Charts and Dashboards: Tornado Charts - Part 1

Welcome back to our Charts and Dashboards blog series.  This week, we consider tornado charts and their uses.

A Tornado chart, put basically, is a Bar chart that extends out on both sides of the vertical (y) axis. They are useful when conducting sensitivity analysis (NOT scenario analysis, as we only want to flex one variable at a time when constructing this chart).

This chart displays how much impact varying an input has on a measured output – which may be depicted in either absolute or percentage terms.

A Tornado chart shows how sensitising each variable causes the output to vary from the “base” case, i.e. with no sensitivities applied.  However, it does rely on variability being monotonic – that is, the extremes will occur at the maximum and minimum points sensitised.

Each variable is ranked by the magnitude of the difference it caused, and this ranking of the variability is what creates the ‘tornado’ layout. 

The assumptions can be varied either all by the same percentage, or by agreed-upon parameters.

Varying all inputs similarly (e.g. ± 10%) is a “deterministic” approach and will highlight the key drivers of the model in these circumstances.  This variation should be symmetrical so that income and expenditure may be compared consistently.

However, this may not be realistic as different variables are unlikely to be equally likely to change by the same percentage (e.g. foreign exchange rates may vary by ±30%, whereas fixed costs may only range by -5% to +8%). When these probabilities are considered in a tornado chart, this is a “probabilistic” approach.

Data Tables can assist greatly in the construction of a Tornado Chart, so please consider reading our thought article on them - here.

We will cover the specifics of building a Tornado Chart next week, in part 2.

That’s it for this week.  Come back next week for more Charts and Dashboards tips.