To stay up or go to bed, that is the question:
Whether it’s better for my mind to witness
The twists and turns of an outrageous election,
Or to save myself a load of trouble
And head off to bed on time. To lie – to sleep
No more; and by a sleep to say I check
My phone a thousand times fearful of a shock
That’s fresh and startles, still; it’s an ending
That cannot now be wished away. To sleep;
To sleep, perhaps to dream – ah, there’s the rub:
For in that sleep what nightmares may loom large,
When we have shuffled off to bed at last,
Must make us pause – there’s the challenge
That could make calamitous this long night.
I stayed up last time until the New York Times prediction graph first showed Donald Trump above 50%, at about 2.30am:
.
I switched off the TV and went to bed still believing that Trump’s surprisingly strong showing would ultimately be unsuccessful and would go down as a monumental protest vote, a severe warning to the political class – not the political earthquake it turned out to be.
It was amazing to see such a sure thing – I mean, Clinton was given an 80% chance of winning! – turn out to be “wrong”. Forecasting seemed to have a pretty bad 2016 all round, actually: five months before the US election in November 2016, opinion polls and expert opinion alike “failed” to forecast Brexit.
Much financial market forecasting was hurriedly turned on its head, too, that night.
What will be the headlines tomorrow morning?
· Biden Clean Sweep: Democrats Take Both Houses To End Trump Era
· Trump Defies Polls Again: Shock Result Cements Republican Dominance On Capitol Hill
· Violent Clashes As Trump And Biden Both Claim Victory
One of the exquisite features of this election is the sheer uncertainty of it all. The apparent failure of opinion polls to forecast Brexit in June 2016 or the US election result later that year is overstated. It is not that the polls were “wrong”; what we should have learned from those experiences is to understand more deeply what probability means.
Tail events actually happen in the real world sometimes. A 10% chance event will happen quite often over time. Sometimes even 0.02% chance events happen: sports fans in the UK had a taste of this lesson in May of that same year with Leicester City Football Club winning the English Premier League title having started the season at odds of 5000-1.
Should we base our portfolios on trying to divine what these extreme events might be? To try to call the tops or bottoms of long cycles with precise timing? No. But nor should we act as if the most expertly-crafted models and analysis, which are right much of the time, can't ever be confounded by events with apparently tiny chances of happening. Robust portfolio construction is an often artistic endeavour drawing on precision and humility.
Perhaps the best way to spend this night, if I must stay up, is to re-read Daniel Kahneman’s Thinking, Fast and Slow or Nate Silver’s The Signal and The Noise: to remind myself of the importance of keeping an open mind, even when something looks like a sure thing.