Posted on

If MINVAR is Crap, What Does “Work”?

Let me admit my “mistake” on this. I am using the Lazy Portfolio. Gold, US treasury bills or their equivalent, longer term US Government bonds and the S&P 500.

The mistake is cherry picking a portfolio. Why gold and not silver? Why gold and not a broad commodity index? Why the US stock market? Why not the MSCI World? Why US treasuries and bonds? Why not a basket of world debt?  Should I be including Euro and Swiss short term Govs where the interest rates are currently negative? Should I be looking at constant maturity schemes for longer term bonds rather than conventional approaches?  If its “All Weather” you are looking for then you need to answer these questions and a few more. Don’t cherry pick.  Its a nonsense and in the long term could be extremely damaging.

Think Trump. Think Kim Il Wotsit. Think any number of other nice people who could blow up their countries or economies in the twinkling of an eye. Do you really want to cherry pick? Or should you go as global and as wide as you can. I know my preference. Only you can choose yours.

For my own investment I would choose a basket of commodities, the MSCI World, and world debt. Or at least major economy debt. Would I include Greece, Italy, Portugal and Turkey in my debt basket? Should I? A question for another day perhaps.

Anyway I chose the lazy portfolio because yesterday I trashed the cherished MINVAR approach (using that portfolio) by pointing out it amounts to trend following and can have you “dangerously” exposed to a portfolio consisting at times of 100% stocks. Probably not very bright.

Is there an alternative that appeals to me? Ignoring for the moment my dislike of the Lazy Portfolio.

A useful benchmark is the UCRP – uniform constant rebalanced portfolio. Or equal weighting with constant rebalancing to those of us who don’t go for poncy names.  Instead of constant rebalancing let us assume monthly rebalancing.

Here are the results:

Stat                 UCRP        S&P 500
-------------------  ----------  -----------
Start                1988-01-04  1988-01-04
End                  2016-12-20  2016-12-20
Risk-free rate       0.00%       0.00%

Total Return         665.76%     1593.83%
Daily Sharpe         1.17        0.64
CAGR                 7.28%       10.26%
Max Drawdown         -19.84%     -55.25%
Monthly Vol (ann.)   6.19%       14.27%
 Actually not at all bad, especially in risk adjusted terms. A difficult benchmark to beat and, other than the choice of portfolio itself, pleasantly forecast and parameter free. And of course no 100% exposure to stocks.
Can un-leveraged risk parity improve on that? “Risk Parity” may have been lionised by AQR and Bridgewater but neither Asness nor Dalio invented the concept. CTAs have used it for years to equalise bets in volatility terms between instruments in portfolios containing everything from the three month Eurodollar to metals and stock indices.
Here I present a simple monthly rebalancing with weighting on an inverse volatility basis. There is probably some fancy name for it. Do people call it naive risk parity? Perhaps, but who cares.
Our Cliff and dear old Ray add leverage to increase the returns and indeed this is very easy to do, very effectively, using the futures market.
Either way, it looks more attractive than many other options and preferable to the UCRP. Lower volatility, lower drawdown (potentially!) for a reasonable return. Looks to compare very favourably to pure stocks and Minvar.
Stat                 Inverse Volatility    Benchmark
-------------------  --------------------  -----------
Start                1989-01-04            1989-01-04
End                  2016-12-20            2016-12-20
Risk-free rate       0.00%                 0.00%

Total Return         552.48%               1395.44%
Daily Sharpe         1.56                  0.63
CAGR                 6.94%                 10.16%
Max Drawdown         -12.47%               -55.25%
Monthly Vol (ann.)   4.76%                 14.37%
And here is the area chart for the portfolio over time:
Compared to the highly unsatisfactory and doubtless unstable concentration produced by MINVAR:
Am I making a recommendation here? No I am most certainly not. My partner and I offer software only and not investment advice of any kind. People are merely invited to explore different portfolios, weightings and rebalancing periods for themselves and to draw their own conclusions.
What I will say is that after many years of looking for Eldorado with my friend Dr Pangloss I am convinced most people are completely wasting their time and money in the financial markets. If you want to “invest” do it forcast free and as parameter free as possible.
If you want to “trade” only do so if you have a structural bias in your favour. For Bobby Axelrod and his mates this means inside information. For the HFT brigade it means the bid offered spread and front running. If you can’t think of a structural edge, don’t trade.

4 thoughts on “If MINVAR is Crap, What Does “Work”?

  1. Anthony,

    It is really a joy and very inofrmational to read your blog. Many thanks for writing it1

    Please do not misunderstand the quesiton I have — it in no way meant to imply any kind of controversy or to be confrontational. I am just genuinly intersted in the subject and I am looking forward to your opinion.

    So here it goes.

    You authored a book on ETF investment and in it (judging by the foreword) you advocate timing when it comes to asset allocation. But your recent posts suggest, at least to me, that you may have changed your point of view since then and now prefer avoiding any prediction of returns and, I presume, of timing them,

    Please correct me if I am misinterpreting your writings or if I am deeply miunderstranding something. Especialy the latter is entirely possible, since I am a true layman when it comes to trading. But in any case I am as always very much looking forward to your comment. I think it is a very important subject, both for professional and those of us
    just dabbling in algo trading at Quantopian ….


    TIm Vidmar

    1. Tim
      Your question is entirely fair and reasonable and you are correct about market timing – I have changed my mind since 2008 as a result of direct trading experience up to 2011 in futures whn my account was sabotaged by Jon Corzine. And also as a result of increased awareness of the ease of curve fitting which in many cases is almost unconscious.

      I have also witnessed the disastrous performance of CTA trend followers since 2008. Like the trend followers, most of my hitherto well performing systems took a turn for the worse in recent years. For instance “Tuning Up the Turtle” went way off course after I published that article.

      I DO believe in trend following but not in the concentrated form I used in the past. I certainly do NOT believe in the Dual Momentum type trend following – or rather, while it has worked in back testing, I can not see that this binary approach is a safe or sensible way to approach markets.

      I have tended to take a more long term approach and considered consistent positive “drift” rather than trend pe se. So, long term stock markets have trended up for hundreds of years with economic growth. Bond markets have trended (“drifted”) upwards with coupons regardless of interest rates.

      Commodities have not, on the whole shown consistent drift. After inflation gold has not done much over the past few decades. Perhaps better methods of extraction of minerals and metals and better agricultural methods have kept commodity prices down.

      In summary, I “believe” trend following on stocks and bonds “works” but I prefer these days a risk parity and long only approach using only such markets (stocks and bonds) which have shown a long term uptrend. I prefer to stick with those markets and account for the disparity in volatility periodically rather than dipping in and out with “signals”.

      So far as traditional Trend Following is concerned, I quite like AQRs approach and similar. Take the top 300 out of 1,000 stocks each month. Don’t concentrate one’s holdings. Or, going back to my 2008 book, I still think the approach was right but I would add a similar number of high quality bond funds of varying duration to smooth the results.

      I also like to look for situations where I can “earn” rather than speculate. Picking up premium and hedging the exposure as for instance with the VIX and long term treasuries.

  2. Anthony,

    Please ignore my pevious comment! I have just come ccross your post on “Why I loathe market timing”, which makes it abundantely clear whar your position is.

    I should have the “due diligence”. My apologies.



    1. Ah, sorry! I should also have checked my own older articles.

Leave a Reply