The history and current state of Esports
Summary
In this document, we will be analyzing Esports data on players, tournaments, game sentiment, and game meta.
The document is broken down into 4 sections:
Game Sentiment
CSGO Game Bias
Esports prize pools and salaries
Player data
Game Sentiment
We will be conducting an analysis of game sentiment for Dota, CSGO, and Apex Legends. We picked these games because they were the three main Esports games available on Steam. The data is obtained by scraping Twitter and steam review data. Firstly, we will look at the most common words used in steam reviews and twitter for each game. We used 1000 Steam reviews from each game and 2000 tweets for each game to conduct this analysis.
This chart shows the most commonly used words in each game on Steam. For Apex Legends, there seems to be a large number of people that mentioned “bugs” which indicates that bugs might be an issue for the game. “Friends” was also a common word in CSGO steam reviews, which means that people tend to play CSGO with friends. In both CSGO and Dota 2, the words “rank” and “competitive” are mentioned relatively frequently, this means that players of these games tend to play the game more competitively compared to Apex Legends.
On twitter, there seems to be a lot of discussion regarding skins on CSGO with words such as “skins” and “csgogiveaway”. Skins are a prevalent part of the CSGO community has been a huge source of revenue for Valve and skin resellers. Many companies also give away CSGO skins in exchange for twitter followers, retweets, or traffic on their websites. When scraping the data, we found that a majority of tweets about CSGO were regarding skin giveaways. Another common word in CSGO is “betting”. This is because there are many CSGO betting companies that promote their websites using Twitter. Usually, those advertisements are done through Twitter bots.
In this graph, a notable common word would be “toxic” as Dota 2 and CS:GO both tend to have much more mentions of player toxicity within the games. This can be due to the fact that teams on CSGO and Dota comprises of 5 people compared to that of Apex where you can play solo or up to three people. This means that people are less likely to play with toxic teammates on Apex. Another interesting point is that there seems to be a high amount of people that talk about “cheaters” in CSGO. This likely means that cheaters are significantly more prevalent in CSGO compared to Apex and Dota.
An interesting thing to note from this graph is that the word “esports” is mentioned quite frequently for CSGO and Dota 2. This is an accurate reflection of the significance of esports on each game. In section 3 (prize pool) we will dig deeper into the specific comparisons of different esports games in terms of the prize pool and average salary.
The first graph shows the sentiment of players of each game in Steam reviews while the second shows the tweet sentiment of each game. The higher the sentiment value, the more positive the overall sentiment is. Both graphs showed varying results which can be explained by the fact that Steam review sentiment is based on reviews of the game while Twitter sentiment can be also based on other factors such as community, and esports. We concluded that CSGO has a larger esports fanbase as well as a overall more positive community while also having a worst-rated game which could be related to cheaters, gambling, and skin scams. Additionally, twitter bots are much more common in tweets regarding CSGO due to gambling websites and skin giveaways which might increase the overall sentiment. These factors are the reason for the disparity between both graphs.
CSGO Game Bias
For our in game analysis we looked at the difference between the offense Terrorists Team statistics and the defense Counter Terrorists related to their overall win percentage. When looking at the CT sides graphs one of the more obvious things that stick out is that the overall win rate is higher than the T side. Another thing that sticks out are how close the points are to the regression line in comparison to the T graphs. When running the correlation tests the only two factors that were not statistically significant were the flash assists per player, and the grenade damage per player per round. When running, correlations on the T and CT side flash assist are slightly more correlated to wins with a p value of 0.263 compared to the CT at 0.457. However the grenade damage is slightly more significant for the CT likely because players can use things like molotovs more effectively on defense. For all of the other factors they had a p value of 0.05 or less, showing the CT bias for the game.
