“To derive the most useful information from multiple sources of information, you should always try to make these sources independent of each other. This rule is part of good police procedures. When there are multiple witnesses to an event, they are not allowed to discuss it before giving testimony. The goal is not only to present collusion by hostile witnesses, it is also to prevent unbiased witnesses from influencing each other.” – Daniel Kahneman – from Farsighted Steve Johnson
This sounds very simple, but it is very important to eliminate groupthink for any investment decision. A poor situation is to have an investment committee meeting and have everyone around the table start to give their views without taking a position, and making sure the chairman or lead manager gives his strong opinions first. Take a guess on what will happen? By some miracle, the committee is likely to have a similar view to the chairman.
If there is an effort to gain consensus before a vote, there will not be much independent thinking. Groupthink reduces independent sources of information and opinion. Conflict with investment decisions is good. Good decision-making does not allow for pre-meetings.
Perhaps the best approach for eliminating groupthink is to have every committee member write down their views and potential action before anyone talks. Unfortunately, this is unlikely to happen in real live. As social beings, there is a desire for consensus.
However, there is an alternative through developing independent quantitative models. There is no socialization. There is no waffling or influence from other groups. Separate models can be done through having different models focus on different sources of information. A simple case would be to have a price-based trend model and a fundamental model that uses macro data. These are two different and independent views. Another model could focus on cross-asset data. Still another model can take a very short-term view versus others that focuses on long-term information. These models can be compared and rated to derive an overall forecast.
These ensemble model approaches allow for independence and a strong basis for comparison. The decision framework is straightforward. If there is no consensus across models, then any action should be tempered. Strong consensus will lead to greater action. The investment procedure of creating independence can be preserved.