False positives and False certainty
October 8, 2020
While notionally about Covid (isn’t everything at the moment?), this is more about data, uncertainty and risk, ie the key elements of sensible investing. In their excellent work on Radical Uncertainty, former Bank of England Governor, Mervyn King and journalist and economist John Kay made the point about the danger of false positives in testing (they were actually explaining why regular testing for certain cancers may not be a good idea), something that is highly relevant right now in the Covid discussions. In a recent radio interview, UK Health Secretary Matt Hancock displayed a common misconception about false positives when he said that they were “less than 1%” and thus dismissed them as a problem. However, when the actual incidence in a population is less than the false positive rate, this can be a huge problem, for the actual meaning of a false positive % is not the percentage of positive tests that are false, but the percentage of total tests that are false positives.
To put some numbers around this. If you test a general population where actual incidence is relatively low (as opposed to a group showing symptoms) then you might expect to see anything from a few tens to a few hundred positive cases for every 100,000 tests. Incidentally, the UK used to report pillar 1 (hospital tests) separately from Pillar 2 (tests in the general population) but for some reason no longer does this. Some of these will certainly be genuine, but a 1% false positive rate – and the tests are unlikely to be any more accurate than this – means that you could also expect to see up to 1000 false positives! How do you tell the difference? For a good beginners guide to false positives it is worth looking here.
The problem then is similar to the one we have in markets generally, we are inputing too much certainty around an uncertain dataset. In markets, datasets such as earnings estimates trigger programme trading algorithms, while hugely generalised (and constantly revised) macro data on GDP and unemployment triggers swing trading in bonds and currencies. As traders and investors we are used to either playing the noise ‘game’ or ignoring it. In the same way we are used to ignoring ‘news’ such as the latest claim of “The hottest September ever” as Global Average temperatures are said to be 0.05 degree above the previous high, a measure of such utterly spurious accuracy as to be meaningless, other than to promote an existing agenda through grabbing headlines. This second example however leads us to a milder version of where we are now. It doesn’t matter whether the global temperature really is rising and in declaring the science ‘settled’ governments are effectively saying they are not prepared to consider any argument to their settled course, what matters is how government policy will respond to the data they have settled upon – however flawed.
Thus on Covid, it doesn’t matter what anyone may say on issues such as tests versus actual cases, the accuracy of tests, the new information that we have on treatments and so on, it is clear that governments have already made up their mind and are now suffering from extreme Confirmation Bias – i,e they will only listen to arguments that support them doing what they intended to do all along. The western economies that locked down hardest, including the UK, are doubling down on their actions and justifying it with reference to this ‘casedemic’. Sat in the UK or US, or even (still in quarantine) in Hong Kong it is easy to believe that it is only the local government wherever we are that is ‘ignoring the science’, but this is clearly a groupthink issue. All of them predicted a so called ‘second wave’ and all of them are pointing to increased test results as ‘proof’ of this despite rates of actual infection and hospitalisations running at less than 10% of April levels. All of them are also talking about a second lockdown and a need to wait for a vaccine with such a strong overlap in terminology as to appear heavily co-ordinated.
This then is the reality, even though it is based on highly uncertain data. Governments are shutting down economies in the manner of 19th century doctors recommending the use of yet more leeches when previous attempts to bleed the patient of bad humours (not surprisingly) failed. Investors are voting with their feet and sitting on the sidelines, having effectively written the rest of the year off.
There are however increasing signs of a pushback. The Great Barrington Declaration is an open letter to Global Governments to change course, signed by many, many eminent scientists and experts. At the time of writing, there are 4,248 Medical and Public Health scientists and 7.962 Medical Practitioners who have signed it – along with over 100,000 members of the public. It’s the first two groups that should really matter in influencing policy and we have to hope that once the great Political logjam of the US Presidential Election is out of the way, whoever wins will recognise the need to ‘move on’.