Chris Gilchrist: Herds, spreadsheets and fund selection

Behavioural finance boffins have made us more aware of herding, which is argued to be a relic of humanity’s formative years on the African savannah.

Many of our hard-wired “rules of thumb” – such as the bird in hand being worth two in the bush – are often a poor guide to the most advantageous courses of action in the financial world.

Herding happens everywhere. It affects those who assemble buy lists of funds. Whether they are rating agencies, panel builders or discretionary fund managers, the evidence from sales data says they are converging on a smaller and smaller number of funds. One result is that funds run by talented managers are swamped with inflows that make it likely future returns will be lower.

In this case, the main villain is naïve faith in mathematics represented by so-called quant screens. There are few funds with conviction managers taking big positions that make it through this process.

Neil Woodford is one exception – yet the low volatility of the old Invesco Perpetual Income fund was, on the face of it, inconsistent with his huge exposures in sectors like tobacco and pharma.

Academics say it takes many more years of data than the average fund manager’s tenure to justify conviction that superior returns are the result of skill rather than luck.

Filtering the same data in an attempt to find winners is a flawed project.

Most of the people generating gigabytes of statistics have lost sight of the fact that they are not reality, and place too much weight on unstable characteristics like volatility and correlation.

If you accept the future is uncertain, then you should not rely heavily on any probability-based system.

Frank Knight and John Maynard Keynes identified the reason: uncertainty cannot be modelled. This is an insight that remains as true today as it was 100 years ago.

The number-crunchers remind me of the Middle Eastern story about Mullah Nasruddin. A neighbour found Nasruddin at midnight on his hands and knees under the street light near his home. “What are you doing?” asked the neighbour. “Looking for my keys,” replied Nasruddin. “Are you sure you dropped them here?” asked the neighbour. “Oh no,” said Nasruddin, “I think I lost them down that alley, but it’s dark there and here there’s plenty of light.”

In contrast, I argue – along with Fidelity founder Ned Johnson – that good portfolio managers are more like artists than actuaries.

Look at Woodford’s or Nick Train’s actual portfolio and try to construct it using quant-y rules. You will fail. The reason is that a good fund manager picks a stock for a portfolio. He does not buy a stock in isolation.

This is the kind of recursive holistic process at which the human brain excels: how does the addition or removal of one of, say, 40 ingredients alter the character of the whole?

But it only works if you create and retain a clear image of what you want the whole to look like, which is not a spreadsheet. Fund selectors ought to emulate this process and pay a bit less attention to the statistics.

Chris Gilchrist is director of Fiveways Financial Planning