- Wealth PMS
In his 2006 research report Painting By Numbers: An Ode To Quant (via The Hedge Fund Journal) James Montier presents a compelling argument for a quantitative approach to investing. Montier’s thesis is that simple statistical or quantitative models consistently outperform expert judgements. This phenomenon continues even when the experts are provided with the models’ predictions. Montier argues that the models outperform because humans are overconfident, biased, and unable or unwilling to change.
Montier goes on to give diverse examples of the application of his theory, ranging from the detection of brain damage, the interview process to admit students to university, the likelihood of a criminal to re-offend, the selection of “good” and “bad” vintages of Bordeaux wine, and the buying decisions of purchasing managers. He then discusses some “meta-analysis” of studies to demonstrate that “the range of evidence I’ve presented here is not somehow a biased selection designed to prove my point:”
It’s an interesting read; an example displayed is using statistical generators to figure out if a patient is neurotic or psychotic – “experts” seem to get it wrong more often than the statistical generation of the result.
Montier argues the concept applies to investing as well – using quantitative models might remove all these crazy biases we have as humans. I have at many points cursed myself for not following a model, even if the result was beneficial to me; the point being, on average my discretion does nothing. Yet, I feel the need to prove my intelligence over the computer every once in a while, and for that I create a separate “education” fund – every time I override the system, the trade goes into my education and currently I’ve paid over a lakh for my education. There.
How does one invest in a “quant” way? Different time frames. Stocks have fundamentals – you must trust these fundamentals, so you first screen stocks where you believe there’s a reasonable element of trust in the declared results. Then you use a ‘formula’ to invest – buy stocks where EPS growth is higher than P/E, or buy stocks where 5 year profit growth is over 2x the Industry average, or where a combination of EV/EBIDTA, replacement value and book value is substantially better at current prices than comparable companies. It’s a pain to get all the data organized but once it is, writing such screens is not difficult. In fact, most of quant is not rocket science – and what might appear complex might actually be seriously flawed in the real world (eg. Black-Scholes).
Another aspect of quant investing is automated position sizing. It’s debatable that humans will beat computers on entries and exits – you could have either one win. But the power of computing is also in appropriate sizing of positions – do you buy 100 shares of this stock, or a 1000? Humans tend to focus on round numbers and on whatever they earlier traded. A person who would buy 1 lakh worth of every stock – say 5% of his portfolio – will buy the same 1 lakh worth even if his portfolio doubles. Worse, he’ll buy the same level when the portfolio is DOWN 30%. It’s human nature – we find a comfort level and stay with it. An algorithm, though, can throw out quantities based on stop loss risk, portfolio size percentage, strength of the entry/exit, pyramiding etc. In the long run, position sizing contributes substantially to returns.
I find that I’m more comfortable with a formula and back-test behind my investing. Since I like to stay human I still do some discretionary trading too, which has done well in stages and horribly in others. But I find the discretion harms me mentally far more than a losing position in a quant strategy.