In 2020 I decided to create a CSGO betting model. It was the first time I was involved in any esports betting. But as a person, who takes the sports betting industry seriously, I wanted to take the advantage of this growing market.
I found that there is a growing CSGO (Counter-Strike: Global Offensive) gambling community and more and more esports bettors are interested in betting on CSGO and other esports markets.
How I started betting on esports
Because of the COVID situation at the start of 2020, when all sports and leagues were shut down, I decided to start researching CSGO betting opportunities and I decided to create my own CSGO spreadsheet, where I will get my probabilities of possible outcomes of CSGO games. With this, I can find value bet situations and place wagers on CSGO events.
Steps I used when creating CSGO betting model
The steps to build such a predictive model is similar to other sports betting models.
1. Learn and research esports leagues
First I researched esports as a whole and then what kind of bet types and leagues are available.
Found that the most popular leagues are:
- League of Legends
- Counter-Strike: Global Offensive (CSGO)
These three leagues are most popular and I found that we have basically teams, players who are paid a pretty decent amount of money (Johan “N0tail” Sundstein – $6.9 million (Dota 2)). The teams have players, lineups, there is a lot of money involved, we have fans, we have people who watch this, their players are stars, there are even some trades or players changes,… a lot of similarities with our traditional sports and leagues.
2. Look for information and the statistics
Before I build a betting model, I must research what types of statistics and past data are available on the internet.
I explored available resources and the stats from all three leagues. CSGO statistics was the best fit for my needs. I found that the best-organized player’s statistics and information about CSGO games (counter strike) were on the site HTLV. Players and CSGO teams statistics are very well organized and useful for building my first CSGO betting model.
3. Building a predictive model
The next step was to build a predictive Counter-Strike betting model, that can project winning probabilities, so I can compare with bookmakers numbers.
CSGO Predictive Spreadsheet
Organizing and analysing available CSGO statistics in a spreadsheet was my first step. Despite we have some good information about teams and players, it is still not perfect. But taking into account that esports is a new betting market, that was good enough compared to other leagues.
So, I decided to create all teams with lineups and players stats and then use simple logistic regression.
And the rest is turning these information into probabilities, which gave me directly my estimated odds.
If we look at the CSGO predictive spreadsheet above you can see:
- My Win % – my estimated win%
- My odds – win% turned into the odds (1/win%)
- Bookie odds – I test all my models against sharp bookmakers (Pinnacle)
- Qualified or better – this is the last price (odds) when my bet suggestion is still qualified
- Value/AdjkValue – I set the criterium at 1.50 (if I get adjk 1.50 or more – then this is qualified bet. If not, then it is not qualified)
4. Daily betting analysis and tracking results
The next thing was something I call a dirty job. Showing real-time results against real betting markets. It’s a daily analysis of CSGO games and tracking results.
It’s nice to talk about betting theory, about betting models, … but what I did here is to test (and bet) real-time and publicly show this. Wake up, check the odds, do the job. And then next day the same.
I decided to test my CSGO betting model against opening odds, against real-time odds and closing odds to see the results.
What I have learned about betting CSGO?
Esports is a new big thing in the betting World, because all indicators are showing that it is growing and it will not stop soon. Betting on CSGO and building predictive model is similar to other sports.
We have teams, we have games, they play against each other and good CSGO model helps you to identify value bets. There are different ways to build a betting model for CSGO, but my variation is as a predictive spreadsheet in Google Sheets.
The goal of my CSGO model is the same than any other betting model I use. It’s about projecting the odds and probabilities, then comparing with bookmakers odds. If the model finds a value, it is a potentially a good bet.
Things I learned about CSGO betting and using my predictive model:
- The competition is almost every day (at least so far) and the good thing is that we don’t need stadiums, the weather also don’t affect the game, which makes this much easier in these uncertain times of Covid-19.
- All trusted and respected bookmakers offer eSports now. They also know that this is big thing and huge market.
- The market is still soft I believe. There were some nice line movement and I didn’t see yet a betting model for eSports. I am not saying that it doesn’t exists, but at least I didn’t saw it in betting communities. It is all based on eSports lovers and writing analysis, which is basically the same, what most sports fans do in other leagues and sports. Because of that, there is no such competition like in NBA betting or English Premier League betting for example, where we have a lot of sharp bettors and a lot of betting models/algorithms. This is definitely a chance for many.
- I also learned, that csgo games are played in different maps and the information about maps come out later. Most of the times teams play on 2 wins (winner out of 3). In grand final they can play up to 5 games (winner of 3) and in some cases they play only one map. Some teams are better with some maps and the other teams are better with other maps, which makes sense. So, I stay away from games with only one map, because it can be a gamble. I lost bet or two, when I was testing the model, because of this, but later I stayed away from one map games.
- In-game betting can be interesting if you focus on maps model. The maps information come out at the start of the game, so after the first map, we know which map they will play next. So, the model can be built also based on maps and those who love live betting can create a model for this second map betting. I am not big fan of in-game betting, but my model is part of our membership, so anyone can download it and with some additional work add this feature with maps.
- The margins from bookmakers are still little bit bigger than on other sports (4.79% on my example), but this is completely normal at this stage now. It is a new sport for bettors and bookmakers take little bit more commission to secure their business. But on the other side (I believe – maybe I am wrong) there is not many “sharp esports” bettors, so this margin can be beaten probably easier than 2.5% vig on sharp markets, like NBA.
- I also learned that teams will change players from time to time. This is why it is also important to follow these changes and trades, so you can incorporate this information into the model.
- Almost all games are live on TWITCH, where you can watch two teams shooting at each others, with commentators and many viewers. I watched 2-3 games just to get the feeling about it, but in general I don’t care about excitement and watching games. For me most important is to find the value. This is the only focus I have. But if you like watching these games, they are live and can be exciting too.
Esports and CSGO is not going anywhere. CSGO betting and different variations of Esports gambling will be there for many years. More and more bettors are looking how to build esports predictive models, which can be a real advantage in this market.