When I left the office last night the probability of Trump winning the US presidential election was 15%. We had some very keen people in the office tracking the odds as shown below:

In the night the probability of a Trump win dipped to 9% – we all now know what happened next.

I am sure you will all be reading blogs on the political impact of the US election but for me, the lesson from the election is a continuation of a theme. Assumptions and probabilities are not hugely useful in a single event; we have seen this numerous times.

So we have seen that even a 9% probability can still not give us confidence in a result. You have a similar principle applied in the calling of states during the election. This, typically, is only done once the probability of an outcome is 99%. What this means for me is that any probability between 5% and 95% is not as useful a guide to potential outcomes as it looks.

Now, the nature of probabilities is that if the election was re-run hundreds or thousands of times then Clinton would have won more. But in a single event they are pretty much useless.

We have blogged on this before.

Perhaps the issue for pension schemes is that even if you have a strategy that has a 70% chance of success, you must be prepared for the eventuality that you will not succeed in practice.

Irrespective of your political views, what we have learned from elections this year is that we need to sense-check any behaviours that are based on probabilities.

For the large numbers of people worried about market impacts of the election result, there is some hope (ironically provided by this result). All the assumptions about market movements from here are explicitly or implicitly probability-based. One thing to learn from this election is that when we rely on assumptions and probabilities there is, by definition, a chance we’ll be wrong.