When it comes to understanding the fundamental principles behind odds and probabilities, there is perhaps no better person to turn to than Andrei Marius Popescu. This esteemed researcher has made a name for himself throughout many scholarly communities, and his influence on the sports betting sector is particularly impressive. What does he have to say about the role that cognitive biases may play in relation to predictive modelling? While this is a decidedly niche subject, the answers are nonetheless quite enlightening.
A Quick Definition
It is wise to begin by defining the term “cognitive bias” in relation to the outcome of a specific event. Thankfully, a considerable amount of research that has been undertaken by Andrei Marius Popescu focuses on this subject.
“We can define cognitive biases as a type of mental shortcut,” he begins. “Our brains can sometimes process information too quickly. This may lead to incorrect assumptions; especially when the results go against objective facts.”
One example of this way of thinking involves placing a greater degree of importance on an unlikely event (such as a plane crash) simply because a similar story was recently featured in the news. In other words, our minds can sometimes distort the facts based on preconceived notions, or personal beliefs (this is why the term “bias” is used).
Now that we have an understanding of this basic principle, how can it be applied to sports betting odds? Andrei Marius Popescu was kind enough to provide us with three relatively common scenarios.
The Gambler’s Fallacy
“This first situation is essentially a false prediction of the future,” he begins. “We can sometimes believe that if something happens frequently in the present, it is less likely to occur in the future.”
Simply stated, individuals often have the tendency of assuming that things will “balance out”; even if there is little evidence to support this belief. This can cloud our judgement when predicting the outcomes of specific events (such as a football competition).
Overconfidence
Another human trait that can often lead to confirmation bias involves the rather simple notion of overconfidence. For example, we can sometimes be led to feel that we have more control over the outcome of an event than is realistic. The problem here is that this can also cause us to downplay the role of chance.
“One of my previous studies involved measuring the influence that overconfidence had on betting strategies,” Andrei Marius Popescu continues. “I found that those who had more experience were often led to believe that skill alone could exert an influence on the outcome of a match. While this was often on a subconscious level, it still impacted the type of wager that was placed.”
The Notion of “Clustering”
A further example of confirmation bias comes in the form of a habit referred to as “clustering”. In this case, we can sometimes group otherwise random events into different categories (or clusters). This leads us to believe that there is some type of underlying pattern; even if one does not inherently exist.
“We need to remember that gambling enthusiasts are generally unaware of this habit. This makes it much more difficult to detect. Others may even go as far as to deny its existence entirely,” Andrei Marius Popescu states. “It can then be much more difficult to correct.”
A Measure of Reality
These are only three examples of how cognitive biases can often distort our perception of randomisation. So, are there any ways in which we can combat such belief systems? “I asked this question to my followers on X, and the responses varied. Some feel that bias is simply a part of the human condition. Others insist that the role of automated betting algorithms can do away with such tendencies.”
However, he also states that bias is a factor that needs to be included within any type of predictive analysis. This is especially relevant when we remember that even seemingly minor biases can have a massive impact on the way we perceive an event.
“Ultimately,” Andrei Marius Popescu concludes. “We have not yet come to the point when any type of machine learning algorithm can account for the nuances attributed to cognitive bias. I am not convinced that these systems will ever become advanced enough to account for these types of outliers.”
So, it seems as if (at least for now) cognitive bias will remain embedded within the majority of wagering strategies. Those who are able to appreciate its influence may still be able to bend the odds slightly in their favour. Either way, it will be interesting to see what data emerges in the future regarding this interesting phenomenon.
This is a submitted article
