Decision making is noisy when there’s a high degree of variability. When different radiologists analyse the same X- Ray, their diagnosis vary. Likewise in areas of medicine, law and finance, there are divergent opinions when experts are presented with identical data. Worse still, when the same data is presented to the same expert for a second time, he may arrive at a different conclusion depending if it’s before or after lunch!
Yet, in 2016, the Harvard Business Review (or HBR) reported that Nobel laureate Prof Daniel Kahneman observed that chess masters tend to share similar assessments when presented with the same game. Drivers too, when they assess traffic patterns at junctions and roundabouts.
Why then are humans reliable decision makers in some areas and unreliable in others?
In chess and driving, there is a strong cycle of practice in predictable environments. Actions carry immediate and clear feedback. Both chess players and drivers can learn.
In contrast, the professional world such as law, medicine and finance lacks a clear link between action and outcome. Professionals learn on the job, usually from hearing seniors explain and criticize. Furthermore, financial judgement involves long feedback loops. A poor investment decision unravels months later. Sometimes, luck or chance intervenes. Indeed, which parts of the outcome can truly be attributed to good decision making?
The same HBR article revealed that these professionals are overconfident in the reliability of their peers’ decisions. Experienced professionals making decision on the same data can diverge by a whopping 40 to 60%!
So how do we reduce noise?
By using a set of simple rules. When we present the same inputs, the decisions coming out must be the same. Algorithmic decision making can be simple, but the key advantage is they reduce noise. The same formula applied to the same data will consistently generate the same output. 2+2 = 4 not 7 , regardless of whether it was presented before or after lunch.
By focusing on high quality businesses, we are already aiming for the center of the bull’s eye. By using simple algorithms to manage position sizes we want to greatly reduce the range of noisy outcomes.