On the last page of his book, Homo Deus, Yuval Noah Harari poses the questions:

“What’s more valuable – intelligence or consciousness? Intelligence is decoupling from consciousness. What will happen to society, politics, [investment] and daily life when nonconscious but highly intelligent algorithms know us better than we know ourselves?”

I have added the word investment.

This book certainly sets the mind noodling.

We are often being told – mainly by academics and parts of the FT – that there is little point in active asset management. Their conclusions, often as not, are based on the average performance of the average manager.

We think not and wish to rebutt such a reductive line in reasoning. Who would ever be interested in the average outcome from an unsampled universe? In football, there is little point in supporting an average team – unless you happened to come from the town itself. That’s why people all over the UK (and wider) support Liverpool, Manchester United and Arsenal and why Mansfield Town – who finshed exactly half way down League Two last season – has an average home gate of 3,774.

To our minds, the universe of – equity managers in this case – that should form a basis of enquiry and statistical analysis should exclude managers who hug closely to indexes and display skittish behaviours, easily measured by fund turnover.

A 2015 paper by Martijn Cremers and Ankur Pareek advanced evidence that the average outperformance of high active share portfolios (meaning non-benchmark) with long holding periods was 2% per annum, after fees. Fergus Shaw reflects on their work:-


It is common to think that the US based tracker manager Vanguard will eat the collective middays meals of the active asset management industry. But is not Vanguard at more significant risk of disruption? From Google, for instance?

We attended a conference last year at which Demis Hassabis talked through how his algo, which is called Deep Mind, beat the world champion of the ancient Chinese game Go. Go is many times more difficult for either a human or a machine to master than chess on account of the many more possible combinations of moves. This party trick was impressive enough for
Google to buy his company for US$500mn.

What is Google up to? A myriad of things. And retail investment management may be in its play book – a few sections behind the driverless car.

In the search for a crisp definition for what long term investment entails the best we have come across so far is the term “decision making under conditions of uncertainty”. The expression does not hale from finance but from foreign affairs and presumably makes up a large part of The Minister for BREXIT’s in tray. Decision making under conditions of uncertainty in an investment context goes a bit further than the challenges faced by the driverless car.

How does the driverless car manage uncertainty? For example, the kind of uncertainty provided by dirty road signs? The real-world problem here is that no two dirts will be the same but the car needs to recognise that a sign is dirty and decide, by referencing its own database of past observations, what lies behind the dirt.

Taking this issue across to investement, in a capital market transaction, whether company A acquires company B may or may not be a good thing and this is not knowable with a high degree of certainty. The managers of the companies in question are vested so their reports may not be accurate or honest. Circumstantial evidence abounds but do the variables submit to computerised learning even if the computer can learn from itself?

It is a merit of computational processes that they look upon correct answers and mistakes devoid of emotion. It is the nature of human beings to freight their decisions with emotion and then discard part of the essential learning from past mistakes. A good asset manager, we suggest, has a less than average emotional reaction to his or her mistakes (which may not be mistakes but just the limitations of inadequate foresight) but more open than normal candour is exploring the prior reasoning behind these mistakes.

Russell Napier, of our Investment Committee, set up the Library of Mistakes in Edinburgh to address the fact that investment is one of the few human areas of human endeavor where many participants willfully choose not to learn from past mistakes.


Indeed, pattern recognition is a necessary feature of both the human and computer minds when applied to investment. Both conscious and unconscious processes exist to bring forth the thought “this may not be the same as before but it bears similarity and is probably relevant”.

The question is, and apologies that no answer is provided here, whether that similarity can be reduced to a number in every part of the process. If so, then computers have a shot.

Quants are not new in investment. They pre-date the computer age. Humans have been operating quant tables since the Roman’s shifted salt around their empire.

In the Global Leaders equity investment programme, we look to identify and invest in leading companies in long term growth sectors, we use quantiative factors to reduce the number of listed companies on world equity markets (253,885 traded equity securities accessible through a Bloomberg terminal) to a manageable number for further qualitative research – done by humans.

Whether it is the festering mess off Europe’s southern borders or the argument we in Britain have picked with Europe or Trump in the White House, these things serve to remind us of the folly of nations. In investment terms, we’d rather bet on companies not countries.

If you would like to obtain a summary document on the Global Leaders in fund form, please email Tom Milnes.

James Spence
Managing Partner