AI FORECASTING REPORT: The Reality, Part 2

374

By Terry Kades, BlueWave Forecasting

If AI is not the magic wand, what is?

How many technology savvy investors believe that they can outsource investment decisions to AI, some level of automated trading where the rule in embedded in a learning algorithm?  At this point in its genesis, AI is in relative infancy and human experience and judgment still trumps AI systems. However AI is still very useful in sifting through large amounts of data and complexity, and guiding humans to investment opportunities and better informed decisions.

Lets move on to industry and business scenarios. A scenario is a “wrapper” that describes a group of drivers. At the industry level, these drivers are in put into categories such as macroeconomic, regulatory, political, legal, duties and taxes, transportation, labor etc. Examples include GDP, inflation, interest rates, labor rates, fuel prices, customs duties for goods related to an industry and so on. It’s best to be pithy and include only those drivers that really have a material impact on an industry. Weighting of drivers within a scenario is necessary and requires running several statistical techniques that are rather complex. We suggest between 6 and 9 drivers so that each driver counts with sufficient meaningful impact in a statistical sense.

Forecasting the drivers is another step in the scenario creation process. Each driver must be forecast according to the scenario of the driver itself. For example, if GDP, inflation and interest rates were drivers, one would need to forecast various scenarios of GDP, inflation and interest rates and slot them into their respective scenario containers. One version might be optimistic GDP and low inflation, while another might be the opposite, and many permutations and combinations can be created. The label of a scenario is what describes the version of the data that the drivers represent.

A business scenario is formed from the industry scenarios in which the business operates. Doing it this way will provide the best opportunity to give context to scenarios for a business. Scenarios are used to drive forecasts and so its important to base a business scenario in the context of its industry. When a business approaches a bank or investor and asked to produce their forecasts, at least they will be able to say “these are our forecasts and they are based on the scenarios for our industry”. What makes a business scenario different from an industry scenario? The difference is the inclusion of a “bucket” of drivers that pertain to the business such as it’s products, markets, partnerships, channels, investments etc. Again these are also weighted according to their importance and will be included in the drivers of the forecasts make for the business.

It becomes clear that using AI to forecast with any degree of accuracy requires the algorithms to know how the drivers of the forecast are wired. This means the inter-relationship of the drivers must be stated and if anything changes, then changes to the algorithm need to be made, and this requires judgment based on experience. While it is still early days in the evolution of AI, eventually forecasts will be based on the drivers in the context of scenarios in an industry. But this is complex and will take some time and a good deal of testing to get is formed and calibrated. In the meantime we use AI to shortcut and improve the path to make our task as human beings a bit easier.

The next article will discuss scenarios for key customers, those few customers who represent over 65% of the revenues generated. 


Terry Kades is the CEO and co-founder of BlueWave Forecasting, which provides industry and business scenarios and forecasting. An experienced director and senior management consultant he is expert at identifying the drivers and creating scenarios that guide business plans, enhance analysis of investments and provide accurate forecasts. He is convinced that the true power of AI will flourish when it incorporates properly constructed scenarios that fertilize and boost the intelligence aspect of AI.