Passive investing can beat a stock picker, but can it beat a quant?
As a follow-up on the last post about trusting your model, here’s a striking comparison of a passive ETF version of my portfolio to the actual one.
I’ve taken my current allocation goal and replicated this using ETFs. The allocation is as follows:
- long EU bonds 15%
- intermediate EU bonds 5%
- global bonds 5%
- gold 5%
- cash 10%
- EU and US stocks 60%
For the year starting May 2014, the result is that the ETF portfolio gained 14.4% vs my own actual portfolio performance of 12.3%. So the ETF portfolio was clearly better, even though this may not mean much based on evidence from just one year.
But I repeated the comparison since the start of this year because I changed my stock allocation on January 1st. I added more stocks, made the positions smaller, and I now have 27 stocks in my portfolio, which is twice the number of positions I had in 2014. This means that the individual importance of every single stock is decreasing further and the overall performance I’m seeing in 2015 is more likely to be attributable to the underlying properties of the whole group of stocks and not just 2 or 3 of them.
What this looks like, compared to the passive ETF portfolio? Have a look:
The obvious observation is that my own line is much smoother (shallower valleys and peaks) and that it has risen about three times as much as the ETF portfolio tested, which is quite a big difference. Since about half my personal portfolio is in EUR denominated instruments and half is in USD denominated instruments the devaluation of the EUR is not responsible for the extraordinary gains (since one half would more or less lose the same amount of value that the other gains).
In academic papers it’s often said that indexes (and by extension, index ETFs) outperform active stock picking by a wide margin. I believe that to be true. However, I have effectively switched from active stock picking with a big quantitative component (which I used in 2014) to a purely quantitative model that buys whatever my model spits out without doing any research on the companies. It appears to have been the right choice so far. So while an individual stock picker might not beat an index, a quantitative model very well may.
The secret, then, is in having the stamina to stick to a model without faltering, no matter what happens. I haven’t yet seen a real bear market so I can’t promise I’ll stick to it. But the data I’m getting from these comparisons is reinforcing my belief that it would be the right thing to do.