In today’s post I want to look at long term TAA model performance in a different way. I think intuitively most investors realize that any strategy they pick will have periods of outperformance and underperformance. But decision making under real circumstances exposes us to all kinds of biases which often cause is to make intuitive, gut decisions based on incomplete or recent data. There are statistical ways to deal with these issues to help us make better decisions, in this case, to maybe choose the TAA strategies that not only perform the best over the long term but do so across different market environments and time periods. Many statistical prediction techniques use base rates . Base rates are basically batting averages. And in general, it is good to pick things with high batting averages – it is an indication that the higher performance (or lower performance) is not luck, that it is persistent, and will continue into the future.
** Note: in this piece I’m going to focus on returns which is only one metric to consider in choosing a TAA strategy. Risk adjusted performance, drawdowns, and turnover and other key parameters that need to be considered.
A few years back, about 2012, I learned how to use base rates in investing to help me pick the best individual stock quant strategies . I learned this from Jim O’Shaughnessy from What Works on Wall Street . Here is an excerpt on base rates from his must read book.
What I want to do here is to take a look at the base rates, aka batting averages, of various popular TAA strategies. I’m also going to use SPY-COMP, DM-COMP from the Economic Pulse Newsletter and compare them to a selection of TAA strategies ( with data from Allocate Smartly ). I’m using the longest period I can, 1973 through 2018.
One way to calculate base rates is took at performance over the whole period and various sub-periods. This helps provide context to investors. Ideally, we’re looking for strategies that outperform over all periods. I’ve chosen various buy and hold, TAA strategies, and various time periods. I then compare performance of the TAA strategies over the given period to the 60/40 benchmark portfolio. Underperformance is highlighted in RED.
- Avg TAA is the average of the TAA strategies in the table
- Meta is Allocate Smartly’s Meta Strategy which dynamically chooses among the TAA strategies
As the table shows over long periods of time, anything over 20 years, and over recessionary periods most of the TAA strategies outperform the 60/40 buy and hold portfolio. Some much more than others. No surprise there. But in time periods of 10 years or less, in particular the last 10 years this is not the case. You can also see there is quite a difference amongst TAA strategies, with some having no underperformance periods and other having many.
Now, let’s go a bit further. 10 years is a long time and my selection of specific time periods adds context but is most likely biased in the selection of a specific period. Let’s use ALL rolling 3-year periods and calculate the base rates, the number of periods the strategy outperformed over 60/40 as a percentage of all rolling 3-year periods. The higher number, the higher the batting average, the better. The other thing I did was to calculate base rates with respect to something else besides the standard and very US centric 60/40 portfolio. I used two other benchmarks. A more global portfolio, GAA , which has a much higher foreign stock allocation and also allocates to real assets like real estate and gold. And I used the Vanguard Capital Markets Model , which basically takes the 60% allocation to stocks in 60/40 and allocates 40% to international stocks. As the largest asset manager in the world, the Vanguard model is a much more representative portfolio of the typical investor portfolio than the standard 60/40 portfolio. The base rates are shown in the table below.
I’ve chosen 70% as a good base rate (choose your own, I’m probably biased) and highlighted those strategies with 70% or greater base rates. That means that the given strategy outperforms the buy and hold model in at least 7 out of 10 rolling 3 year periods. As you can see, certain strategies are clearly better than others on this metric. VAA, GEM, SPY-COMP, and DM-COMP have the highest base rates across the different benchmarks. The last line in the table looks at the underperforming periods and calculates by how much the strategy underperformed in that period. When you do underperform ideally you don’t want to underperform by much. For example, for VAA in the periods it underperforms, it underperforms by 40% vs GEM, which when it underperforms it does so by only 26%. Finally, I wanted to show you the whole set of rolling 3-year periods from 1973 to 2018. See the chart below. There is a kind of trend with TAA strategies. When they outperform, they do so for a while, and vice versa, when they underperform, they tend to do so for a while, like over the recent few years. This is completely normal and should be expected.
In summary, base rates are an important tool in analyzing the performance of strategies. The higher the base rate, batting average, the higher the odds are that the strategies outperformance is not a matter of luck. Base rates are not the only criteria in choosing TAA strategies, but I think they are an important one that investors should consider.