How reliable are advice-based algorithms?

June 23, 2020

Article by Gabrielle de Neve, Head of ASEAN Markets.

As I’m driving to another city one morning – back when working-from-home wasn’t the norm yet, and my morning commute was littered with congestion – I was thankful for my navigation application. A trip that would have taken me 30 minutes outside of peak hours, would only take me 45 minutes and a few detours with the application instead of 60 minutes (or more) without the application or any detours from the route I know by heart.

Almost everyone has been in a similar situation, and the parallel between navigation application algorithms and investment advice algorithms is an easy one – maybe too easy – but it got me thinking about how I was managing my investments. Here I am, relying every day and for all my trips on an algorithm to avoid being slowed down; shouldn’t I be doing the same thing for my investments? Would I ring up a single person to ask advice on the best route to take when I’m stuck in traffic? Probably not. How about if I knew this person was using a navigation application when he advised me on the route to take? Well, that changes things.

It is very easy to see why advice-based algorithms are increasingly used, whether it is for self-serviced investment advice (robo-advisors) or for advisor-led investment advice.  In the same way a navigation application can give you advice on the optimal route to get to your destination by collecting current and historical data such as the time and day of travel and the expected weather conditions, advice-based algorithms can provide investors with valuable investment advice by collecting data on the investor’s investment preferences and beliefs, and on current and historical market data.

When unforeseen events create turbulent times, whether on the road or in financial markets, algorithms have the capacity to automatically recompute the best route, taking you off the most congested roads (usually the ones which are most known by the drivers) and onto roads you might have never taken before. If you didn’t have your app guiding you through the process, would you have gotten off the highly congested highway to go on a road you didn’t know and in a region you might not be very familiar with? Although not impossible, you might not have taken the decision as easily as with your app and you might take a few wrong turns in the process.

Will investment advice algorithms predict the future and guarantee you will never lose anything on financial markets? It’s unlikely today, there are always unforeseen elements that can sway a prediction off course. Just like navigation apps cannot predict with 100% certainty in how much time you will reach your destination because they cannot (yet) predict obstacles likes unplanned protests or accidents involving a highly inflammable chemical vehicle. However, these apps will immediately incorporate information of previously unforeseen obstacles (the road is blocked, congestion is piling up very fast) and recompute the new optimal route. Investment algorithms do the same thing: unforeseen obstacles are immediately reflected in the data used by the algorithm which will recompute the new optimal portfolio given the new information that is available.

Until we figure out how to predict financial markets with certainty, the most important thing an algorithm can do is react: detect market movements, anticipate the moves that need to be made, and execute them as best possible.

If both my navigation application and my investment advisory application can guarantee that they can give me the best advice and reroute me on time in case of unexpected events it’s only natural that advice-based algorithms are on the rise, and investors are rapidly turning to them in turbulent times.