Why we sometimes overvalue the favourite in eSports

Why we sometimes overvalue the favourite in eSports

By Montse Garcia Sep 24, 2013

Tweet

Whenever people look at potential eSports bets, or simply evaluate a match and predict a winner, more often than not there is a tendency to overvalue the favourite. This is a natural human reaction because of the difficulty we have in recognising randomness, over valuing players’ reputations and misunderstanding basic probability.

Korea vs. the world – Reputation or randomness

In international tournaments, particularly when foreign players play against strong Koreans, people tend to forget that upsets can occur due to the randomness that is inherent to these games, as well as external factors that have nothing to do with a difference in skill – jetlag, having a bad day, illness and social influences could all have an impact.

Once the game starts, a bit of luck early on (think of build-order advantage or a spot-on scan) can snowball heavily and end in a win for the underdog.

Differentiating heavyweights

In Korean vs. Korean matches, although the overall skill level can be higher, the reality is that the talent gap amongst them is often smaller than eSports bettors realise.

For example, many players’ carry a reputation that is impossible to disregard when trying to make a fair judgement of a match – think of Flash, MVP, etc. This can easily bias bettors in their favour, although most of them are actually tied to similar practice regimes, and have access to equally talented partners and coaches to help them prepare for that match, while a perfectly executed cheese might be all it takes for a player to claim victory.

Many players’ carry a reputation that is impossible to disregard when trying to make a fair judgement of a match.

Overweighting reputation – halo effect

Because of the history and nostalgia that comes with certain players, it is also a common mistake to judge them too heavily by their past performance. In general, it takes longer than it should for people to recognise bad streaks or a considerable downswing in a player’s performance. Through this adjustment period, many inaccurate predictions regarding the player’s ability tend to be made.

Sure some players have had remarkable success in the past, but at some point new names emerge to take the throne. Starcraft 2 for example, is well known for being a “history of eras”. This bias is known as the ‘Halo Effect’, read more about it here.

Fooled by randomness

So what about evenly matched players? Let’s make an experiment with Starcraft 2. Imagine a tournament with 16 players, all of which are equally skilled, for the sake of the experiment we can clone Flash 15 times and have him play only the same match-up, TvT; each game is independent with skill level the same. We start out with a Code S style group stage. Obviously, someone is going to win, but we shouldn’t infer superior skill because of success, as the results are essentially determined by the random factors mentioned earlier.

Nevertheless, what we would likely see is high praise from by the casters and the community for the people who advance, especially the ones going 2-0: “such a smart choice of build order”, “he punished the greedy play of his opponent with a well executed all-in”, “he read his opponent like a book” and so on.

On the other hand, if you were to read what people have to say about those who dropped out of the tournament, you would likely find classics such as “he should have seen that coming”, “he played way too greedy”, or “he got outsmarted so badly” and similar comments.

Next we have an eight player single elimination bracket. Quarter-final and semi-final will be best-of-five, and finals best-of-seven.

Assuming every match is 50/50, it’s quite likely that there is one 3-0, one 3-1 and two 3-2’s. Even while quick maths show you that the chance of a match between two equal players ending 3-0 for either player is actually 25% [(0.53)+ (0.53)], you can imagine all the hype that the casters and viewers will create around the person who wins 3-0. Surely he must be the favourite to win the tournament, right?

We’re now at the semi-final. The player who won 3-0 in the QF plays against one of the 3-2s. Suddenly, the hyped player loses 0-3. He must’ve choked, right? The other semi-final is won by a score of 3-2.

You can probably guess who is seen as a big favourite in the final. People will probably bet until the line is at 1.50. All the while, it’s really still all 50:50 (odds of 2.00) but perceptions have been skewed by two 25% situations occurring.

And then someone wins. A caster might claim, “he is clearly the best player in the world right now”. And why? There are two processes going on here – survivorship bias describes how we mistake a random process for something which has correlation, on top of which we are generating a narrative to explain the correlation.

When we watch games, our brain is often overwhelmed by what we see and hear. It’s almost impossible to see the underlying probabilities. Instead, our minds try very hard to create stories around what we just saw. This makes things easier to comprehend, but more detached from an objective truth of probability.

Humans simply aren’t built to adequately deal with randomness. We like stories, we want things to be clear, because that’s easier to grasp… and much more entertaining. No one would watch an eSports tournament where people flip coins to decide who wins a match, and though our tournament example is hypothetical the principle does apply in reality, just not quite so literally.

So remember when you think you have witnessed an upset that perhaps it was nothing more than a random factor at play. As an eSports enthusiast who is perhaps new to betting, this idea is just as important to understand as an in-depth knowledge of the games, players and strategies.

If you have feedback, comments or questions regarding this article, please email the author or send us a tweet on Twitter.