Diversification, Now More Costly Than Concentrated Portfolios?
Updated: Oct 5
2019 and we need to talk about diversification. Why? Because it no longer is the happy go lucky approach that renders one in the magic nirvana of lowered risk (than a concentrated portfolio) and excess returns: read "alpha". To be clear, we are referring to diversification within an asset class, not across asset classes. For example, the SP500 would squarely sit in the highly diversified equity asset class, but problematic category.
Case in point: our ARBi100, an active quant AI-like approach applied to 100 of the largest stocks in the US. After many years of superlative "alpha" performance, it seems to be struggling to deliver alpha while our ARBiIND, 30 mega cap stocks, is delivering alpha in droves over its benchmark (the DOW30 index). Adding more insult to injury, ARBi100 is daily rebalanced vs that of the ARBiIND, which is monthly! That is indicative enough (to us) that something is amiss in the world of highly diversified portfolio management in liquid markets.
So, if our ARBi monthly rebalancing is generating alpha on a more concentrated portfolio of 30 stocks, then it is obvious to us that the markets are continuing their unabated move to greater efficiency. Therefore, future alpha will be more a function of a solid, unbiased strategy applied against a smaller portfolio and, in so doing, the alpha bleed of excessive diversification (like 100 stocks) can be curtailed.
While we have been reviewed this for some time, 2018 made it quite clear that the nature of all that ARBi100 diversification is that it can lead to a lot of dead money. That is, stocks that generally are not moving enough such that the system can extract some alpha via daily rebalancing. So, the capital sitting on those stocks ends up eating into the total return and otherwise diluting the alpha appeal of the product with a corresponding increase in risk to boot! A double whammy, which pushes an otherwise stable adaptive investment process into a more passive-like risk/return profile.
Research terms abound for this topic and we like Intech's (owned by Janus Henderson) discussion on the topic (see their measurement tool too) that otherwise describe how these dynamic elements can surface in broad baskets of stocks. S&P also puts out similar research and calculations but the point is that if the market has now moved to a world of concentrated exposure (e.g. more efficient) and, generally stays that way, large, highly diversified baskets of stock can deliver on the beta premise (track the overall market) but are hard pressed to beat it altogether.
Our chief takeaway is that less diversification/more concentration provides a better natural angle for an active manager to deliver on the alpha premise. That said, we don't want to sacrifice risk reduction by being too concentrated so our focus will be on delivering product portfolios that generally don't go more than 30 individual positions. As a sidenote, large baskets help with capacity but as an investor, do you want to pay for that capacity? Well, you are forced to pay this 'tax' if you invest with multi-billion investment managers who need more stocks to spread your money out on. You should be concerned for your money only or your client's (if you are reading this as an advisor).
At present, our chief products are now all geared as such and are performing at levels most would consider very attractive relative to that of their representative index and certainly that of most comparable active management processes we are aware of.
Moral of story, don't get sucked up into the excess diversification trap. ARBi100 rest in peace, long live ARBi030 or less.