Every now and then, you will run across a mathematical analysis where if you use a certain screening, trading, or other investment method, it produces a high return in hindsight.
And now, you know about it, because it was just published.
But wait. ?Just published?
Think about what doesn’t get published: financial research that fails, whether for reasons of error or luck.
Now, luck can simply be a question of timing… think of my recent post:?Think Half of a Cycle Ahead. ?What would happen to value investing if you tested it only over the last ten years?
It would be in the dustbin of failed research.
Just published… well… odds are, particularly if the data only goes back a short distance in time, it means that there was likely a favorable macro backdrop giving the idea a tailwind.
There is a different aspect to luck though. ?Perhaps a few souls were experimenting with something like the theory before it was discovered. ?They had excellent returns, and there was a little spread of the theory?via word of mouth and unsavory means like social media and blogs.
Regardless, one of the main reasons the theory worked was that the asset being bought by those using the theory were underpriced. ?Lack of knowledge by institutions and most of the general public was a barrier to entry allowing for superior returns.
When the idea became known by institutions after the initial paper was published, a small flood of money came through the narrow doors, bidding up the asset prices to the point where the theory would not only no longer work, but the opposite of the theory would work for a time, as the overpriced assets had subpar?prospective returns.
Remember how dot-com stocks were inevitable in March of 2000? ?Now those doors weren’t narrow, but they were more narrow than the money that pursued them. ?Such is the end of any cycle, and the reason why average investors get skinned chasing performance.
Now occasionally the doors of a new theory are so narrow that institutions don’t pursue the strategy. ?Or, the strategy is so involved, that even average quants can tell that the data has been tortured to confess that it was born in a place where the universe randomly served up a royal straight flush, but that five-leaf clover got picked and served up as if it were growing everywhere.
Sigh.
My advice to you tonight is simple. ?Be skeptical of complex approaches that worked well in the past and are portrayed as new ideas for making money in the markets. ?These ideas quickly outgrow the carrying capacity of the markets, and choke on their own success.
The easiest way to kill a good strategy is to oversaturate it too much money.
As such, I have respect for those with proprietary knowledge that limit their fund size, and don’t try to make lots of money in the short run by hauling in assets just to drive fees. ?They create their own barriers to entry with their knowledge and self-restraint, and size their ambitions to the size of the narrow doors that they walk through.
To those that use institutional investors, do ask where they will cut off the fund size, and not create any other funds like it that buy the same assets. ?If they won’t give a firm answer, avoid them, or at minimum, keep your eye on the assets under management, and be willing to sell out when they get reeeally popular.
If it were easy, the returns wouldn’t be that great. ?Be willing to take the hard actions such that your managers do something different, and finds above average returns, but limits the size of what they do to serve current clients well.
Then pray that they never decide to hand your money back to you, and manage only for themselves. ?At that point, the narrow door excludes all but geniuses inside.
How would you test a screening strategy when you suspect that returns over the last 40 years have been boosted by the secular decline in developed world inflation?
Perhaps a regression analysis of the returns of interest-rate-sensitive sectors versus the rest, and separate out the returns from interest rate movements versus other return factors.
Particularly over the last 35y, there’s a a sample size problem. Unless there’s a shorter than 3m horizon, oh the co lin y air a tee.
Going along the same thought process, do you see a similar oversaturation happening with the passive indexing boom that we’ve seen over the past decade or so?