If you drop water into sulfuric acid it’d be wise to do so wearing gloves, goggles and a lab-coat. This advice is applicable whether you intend to do it in Belarus, Botswana or Belgium. My point? That’s science where the same thing happens every time no matter who performs the experiment or where.
Then there’s financial theory. It’s called theory because most of it doesn’t work (reliably), the results are rarely [If ever? Ed.] the same and the outcomes vary depending on who performs the experiment and when and where it’s performed.
Notwithstanding the unreliability of financial-theory-models multi-trillion asset allocation businesses are built on their shaky-foundations and Nobel prizes are regularly handed out to its high priests.
Pioneered by Eugene Fama and Kenneth French (the former a Nobel prize recipient in 2013) one particularly wobbly theory continues to fascinate portfolio managers and academics alike and that’s the so-called Factor Model.
Briefly, the Factor Model (there are now many variations) says that stock’s returns can be explained by (originally three: market risk, stock size and valuation) a small number of variables. Understand these and you’ll know why things happened the way they did and, of more use, you’ll have an idea how trends may progress. There’s just one problem. The theory doesn’t work.
Not only didn’t the original model (from 1993) work prompting an overhaul (eventually and no doubt grudgingly?) by Fama and French in 2015 to incorporate more factors but also the original model and it’s 2015-reconditioned-version have been found not to work in other markets. Undaunted though fans continue to tinker with it hoping they’ll find a version that does work and the paper highlighted today is good example of this process in action.
Yang Liu (et.al.) from the Tsinghua University is the latest mechanic to take a fresh toolkit to this Cuban-clunker-theory and see if can be pressed into service in China.
Liu and friend’s work is not without merit and shows that an analysis of ‘Trend’ as a factor provides a lot of explanatory power for data collected on China’s domestic stock markets between 2000 and 2018. That conclusion jibes well with the feeling most China stock investor practitioners have about the trend being your friend, at least in China (in the U.S. BTW, empirically speaking, it’s not).
The big problem with this analysis, like the earlier work by F+F is it’s most likely a data-mined backward-looking conclusion and despite the fancy math ignores one of the most useful pieces of advice in the investing game i.e. that the past is no guide to the future.
You can read the paper in full via the following link Trend Factor in China and, notwithstanding my ranty preamble, it’ll be regarded within the industry as significant. So we should at least be familiar with the argument without necessarily buying into the conclusions.
Factor modeling is a shaky theory and the direct descendant of an even wobblier and more venerable theory that just refuses to die, the Capital Asset Pricing Model (CAPM). The CAPM, despite being perhaps the wobbliest of Cuban-clunker financial theories, is nonetheless still being taught despite over 50-years of hard scientific evidence that it too is just plain wrong.
Here’s a final thought from the world of practical investing on all this. Just because a lot of people believe something it doesn’t make it right.
Happy Sunday.