Many “casual” bettors choose their bets through “gut feeling” about what they think will happen during an event, an approach which does not lend itself to profitable betting. Does this mean more skilled bettors can rule out utilising their intuition entirely?
In his book “Blink: The Power of Thinking Without Thinking” Malcolm Gladwell looks at the concept of intuition. Why do intuitive, unconscious decisions made on the basis of seemingly very little information, so often turn out better than better-informed, more thoughtful choices?
Take, for example, the talent of tennis coach Vic Braden. Braden could tell when a player is about to double-fault before the tennis racket even meets the ball. Vic did not know how he came to this conclusion it was merely a snap judgement that came to him intuitively.
When he watched a match at Indian Wells, Braden correctly called 16 out of 17 double faults before they happened despite 91.1 percent of second service attempts landing in.
Big data is defined as “extremely large data sets that may be analysed computationally to reveal patterns, trends, and associations, especially relating to human behaviour and interactions” which is about to as far from forming immediate intuitive judgements as it is possible to get.
Big data can highlight areas where intuition and perceived wisdom can be inaccurate. Notable examples include baseball scouting (moneyball), the hot-hand fallacy and highlighting how intuitive judgements can lead to confirmation bias.
However, this does not mean the two are so different.
In the book “Everybody lies” Seth Stephens-Davidowitz argues that Vic Braden’s talent for double fault detection was, in its own way, a form of data analysis.
Braden had seen millions of serves and could intuitively spot the signs of a double fault before the error happened. He was able to calculate the chance of a double fault by analysing a player’s swing and comparing it to the countless service attempts he had seen before.
If it could utilise the inputs Braden was able to, a data-driven approach to predicting double faults would work in much the same way. It would essentially take those inputs and compare them to the same data on all other serves in the database before determining its similarity to other double faults.
Look at the below odds for a standard La Liga matchup between Barcelona and Real Madrid:
Real Madrid odds
If you have any experience betting on sports then you will immediately notice something is amiss with the odds. Intuitively someone who knows anything about assessing the probability of soccer matches can see that the implied chance of a Barcelona win is rated far too low by the odds.
Anyone who came to that conclusion would be correct. These odds were actually taken from Pinnacle’s line for the Tunisia vs. Belgium match at the 2018 World Cup. The bettor intuitively knows that valuing Barcelona’s chances of a win against Real Madrid at the same likelihood as Tunisia’s against Belgium is a falsehood without referencing models or delving into the data.
Considering Vic Braden’s talent was an example of intuition in action, perhaps a variation on his ability could be applied to sports betting. After all, if an expert bettor’s brain can function like a supercomputer then it is possible that bettor could have a very accurate grasp of probabilities.
Certainly, there are successful bettors, such as Lewis Deyong, who attribute their success to an intuitive grasp of probability.
If a bettor could reach a similar level of skill then simply betting intuitively could be profitable, but is that a realistic prospect?
The problem with betting in this way is there are so many things that need to be monitored and so many events that need to be bet on to secure long-term profitability. It is not realistic to expect odds to be out of line to the extent of those above.
By looking at the line a bettor should be able to see that there is a possible flaw in the model and could potentially refine the selections to ensure improved accuracy.
Possessing the knowledge to bet with an advantage across a wide range of events would be akin to Vic Braden attempting to call double faults across hundreds or even thousands of matches played simultaneously.
Braden’s intuition may be able to call the double faults in one specific match but he would quickly become overstretched attempting to apply that intuition across multiple events.
An example that demonstrates this issue comes from an anecdote told by soccer analytics expert Ted Knutson about a conversation he had with former US national team coach Bob Bradley.
Knutson explained the use of expected goals to Bradley who pointed out some potential flaws with this data-driven approach to analysing soccer. Bradley argued that by watching a scoring opportunity he could intuitively grasp the chance of a goal being scored better than the data could.
Knutson understood this but correctly pointed out that “Bob’s eyes can’t evaluate every touch in every game across 27 different leagues”. Whilst Bradley’s expert intuition might have been more accurate than the data in isolated circumstances, it could not be utilised across the sheer breadth of games needed to compete with the insight offered by big data.
Another limitation of betting intuitively is the inability to test those predictions. A data-driven approach can be applied to historical fixtures and tested across numerous matches, whereas the sample size of an intuitive bettor may never reach a level where they can confidently claim to be profitable.
Perhaps a more intuitive bettor could find a big enough edge to be consistently profitable with a smaller set of bets but it would certainly be a difficult task.
Whilst a betting strategy built solely around intuitive judgements is highly unlikely to be successful there is certainly a strong argument for intuitive judgements being applied to betting models.
Imagine for some reason a model is suggesting a bettor should wager on Real Madrid in the match outlined above. By looking at the line a bettor should be able to see that there is a possible flaw in the model and could potentially refine the selections to ensure improved accuracy. By doing this the bettor is essentially applying his own intuition to the process.
Equally, the data may flag up that the bettor’s intuitive judgement about the match is in fact wrong. Perhaps Barcelona have been very poor this season or have lost a range of key players. The bettor’s assumption that “Barcelona are a good team” may no longer apply.
“Gut (or intuition) is shorthand for many of the remarkable qualities of human cognition: an ability to rapidly spot patterns, make associations, combine a rich set of personal experiences and innumerable data points to form a judgment.”
Removing the power of intuition from betting strategy entirely would eliminate a huge data source. Equally, to rely on intuition alone would be a very risky approach to betting as it is solely reliant on the accuracy of the bettor’s grasp of probabilities, which may well be less accurate than he assumes.
As with many topics, a combination of the two approaches is perhaps the strongest approach to forming a successful betting strategy.
Combining the intuitive talent of the sports betting equivalent of Vic Braden with the breadth available to a good data led model would be the ideal scenario. They could both benefit from the insight of the other. It would, therefore, seem churlish then for bettors to write off intuition entirely.