Face Clouds, Patterns, And P-Hacking

Without a pattern you're just guessing

Chimp Investor is ranked #3 of 90 by Blog Overview and #10 of 100 by FeedSpot in their relevant categories. A review on WealthTender would be much appreciated.

I remember with amusement a presentation some years ago by a technical analyst, a so-called chartist. Having been asked by a member of the audience to analyse one chart, he proceeded to draw on it various lines and triangles, concluding the price would fall over the subsequent months. It was then revealed to him that the chart had been created using a random number generator which rendered it utterly unpredictable.

There is of course a serious side to this story. All investing involves making predictions about prices, whether directly or indirectly. Even passive investing involves predictions; if you invest in a particular passive fund, you are implicitly predicting it will perform in a certain way.

It is therefore important that there is a sound basis to your prediction. In other words, that you have identified - recognised - some underlying pattern, not an appearance of pattern as had the technical analyst.

Pattern recognition is the basis of our perception of reality. When we look at a tree, our brains register the pattern of what we call photons bouncing off the tree as well as off objects behind and around it. Then, by comparing this pattern with the countless number of patterns of other objects stored in our memory – umbrellas, bacteria, elephants, trees, etc we recognise it as a ‘tree’. Given that we are, in effect, pattern-recognising machines, it is not surprising that from time to time this process goes wrong.

For example, we might see a face in a cloud or in a piece of toast, or shape to a random ‘price’ chart, and attribute meaning to them.

There are statistical techniques – hypothesis tests - for judging whether a pattern is real or apparent. In the case of face clouds, you build a theoretical model for all clouds based on how particles of water vapour aggregate.

You then determine how many of them should contain faces and compare this number with the observed number of face clouds. If the probability of seeing the observed number of face clouds is below a certain level, 1% say, you may be onto something.

Of course, you may also be hallucinating, as you might be if you saw an ice cream van – or a tree - in the middle of a barren desert.

Even discounting this latter possibility, you can never be 100% sure that the face of the Virgin Mary on a piece of toast is not there by chance – ie that it is a divine message. Or that a tree is a tree – it may be some freak collection of photons. However, statistics will get you there, practically speaking. Unless, that is, you cheat.

By ignoring clouds that do not contain faces, the observed proportion that do would be much higher. It has been suggested that this behaviour - known in the trade as p-hacking - routinely happens in academia, where scientists can improve their promotion prospects by publishing interesting findings.

I encountered this sort of thing first-hand while doing a dissertation last year as part of a science degree. I, politely though perhaps naively, challenged the authors of a paper that claimed there was a pattern in the ages of a handful of South African cave rock sediments. This pattern, according to the authors, meant an interesting conclusion could be drawn about our ancestors, a tiny number of whose remains had been found in the rocks.

I asked the authors for details of the paper’s rock age calculations so I could try to replicate them. In return, I was accused of conducting a witch-hunt and of being out to ruin reputations.

Being able to replicate a paper’s findings is the basis of empirical science, which is why the issue of the allegedly widespread p-hacking – one paper is titled Why Most Published Research Findings are False - is called a replication crisis. It is also the basis on which investment funds particularly systematic funds such as ETFs and quant funds should be appraised.

Last year, Campbell Harvey, a former editor of the Journal of Finance, an adviser to investment firms such as Research Affiliates and Man Group, and the writer of 150 finance papers, several of them prize-winning, published a paper titled Be Sceptical of Asset Management Research. Harvey analysed the performance of quantitative investment strategies and found that 400 of them were bogus; while their back-tested pre launch performance was strong, this ceased upon launch.

My suggestion is to look at the assumptions that underpin your investment portfolio and to challenge them. A good place to start might be your implied assumptions about the long-term outlook for inflation.

If you have a balanced portfolio, one with a meaningful bond exposure, you are assuming, whether you know it or not, that the currently high inflation will be transitory.

There are some interesting parallels with the late 60s when excessively loose fiscal policy in the US saw inflation rise from 1% to 5%.

As I noted here:

"From 1965-1969, government expenditure increased at a rate of 11% per annum. The inflation rate in 1964 had been 1.2%. By 1969 it was 5%.

"Fiscal and monetary policies were tightened in 1968 and 1969 which caused a recession in late 1969 and early 1970. Inflation peaked at 6.4% in April 1970 but then fell only gradually, reaching a low of 2.9% in August 1972. However, it did not stay low for long. By the end of 1974 it was over 12%, exacerbated by the 1973 oil crisis. It then averaged 9% for the next 7 years, a period that included another oil shock.

"Many attribute the high inflation in the 70s to the two oil shocks. However, these just exacerbated inflation that was already elevated. The inflation genie had been let out of the bottle in the mid 60s and it would not get put back in until the early 80s when Paul Volker showed up on his white horse."

If there is a recession in the coming months as there probably will be, monetary policy induced or otherwise, bond yields are likely to fall as they did from 1970 to 1972 (see chart below). Back then, this was an opportunity to sell or reduce bond exposure ahead of the real carnage of the 70s.

Source: Credit Suisse, Yahoo! Finance

The views expressed in this communication are those of Peter Elston at the time of writing and are subject to change without notice. They do not constitute investment advice and whilst all reasonable efforts have been used to ensure the accuracy of the information contained in this communication, the reliability, completeness or accuracy of the content cannot be guaranteed. This communication provides information for professional use only and should not be relied upon by retail investors as the sole basis for investment.