The FT’s John Authers is a long-standing sceptic about market timing. But in his Long View column on 22nd February he sings the praises of an approach that sounded very much like ours, making adjustments to weights in different markets as a function of new calculations of probable returns as markets move around their long-term return trends. The surprise is not that it makes sense but that anyone would think it makes sense as a market-timing device or allows you to compete with any advantage in the great performance competition. It can make sense as a form of risk control, such as where outcome probabilities are important to giving clients confidence you will observe constraints they choose, such as meeting a minimum spending target, but not as a means of competing on performance.
The implication of adjusting allocations to changing future real return probabilities is constant and often only gradual averaging up and down, or in and out. Though a positive feature in a risk-control process, it is not a very sensible strategy if you are seeking to outperform other active managers or if that is what your clients expect of you.
Having run US pension and endowment money on this basis in the ’90s, as the first to offer what became known as Active/Passive, it was clear that the constant averaging up and down that results from allocating based on mean-reversion derived expected returns leaves too little alpha to compete with conventional active portfolio managers. It required an overlay of other rules drawn from the market timing toolkit, such as degrees of deviation from trend or momentum. It was therefore no surprise to find that the firm John Authers singled out, Elm Partners in the USA, has chosen to surround its mean-reversion thesis with degrees of deviation from trend and momentum. The first is logically a form of applying the second, the implication of slowing down the averaging process being that it will only increase returns if there is positive momentum.
The problem with this is that dependence on momentum does not give an active manager a competitive advantage. It works until it doesn’t and when it doesn’t it tends to blow up. It can work as an asset gathering strategy, because investors chase your winning streak, but not if you want to stay in the game. Besides, it were a long-term success strategy, everyone would adopt it. Or maybe they did and that is why it only works the way it does. Think how catastrophe insurance works, as that is probably the function long-term investors provide in public markets. Not recognising this is equivalent to farmers on the slopes below Vesuvius not questioning before AD79 why they got so many harvests each year.
Moreover, it is highly likely that while it works, it would work better if there were no input from any valuation measures. This was the unfortunate conclusion of back testing at my previous business. We we found the same was true when we tested the technical, momentum-based rules that Jim Slater adopted to modify selections based on his Zulu Principle: limiting the sample to the shares that passed a fundamental test weakened the results of applying only the technical tests to the entire population. But of course, the fundamental inputs serve a different purpose, describing an investment philosophy that sounds clever rather than naive, because that is what investors are more likely to buy.
Since we have so much experience of managing money at Fowler Drew (since 2005) on the basis of rebalancing to new probabilities without the aid of any technical market-timing tools, I thought I should share some numbers with the FT readers by posting a comment on I quote.
‘To give an idea of scale of the return-only benefits, there are two sources of alpha. The first is from the construction rules, as a probability-matching rebalancing approach will drop the bias to market cap (and so needs its own benchmark). There is potential for small efficiency gain from doing so, well-evidenced in academic studies. Then there is the tactical alpha relative to that policy benchmark. Without market-timing tools, that is probably only capable of generating about 1% pa. Hence my observation that the main source of added value for clients is the planning of limit-constrained, utility-maximising goals and managing the money to be constantly consistent with the plan. Alpha is icing but it cannot be the cake.
‘There are a number of other practical issues that need to be addressed before turning this into a valuable service, not least how to deal with currency risk and returns as you cannot obtain the mean-reverting local-currency returns without either hedging costs or currency effects which are very large relative to the limited alpha. There is also the question of what the appropriate risk-free asset is which is not necessarily cash if the objective is better risk control as well as alpha. But it has to start from a solid foundation and mean-reverting real equity returns derived from an index with known and constant decision rules are just that. There is a lot you can do with what little you know but market timing is not one of them.’
On reflection, I think John Authers’ scepticism has stood him in good stead. It’s why he is so widely read and well thought of.