Man or machine? Which would you trust to look after your money? Thirty years ago, there would have been no question that a human hand would get the vote. Now, investors are having to get used to the idea of having computers run their portfolios, albeit still under the watchful eye of a sensible fund manager.
Two forces have driven this trend. One is simply that the technology is now in place. It is no longer a technical marvel for a machine to do the following: 1) screen a universe of available stocks; 2) pick some given proportion of them as candidates for a portfolio; 3) calculate what impact it will have on the portfolio's risk-reward characteristics if you combine these candidates in different proportions; 4) optimise the portfolio so it meets the statistical specifications you have set it; and finally 5) buy and sell the shares in a single programme trade, executed by an automated trading programme. All these things now happen, on a scale you might find surprising - and in the blink of an eye.
The second ingredient has been the gradual acceptance by providers and consumers alike that using machines to run investments is not only possible, but in many ways desirable.
For one thing, it can be much cheaper and more cost-effective to operate an investment programme this way. For another, there is a powerful body of experience to suggest that machines are rather better equipped than humans to do the job well. It is this second leap of faith that is arguably the more significant.
What computers have that the average human fund-manager lacks are two essential qualities: discipline, and an absence of emotional distraction.
Machines don't easily get panicked, or fall in love with what they own. And they only do what they have been instructed to do. They have rules and they stick to them. Of course they will do stupid things if their instructions are stupid, or their methodology is flawed - which is what brought down the eggheads at the hedge fund Long Term Capital Management eight years ago. But discipline and emotional intelligence, so behavioural finance theorists have found, are proven strengths for any fund manager, robotic or human.
Two recent case studies show how vital these strengths can be. One is the success Philip Wolstencroft and Peter Saacke at the fund management house Artemis have had with their "SmartGARP" methodology. This is a stock screening system originally developed by Wolstencroft when he was the market strategist at Merrill Lynch (originally Smith New Court). So well did this computerised stock-screening system work when used to pick model bull and bear portfolios that it was logical for Wolstencroft to take it from the sell side (stockbroking) to the buy side (fund management).
Artemis gave him the infrastructure and, eventually, the funds to allow him to test how his system worked in practice. The answer is: remarkably well. Wolstencroft's original fund, Artemis European Growth, has grown from £50m at its launch five years ago to £1.2bn now. Two years ago, Artemis launched a second fund of a similar kind, a Global Fund managed by Saacke. Its performance has been even better.
It would be wrong to describe either fund as 100 per cent quant-driven. The shape of the European fund, Wolstencroft reckons, is about 75 per cent driven by what the SmartGARP model suggests, and 25 per cent by adjustments its two managers choose to make.
The Global Fund, which owns 75 individual stocks (there are 50 in the European fund) is about 90 per cent the work of the model, says Saacke. The two say that in cases of doubt they have learnt to stick with what the model suggests.
The simple reason is that the model, which scores every stock in its universe on a wide range of criteria (the main ones being measures of value, growth, momentum and earnings revisions) has earned its spurs in confrontation with the market. By its nature, the computer is immune to the distorting influence of market "noise" and news headlines, let alone the market's twin poles, fear and greed.
Only at moments of abrupt changes in market direction, such as in March 2003 when this bull market began, does the model tend to falter - it takes a few weeks to pick up all the new signals. Even then, it tends to recover very quickly. The main job of the two managers is to monitor the output of the model and check its findings are robust when measured against common sense and, occasionally, their better judgment.
A second team of machine-aided fund managers I have started to take an interest in are at JP Morgan Asset Management. Their investment process, which picks European stocks and combines the best in a range of stock portfolios, has a record of consistently strong risk-adjusted performance since it was first developed in the 1990s. The JP Morgan process is the engine that drives the recently launched Growth and Value funds at Norwich Union, for example.
Can the machine-aided teams match the best active fund managers of the old school? It will be interesting to see, especially if we are now approaching some kind of bear market setback, as Anthony Bolton, for one, appears to believe is likely.
His stance contrasts with that of another fund manager whose views I follow, the American Ken Fisher. Fisher thinks the bull market in equities will remain robust well into next year. For what it's worth, the Artemis model SmartGARP appears to agree; it is still positive, for example, on a number of resource-based stocks. So far, the two Artemis managers have resisted the temptation to override.Reuse content