Correlation Is Not Causation

Drummed into applied math students everywhere. It even has its own website, with this gem on how margarine consumption is correlated with divorce rates in Maine.

correlation

Should be true enough in markets as well. But in reality, at least in pockets, it isn’t always true. Stocks have always in part been driven by relative valuations. Stat-arb was a big thing some twenty years ago when computing power was starting to be applied to stocks. Pairs trading based on common risk factors makes some sense, Ford and GM operate in the same business after all, it makes sense they should be broadly be impacted by the same broad industry and economy trends. When computing power jumped later, factor investing came to dominate. Grouping stocks based on different attributes has some merit. At their heart, the old quants had valuation firmly in the mix of parameters. Many of the newer factors and machine learning quants have thrown out what ultimately matters. Price – or rather ‘value’. Low vol investing doesn’t care whether a stock is priced for perfection or not. Quality takes no account of what that pricing implies going forward, just that its metrics are stable. Momentum will push junk yields far below default rates and not even notice. As long as the quants see the property they like, regardless of valuation, away they go. They operate as if they are just observers, quietly taking a look from afar and being able to interact without impact. The Hawthorne effect is the phenomena where the behavior of subjects is altered due to the awareness of being observed. The quants in places are not observing any longer, and their impact is self-fulfilling, for a time anyways. There is plenty to be gained from applying stats and metrics to markets, but it is surely important to not take it too far.

You can see this today (September 9, 2019). ‘Value’ stocks are up a lot, not particularly based on the merits of the underlying businesses, but because other types of stocks are down. When stocks are held for their correlation properties, strange things happen. Like the butterfly that flaps its wings and causes a distant thunderstorm. It’s easier to make a case that at least today, retail stock Gap is up big because Boris Johnson chose to shut down parliament. Not often thought of as a butterfly, but bear with the logic here. Boris shut parliament…which catalyzed votes to stave off ‘no deal’ Brexit…which caused Gilts to fall…which drove global bonds to fall…which pushes growth stocks, utility stocks and REITS down…which makes value stocks jump. Seem strange? It should. But the stock market acts this way more and more. Factor investing and ETF baskets that segment stocks into groups are big drivers of prices, particularly when smaller names get larger weights in factors. We need to get back to a more fundamentally driven world.