Machine Learning and eSports

Posted on July 19, 2019 by jamesdmccaffrey

I know what eSports are but I don’t fully understand all of the nuances. I’ve always been a big fan of games, but only classical games like chess and poker. I’ve only ever played video games a handful of times – the original Doom many years ago, StarCraft when it was first released, and a couple of others. I get bored with video games quite quickly, probably because I’m not very good at them.

I have a good friend, KL, who lives and breathes video games. He’s a very senior guy at Microsoft’s Xbox. I was chatting with him recently and my first question was related to the similarities between watching games like baseball and football, and watching experts play video games. KL explained to me that even though there are thousands of video games, only a few are used for eSports and that they’re popular games played by millions of people. OK, that makes sense to me.


I will be presenting on a panel about machine learning at the upcoming CEC conference, Sept. 4-5, 2019. I plan to pick the brains of the academics there from the University of Nevada.

KL also explained to me that most of the video games used in eSports require roughly equal amounts of strategy and physical skill. This, in principle, makes the idea of watching eSports more interesting and plausible.

Perhaps the idea is similar to most other sports. People who play a sport or game (or used to play when they were young) like to watch professionals and experts play those games. Men that grew up playing ice hockey (or whatever) like to watch professional ice hockey, but people who didn’t play ice hockey (or whatever) as young men likely have limited interest.

This could explain why eSports could become wildly popular. It could also explain why the abomination that’s women’s pro basketball has no chance of becoming popular among normal women. But I still have a lot of questions in my mind. Professional sports is monetized by TV advertising, which means the audience has to have money and be inclined to buy what is being advertised (typically things like beer, automobiles, etc.) It’s not clear to me what an eSports audience can support in terms of monetization.

Well, what’s the connection between eSports and machine learning? I’m not 100% sure. Machine learning is all about prediction, and sports prediction is useful if there is wagering involved. I do know there is intense activity in state governments surrounding efforts to create all kinds of new wagering scenarios for professional sports. If eSports gains traction, then it’s possible that machine learning can be applied in much the same way that it can be applied to traditional sports wagering.

In the end, everything is about economics and money. But not really. In the end, money is all about what good it can do for others.


Women’s professional basketball attracts dozens of fans. Professional tuna tossing. Semi-pro toe wresting. Unicycles plus basketball equals strange.

This entry was posted in Conferences, Miscellaneous. Bookmark the permalink.

3 Responses to Machine Learning and eSports

  1. Thorsten Kleppe's avatar Thorsten Kleppe says:
    July 20, 2019 at 5:00 am

    To fully understand the eSport on each game you need to play it.
    On of the most impresive moments I know is the game this year between ninja in pyjamas (nip) vs fnatic on the IEM in sydney, both swedish teams exist for decades. Couter strike exist 20 years now and its the highest competition class of strategie shooter.
    This minute shows all the intense and emotions of the last round, for nip means this a draw for overtime or a loss. This match shows a lot of this fascination.
    https://www.youtube.com/watch?v=s5DowrFIJ0w

    I was a pro player 2003/2004 in tom clancy’s rainbow six raven shield, it was a amazing time with the best team I could ever had, we named us “good game”. We played 2004 in Berlin on the Cubix cinema, that was our biggest event.
    But Ubi Soft, the publisher, was a really silly company, they let this potential eSport game die and today the big player in eSports are other companys, a sadly end.
    Because they build a grafic engine with a demo function, and take this function out, but that’ elementery function to record the games for the viewers, so it was not really possible to publish the gameplays.
    Today in csgo they fill the biggest halls. This weekend in chicago with the Intel extreme masters.

    The monetized aspect is easy to tunderstand if you know how much money the gamers giving for there sport. A nice state of the art pc today is costly, with a cpu around 500$, a gaming TFT with 240hz = 500$, a gpu = 1000$ and you need a lot more for your gaming experience. We play for over 20 years, last time I was watching the stats, there was over 1 million activ player in the month in csgo only.

    My mouse is a zowie fk2 for 60$, keyboard spectrum gram for 120$ and a mousepad steelseries NP+ =40$, headset siberia v2 = 70$, and in 3 or 4 years maybe I will update some devices again.

    On steam, the biggest platform for gamers, you can find stats about the distribution of the hard and software configurations of the users. There is a big market, without gaming the desktop pc would not be the desktop anymore.

    Ok, the most impressive ML moment was in a strategie game named dota 2, OpenAI Five from DeepMind was playing against the best player in this game.
    https://www.youtube.com/watch?v=tfb6aEUMC04
    The bot was able to create new strategies which never used by humans. The bot was searching for the enemy units on the map and sends decoys to them and wait till the enemy was attacking, and then he was escaping with low energy but still alive, so he did disturbs the ordering from the human and could win against them, wow.

    A nice prediction system in csgo could be a ML method like a LSTM, the demo function saves all the data about the games, so a nice system can create a fingerprint about the players, the teams and the matches.

    Many pro teams are sponsored by betting company’s.

  2. Pingback: #machinelearning and #esports – Global Life

Comments are closed.