Author Topic: Using Sharpe and Sortino Ratios to determine an Asset Allocation  (Read 4019 times)

MoNoGro

  • 5 O'Clock Shadow
  • *
  • Posts: 5
  • Location: ABQ, NM
I recently discovered MMM and been skulking around the forum for several weeks. In the past I have tried picking individual stocks, taken some silly risks and have little to show for the effort.  Better late than never, I found the discussions here very informative and have come to  realize that beating an index or timing the market is a fool’s errand.

There are a number of opinions regarding AA in the forum, but I haven't found much discussion about risk mitigation beyond diversification using total stock funds.  I am by no means an authority on Sharpe or Sortino ratios, but what I understand is they allow one to compare downside risks among individual stocks or AAs in relation to potential gains.  I’m curious if there are any Mustachians who have used these ratios to guide their asset allocation?

I recently backtested the built-in passive AAs available at portfoliovisualizer.com (Swensen, Ferri, Swedroe, and Bogle) and compared them to the basic 90/10 AA and jhcollin’s recommendations.  My goals was to see what AAs mitigated downside risk with potentially higher returns.  For no real reason, I limited my analysis from 1994 to 2014, rebalancing annually.  The results are in the attached excel file.

To me, what is most apparent are the AAs that include REITs (Swensen’s Yale and jhcollin’s original) generally have larger CAGR, lower standard deviations (less volatility) and higher Sharpe and Sortino ratios.  I am interpreting this to mean greater returns with less downside risk compared to simple stock/bond splits. The response with the inclusion of international stocks is mixed  with respect to volatility…I’m wondering if Swensen’s Yale performance is higher because of the REIT? 

In the final column, I created a potential distribution that modifies jhcollin’s 50/25/25 (total US/total bond/REIT) to include a small portion of mid- and small-cap value stocks.  The CAGR and Sortino ratio are the highest amongst the bunch.  Certainly one could analyze an indefinite number of AAs, but I‘d like to get the community’s impression about using Sharpe and Sortino ratios to determine a high yielding AA with potentially less volatility.

innerscorecard

  • Pencil Stache
  • ****
  • Posts: 589
    • Inner Scorecard - Where financial independence, value investing and life meet
Re: Using Sharpe and Sortino Ratios to determine an Asset Allocation
« Reply #1 on: January 14, 2015, 08:05:45 PM »
Why do you care so much about volatility?

rmendpara

  • Pencil Stache
  • ****
  • Posts: 610
Re: Using Sharpe and Sortino Ratios to determine an Asset Allocation
« Reply #2 on: January 15, 2015, 07:11:50 AM »
Different uses.
 
Sharpe is simply a measure of an assets expected return relative to risk, instead of (some) people who just measure performance on absolute return regardless of measuring how much risk was taken on.

To try and make the point, I would avoid calculating these things, because asset classes will experience a very different volatility profile in the next 2 decades than in the past two. US Treasuries, as an example, have been extremely volatile over the past few years.

If your real goal is to minimize volatility, then by all means go ahead with your approach. Your real goal should be to continue adding asset classes with expected real returns (i.e. outperforming a risk free return and/or inflation) within your portfolio which are relatively uncorrelated.

If you're into calculating stuff, I'd suggest you try a macro review of asset class volatility (i.e. US large cap, Intl large cap, Intl munis, US munis, REITs, p2p, etc) and see if you can come up with a portfolio with weights that you'd be comfortable with the total expected return, and that also has a relatively low Beta to the S&P (or Dow/Nasdaq/whatever).

MoNoGro

  • 5 O'Clock Shadow
  • *
  • Posts: 5
  • Location: ABQ, NM
Re: Using Sharpe and Sortino Ratios to determine an Asset Allocation
« Reply #3 on: January 15, 2015, 10:17:49 PM »
Point taken rmendpara regarding the trajectory/volatility of the market going forward compared to the past.  Backtesting has its limitations...

My focus isn't solely on volatility, but from what I've understand volatility can impact the long term growth rate of a portfolio.  The simple stock/bond split recommended by many index investors out there seems to have higher standard deviations which I interpret to mean larger losses in a falling market and the erosion of capital.  My question is, is there a better distribution of asset class(es) that can achieve an equivalent  CAGR to a jhcollins or 90/10 portfolio with potentially less downside risk? It makes sense that some measure of risk, like these ratios, should be part of the analysis.  That said, basing an AA just on avoiding volatility is probably not wise.  What other types of analyses/metrics should I use to chose an AA?   

andy85

  • Handlebar Stache
  • *****
  • Posts: 1060
  • Age: 38
  • Location: Louisville, KY
Re: Using Sharpe and Sortino Ratios to determine an Asset Allocation
« Reply #4 on: January 16, 2015, 09:58:37 AM »
Point taken rmendpara regarding the trajectory/volatility of the market going forward compared to the past.  Backtesting has its limitations...

My focus isn't solely on volatility, but from what I've understand volatility can impact the long term growth rate of a portfolio.  The simple stock/bond split recommended by many index investors out there seems to have higher standard deviations which I interpret to mean larger losses in a falling market and the erosion of capital.  My question is, is there a better distribution of asset class(es) that can achieve an equivalent  CAGR to a jhcollins or 90/10 portfolio with potentially less downside risk? It makes sense that some measure of risk, like these ratios, should be part of the analysis.  That said, basing an AA just on avoiding volatility is probably not wise.  What other types of analyses/metrics should I use to chose an AA?   
with a virtually unlimited number of AA variations, the answer to this is probably a definitely. but while AA #1 may be the least volatile and highest in returns one year, AA #276 may be the best in the 2nd. Nobody can put together the perfect AA every year....and even if they could, the expenses from selling and buying in order to reallocate among the best AA year in and year out would outweigh the benefits of the perfect AA.

So basically, you can never get it perfect. It obviously appears your mix in the last column offers the best return given the risk.