TAA Investing


In today’s post I want to take a look at an alternative measure of risk in portfolios, the Ulcer Performance Index, and use it to rank the various TAA and buy and hold strategies.

First, let’s talk about the Ulcer Performance Index (UPI) . The Ulcer Performance Index (UPI) is similar to the Sharpe ratio in that it is intended to be a measure of risk-adjusted return. The difference between it and the Sharpe ratio, and what makes the UPI a better measure of risk, is that it uses the magnitude of drawdowns (all drawdowns, not just the maximum drawdown) and their duration as the measure of risk. Think of it as a measure of how painful it is to live through implementing the particular strategy. Hence the name, Ulcer Performance Index. Or think of it as a measure of the bumpiness of return path of the strategy. See this post for a good explanation of what the UPI measures and how seemingly similar strategies can have very different UPIs. I particularly like this quote:

The smoother trip is likely to lead to less stress, not to mention wear and tear on the car, which can cause further headaches.

Essentially, smoothness is better than bumpiness when managing your investments. It means fewer mistakes and in general better outcomes. One of the biggest benefits I think of using UPI as one of the main parameters in choosing your TAA strategy is that it is the indicator most closely correlated to higher safe withdrawal rates in retirement (SWRs). Because SWRs are so dependent on the sequence of returns, more specifically the sequence of negative returns, using a measure that takes that into consideration is important when you are entering or are in the withdrawal phase of your life.

The other thing to note about UPI is that since it does incorporate returns in the numerator of the calculation it does vary quite a bit by the time period under consideration. In my comparison below I’m going to use UPIs calculated over the last 20 years, which in general are lower than UPIs over longer periods of time. Why? Mainly because the last 20 years saw some of the lowest nominal returns for risk assets since the mid 1960s. The denominator of the calculation, the Ulcer Index itself has not changed significantly in the last 20 years as compared to longer periods of time. Just to give you an example, the UPI for the Meta strategy over the last 20 years was 2.86 whereas it was 3.77 over the full time period going back to 1973. In other words, just make sure to compare UPIs across strategies over the same period of time.

OK, now let’s look at the data. In the table below I’ve ranked various TAA and buy and hold strategies by UPI over the last 20 years. I’ve also included annual returns and max drawdown for each strategy. There is generally a correlation between the measures but not always. For example, see the difference in UPIs between Vigilant AA – Aggressive and Vigilant AA – Balanced. Buy and hold strategies are in green and my Economic Pulse Strategies are in yellow. Data is from AllocateSmartly and from my own backtests. For data over longer periods of time see AllocateSmartly or drop me an email for my own strategies.

First thing to note is that in general this is an area where TAA strategies really shine, handily beating out the 60/40 benchmark by a longshot and buy and hold in general. But with some better diversification and portfolio construction a buy and hold portfolio, like the All Weather Portfolio, can produce some much better results than the standard buy and hold. Also, there is quite the range of UPIs across various TAA strategies. The higher ranked strategies tend to be quite aggressive in managing risk, or alternatively target higher returns for a certain level of drawdowns, but do not target the lowest level of drawdowns. It is not a free lunch, often the higher level of UPI comes with a lot of trades and turnover but that is a topic for another day.

In sum, I think UPI is quite a good measure of risk-adjusted return, better than Sharpe, Sortino, and max drawdown. This is especially true for investors in the withdrawal periods of their lives where a significant negative sequence of returns can ruin withdrawal plans. And in general it also works for very risk sensitive investors who are looking for a ‘smoother’ ride. I wasn’t such a big fan in the past but over time, and with the friendly prodding from a few of my subscribers I’ve come to like UPI as a risk measure quite a lot. A shout out and a big thanks to them for that.


Tagged , , ,

7 thoughts on “ An alternative way to look at risk in TAA strategies

  1. Is Gary Antonici’s dual momentum the same as traditional dual momentum in your table? If so, any thoughts on why your metrics are so different from the ones in his book?

    1. Yes. One and the same. As for the differences, one is time period. For this post I used the last 20 years of data only. Two, AllocateSmartly uses a very stringent set of backtests rules that includes fees and transactions costs. All of the strategies are tested in the same framework to get apples to apples comparisons.


  2. I can only agree. I found UPI because my backtesting software used it a lot, read how it’s computed, and it’s my favourite performance since. Then I started using Allocate Smartly where it’s also used as a main metric.

  3. It would be also interesting to incorporate the turnover of each strategy. Some of theses strategies can have wild turnover of over 500% a year. Turnover should be a good predictor of the expected differences between backtest vs. real life (mostly due to tax & commissions).

  4. Trading various days of the month can improve the performance, especially for volatile strategies like VAA Aggressive. If you plug that into AS with day 7 33%, day 14 34%, and day 21 33%, you can see this would be the number one performer pretty sure. Or days 5, 10, 15, 21 each 25%, close to the top too, albeit not as good, but you get a sense.

    Combining strategies smooths the ride even further, and goes to the what, the when and the how of diversifying per NewFound Research. UPIs in the very high 4’s, low 5’s can be seen historically. I perform various robustness checks (changing strategy weights, removing each strategy…)

    Anyone not combining strategies is missing a big opportunity to take UPI to an even higher level. Thanks for this piece Paul

Comments are closed.