. Although

ultimates could be used defensively, offensive ultimates were used a lot more. Flash was used

offensively just as much as it was used defensively. Offensively, flash was used many times in

attempts at making a play (an attempt to gain an advantage on the rift by landing a kill, helping

your teammate land a kill, or by disrupting your opponent). Teleport was the second most used

summoner spell and that is for good reason. The ability to instantly teleport to any place on the

map is a powerful tool in war. Teleport is so important because it allows the player to quickly,

and in some cases stealthily, make big plays. Retreating might not have happened as many times

as initiations, but they were present throughout the whole game, showing that smart

decision-making is still one thing that players recognize as a key to winning. Giving the enemy

team extra gold by dying while trying to force a battle that has a low chance of success is a grave

offense in LoL, and this is why retreating is so important. These were all categorized in the

tactical level of war.

On the logistical level of war, non-professional games (lower ranked ladder games) used

a broader champion pool. This seems to show that the professionals stick to a smaller meta of

champions because they are the most effective, whereas lower ranked games leave more room

for champion diversity. There were also more marksmen champions selected (3 a game) than any

other champion archetype in 7/20 games. This shows their importance because of their lethality.

Marksmen champions are also known as ADCs (attack damage carries), the champions who do

the most physical damage in the game, thus are usually relied on to carry the team to victory. The

second most used champion archetype were vanguard champions. Vanguard champions are tanks

and they are usually paired with a marksman champion in-game in order to protect them from

harm. For the operational level of war, wave stacking usually continued into late game. In LoL,

gold and experience equal power. The more of those two things you have, the more powerful

your champion will be. This is why wave stacking was done throughout the entirety of most

games, because the intake of gold is key to overpowering your opponent. Regarding the political

level of war, a lot more emotes were used in lower rank games. Emotes were used as a form of

communication whether they were letting a teammate know they got out of a fight safely or

taunting an opponent. In the professional games, emotes were mostly only observed at the very

end of a game once a team had destroyed the other team’s nexus.

Discussion

LoL is a real-time strategy game where tactics are the main focus in winning. My

research shows that players tend to play more aggressively versus conservatively. Player

attitude’s are geared towards always looking to make a play and being offensively minded. These

behaviors are aligned with the nature of the game because LoL is meant to be a skirmishing

game between 2 teams. Even the most used summoner spells (flash and teleport) have an

important impact on playmaking which is attributed to offense. Retreating wasn't observed as

many times as its offensive counterparts, but it was close behind showing that even in an

offensive-oriented game smart decision making is still as important as playing aggressively.

Xia et al. uses the mood method to show that teams with more kills by one player,

multiple players, and fights initiated show that they have a higher chance of winning. My work

supports this finding because my research shows that initiating fights are one of the most

common forms of tactical warfare in LoL and playing aggressive leads to making plays which

leads to obtaining kills.

The ADC role along with the middle role receive the most assists from their teammates

(Mejorado & Marquez, 2019). This is an assist on a kill which means one player does damage to

another player but is not the one who ends up killing that player. This aligns with my studies

because my research showed ADC champions were the most selected champion archetype

because of their lethality. This translates to them racking up assists in battle because they can do

a lot of damage to champions fast.

Conclusion

By looking at strategy development in LoL through the 4 levels of war, my research has

shown that LoL is an offensive-oriented game. The strategies observed suited a more aggressive

style of play while also showing how important it is to make the right decisions in-game to

mitigate your opponents advantage margin. All 4 levels of war were witnessed but none more

than the tactical level of war. Summoner spells and other tactical strategies were used as

playmaking potential. Professional players were more prone to using the same small pool of

champions versus lower rank players who were more diverse in their champion selection.

Champions of the marksman archetype were selected more times than any other character

archetype.

References

Tang, W. (2020, June). Understanding esports from the Perspective of Team Dynamics. The

Sport Journal.

https://thesportjournal.org/article/understanding-esports-from-the-perspective-of-team-dy

namics/

Wallner, G. (2019). Data Analytics Applications in Gaming and Entertainment (1st ed.)

[E-book]. Auerbach Publications.

https://www.google.com/books/edition/Data_Analytics_Applications_in_Gaming_an/_xa

mDwAAQBAJ?hl=en&gbpv=1&pg=PA49&printsec=frontcover

Harteveld, C., Kleinman, E., Rizzo, P., Schouten, D., Nguyen, T. H., Liberty, S., ... & El-Nasr, M.

S. (2019, August). Teamwork and adaptation in games (TAG) a survey to gauge

teamwork. In Proceedings of the 14th International Conference on the Foundations of

Digital Games (pp. 1-12).

Lee, C. S., & Ramler, I. (2017, August). Identifying and evaluating successful non-meta

strategies in league of legends. In Proceedings of the 12th International Conference on

the Foundations of Digital Games (pp. 1-6).

Stuart. (2021, March 17). How to Become a Pro League of Legends Player. Cyber Athletiks.

https://cyberathletiks.com/how-to-become-a-pro-league-of-legends-player/.

Michael G Wagner. 2006. On the Scientific Relevance of eSports.. In International conference on

internet computing. 437–442.

Nagorsky, E., & Wiemeyer, J. (2020). The structure of performance and training in esports.

PLOS ONE, 15(8). https://doi.org/10.1371/journal.pone.0237584

Xia, B., Wang, H., & Zhou, R. (2019). What Contributes to Success in MOBA Games? An

Empirical Study of Defense of the Ancients 2. Games and Culture, 14(5), 498–522.

https://doi.org/10.1177/1555412017710599

D. A. M. Mejorado and J. E. R. Marquez, "Visual analytics of team performance in online

gaming," 2019 IEEE International Conference on Systems, Man and Cybernetics (SMC),

2019, pp. 3025-3030, doi: 10.1109/SMC.2019.8914387.

Yang, P., Harrison, B. E., & Roberts, D. L. (2014, April). Identifying patterns in combat that are predictive of

success in MOBA games. In FDG.