This talk was given 26, 27, and 28 February 2008
The text here may not be identical to the spoken text
Invesco and Professional Pensions seminars on
Quantitative Investment Strategies
Welcome, welcome. I understand you’re mainly pension trustees. Thanks for coming and giving up that fascinating meeting on whether you should move to the CMI 92 series long cohort life expectancy model.
We’re here to talk about quant. Yes children it’s the maths class. I’m very keen on maths. I always used to annoy my arts graduate friends by saying you couldn’t understand anything that you couldn’t measure or count.
So numbers are good. But does quant work? And if so how?
Just because the last three sentences had four words in them does it mean the next one will? Clearly not! It was 18. And then two and three – oh God you can get obsessive about it can’t you? But that’s the essence of it. Looking at numbers and predicting the next one. And being right even 51% of the time – or even less I’ve been told – can make you money.
That’s what brought down Jérôme Kerviel – who worked in Soc Gen’s Delta One office which dealt in quants. And lost the bank €5 billion.
My colleagues on a programme called More or Less – one of the only Radio 4 programmes about numbers rather than words – interviewed a quant – which I understand is the name for the people who do it as well as the system. Just as a hundred years ago a typewriter was actually the woman who did the typing. Anyway this guy lived in Hampstead and told the programme "My model is one of those that prints money. It doesn’t make astronomical amounts of money. Algorithmic traders tend to make $10 - $20mn a year for their bank and they keep $2-$4m - £1-£2 million for themselves."
The geeks who are the quants who work out these formulae have been described as having no social skills but are clever and obsessively interesting – sorry interested!
There’s a joke at Goldman Sachs that there are two sorts of worker there. The Haves and Have Yachts. Among quants they don’t just have one yacht – these mathematicians are the Geek Shipping Magnates.
But the most interesting thing in that programme broadcast last October was the view that quants were becoming so dominant that people will need maths and physics to get a job in the City in the future.
So are quants really going to take over the investment world? I’ve seen one estimate that by 2010 algorithmic trading will account for half of all US equity deals.
And that’s despite the fact that they don’t always work. Particularly it seems in August. Maybe when there’s an 8 in the month. Anyway last August quants were blamed for the excess volatility of the credit crunch. And I thought it was caused by commission driven sales of mortgages to people with little income and less paperwork.
So. Is quant for you?
I have to tell you my own bias is towards cash. Last year FTSE 100 share prices rose by 3.8% - less than you could get in cash. This year they lost all last year’s gain and more in less than two months. Last night it was still 9.7% down from their high in June last year. Advisors say ‘Yes but shares are for the long term.’ Mmm. Let’s go back nearly nine years to when ISAs began on 1 April 1999. People who invested in cash ISAs when they began and moved their money to follow the best rates have generally done far better than people who invested in shares ISAs. The FTSE 100 index is currently nearly 300 points – about 4%– below where it was on 1 April 1999. And when the index closed last night at 6076 it was pretty much where it was on 10 April 1998. Nearly ten years and no growth.
That is the challenge that quants and other managed funds have to beat. To do better than a flat index over nine or ten years. After all index funds follow markets down as well as up. Many managed funds fail to do any better. Some do a lot worse – the so-called dog funds.
This year I presented the dog awards for financial advisers Bestinvest. Each year their researchers collar the mutts of the investment world. They sniff out the funds that have been really bad. And publish their list of dog funds – that have performed worse than their benchmark index every year for three years and ended up at least 10% below their benchmark at the end of that three years. That’s not just a bad dog. That’s shut outside for the night in the rain bad dog.
Two examples that caught my eye were funds which had actually lost their customers money. The Newton Japan fund. While the market rose by 7% the Newton Japan fund obeyed the law of gravity and fell by 6%.
The managers of CF Canlife Smaller Companies selected the shares to invest in so well that the fund did indeed get smaller. It shrank by 12% even though the FTSE index of small companies grew by 32%.
And although they lost money they still charged their customers an annual management fee for doing it. On average the dog funds charged 1.65% a year to do worse than a simple tracker.
Niels Bohr, one of the greatest physicists of the 20th century and the man behind what we now call quantum physics, said ‘It is very difficult to make an accurate prediction, especially about the future’. And of course that’s what investing needs.
Niels Bohr’s view was supported by the FSA in 2000 when it published research into whether investment funds that performed well one year were more likely to repeat that success the next. It found no persistency in good performance. This year’s star could be next year’s meteorite plunging to earth in a shower of burning banknotes. It concluded that past performance is indeed no guide to future performance.
Of course some people think the answer is not to follow a fund but its manager – and they normally whisper ‘Anthony Bolton’ or of course the wonderful Neil Woodford. But research done by Citywire published in July 2004 showed that out of 175 fund managers with a five year track record, only 11, that is 6 per cent, one out of 16, managed to do better than the market average every year for five years. And there is no evidence that even this élite will continue to do better in the future. We’ll here more about track records later.
So step forward the quants. Can maths do what tracking by its nature can’t do and what human judgment on the evidence fails to do consistently – beat the market? Is Quant the right fashion for the 21st century?
And it is of course quants that we are going to hear about today...