In praise of charts



Andrew Berkley


10 Sep 2014




A quick chart in Excel is a wonderful thing: highlight some data (ctrl + shift + right arrow) and produce a chart (F11). Sit back.  Enjoy the feeling of fingers off keyboard.

For a brief period while considering a chart, the modeller enters the realm of the chief executive. That is what chief executives do, isn’t it? They look at charts. Then they ask awkward questions.

A good modeller will know that they should look out for three things in a chart:

– a step change: when a step change appears in a chart, a modeller’s antenna should bristle. Step changes are not that easy to model unless (i) they are modelled badly or (ii) they are modelled using complex functions or (iii) the step change is explicit in the input assumptions. So a good modeller should be able to explain which is causing the jump;

– a kink: a change in direction of a gradient suggests something other than a constant at work; and

– a spike: a spike suggests an anomaly or an outlier. Chief executives are particularly good at asking awkward questions about spikes.

A good modeller will also know that there are only two explanations that lie behind step changes, kinks and spikes in charts:

– the first is a genuine commercial reason; and

– the second is a modelling error.

Being forced to admit that a modelling error is the cause of a step change, kink or spike in a chart is a pretty humbling experience. But it is not as bad as being presented with a chart and having no explanation at all.

A modeller that has no explanation for their charts is a modeller that has little aspiration to become a chief executive. They are content to crunch numbers; they have little care for how those numbers might be interpreted by others.

A modeller that charts as they model is a chief executive in waiting.

Andrew Berkley
With a background in business education and financial advisory work, Andrew leads financial modelling training at F1F9. He has been with F1F9 since 2013.