When looking at the T side graphs most of them are all over the place except for when it relates to round wins. The T side graphs have a much less dramatic slope compared to the CT graphs. When running a correlation test, the most significant factor related to wins was multi kills with a p value of 0.001 proving the closeness of points on the linear regression line. Teams on offense getting more than one kill at once, is big because it likely leaves the defenders dead at the sight. Now the defense has to rotate to the attackers who are now defending with a numbers advantage. The other 2 plots relate to the teams winning a round after losing a player to first blood, or winning the first blood. These are the most obvious statistics that will lead to wins because the more rounds won the closer a team gets to winning. Overall the T side Has a much harder time getting wins because the only real thing they can do is win rounds and try and get a multikill to leave the other team at a disadvantage.
Overview
When looking at this data overall, if a player is looking to improve win percentage they can do a few things. When it comes to the Terrorist side the best thing a team can do is win rounds where they are at a disadvantage 4v5 after the opening kill. Winning rounds is also significant for the CT side however it is much more important on offense. Winning rounds whether at an advantage or disadvantage is the most important thing when it comes to winning overall. However, just saying win more rounds is not very helpful. When it comes to T side Multikills and opening kills are the best thing teams can do to help them get wins. It may be helpful to use things like flashes but will not make a significant difference. When it comes to the Counter Terrorist side most of the metrics are closely correlated to wins. However trading deaths with the attackers is the most important for defenders because it wont leave them at a numbers disadvantage when taking back the point. Also wining the opening duel is more statistically significant for the defense because it then puts attackers at a disadvantage.
Esports Prize pools & Salaries
Total Amount of Prize Pool Between Different Games
The League of Legends Data continues to be a part of this chart.
For Dota 2, we decided to use only the International Tournament, since their prize pool money itself already beats all other totals of any other game. For Valorant, the game only started to gain a prize pool on their release date of 2020.
The primary trend noticed in this chart is the decline of prize pool money throughout all listed games as of going into 2022. Although it could be speculated that the cause of this is due to the fluctuating economy and inflation after the pandemic. It is especially tragic of CSGO to drop a huge amount in 2022, since it has been consistent for many years.
In 2020 when the pandemic hit, Dota 2 hit 0 because of covid restrictions, but they quickly rose to their highest point of any other game in 2021 sitting around $40 million in prize pool.
Average Salaries of Top Competitive Gamers
Culturally speaking, League of legends has been known to be the most dominating game known to be played on the computer from the years 2010 to 2020. Even with the arrival of other shooters like Overwatch or Destiny, they have been generally short lived compared to the long reigning power of LOL. Counter Strike has existed 8 years before league and has had a consistent fanbase that follows it for it being a first-person shooter.
Counter Strike, along with other first person shooters at the time (Crossfire) had been the kings of the first-person scene. League and Dota have been known to be staple MMO RPGS (Mass Multiplayer Online Role Playing Game) while entering the 2010, with strong correlation with Real Time Strategy Games like Starcraft.
The player count of League is estimated to be around 70 million while Dota sits around 43 million. Therefore it is surprising to see that Dota 2 players are known to have the highest earnings in any competitive game. It is more surprising that the average League of Legends pro gamer earns less than a Chess24 player, as they barely sit comfortably in the top 20.
While most of us were PC gamers while doing this research, seeing how much impact console players have on the competitive gaming scene is also interesting. We were generally surprised to see how there was a substantial amount of Call of Duty games being listed in the Top 20 most paid pro gamers games.
Player Data
For this section, we only analyzed player data for both CSGO and DOTA as they had player data ranging all the way back to 2012. The data is obtained from steamcharts.com. As seen in the graphs, CS:GO surpassed Dota 2 in player count close to the start of 2019. This can be attributed to CS: GO becoming free-to-play in December 2018. Previously, the game would be priced around $14.99, but now everyone is able to download it for free. After 2020, CSGO saw a decrease in player base and has been slowly decreasing since. At the current rate, Dota 2 might catch up to CSGO in terms of the number of players.
Conclusion
In this project, we presented the history and current state of Esports by analyzing the overall current sentiment of different Esport titles , prize pools of different esports, player salaries, and concurrent player data. We also found what makes a good CSGO professional by finding the metrics most highly correlated to increasing win rate.