Stocks, Treasuries, and Gold

Everything should be made as simple as possible, but not simpler.

Albert Einstein

How can only three assets plus cash produce such amazing returns?  Well, all three assets complement each other in way other assets just can’t. They work so well together, there really isn’t any reason for anything else. Simplify as much as possible, but not simpler.

Portfolio Goals

Lets start with the portfolio goals:

  • Maximize long term risk adjusted return
  • Minimize trading costs
  • Minimize drawdowns
  • Simplify

Stocks

Stocks have the highest return of any commonly traded asset.  Since we are aiming for the highest return possible, let’s build the foundation of the portfolio from the asset with the highest expected return.  However, individual stocks have a fairly high annual volatility of around 30%, leading to a fairly poor risk return ratio (the Sharpe ratio of a single stock is around .28 since 1940). 

Values are for demonstration purposes only

But combining stocks together into a portfolio that slowly adjusts its composition over time creates probably the greatest single investment in history, the stock market index.  The market index reduces the volatility down to around 16%, while only slightly reducing the overall return, creating an amazing risk vs return profile (Sharpe ratio around .46 since 1940) for a single investment. 

Furthermore, with mutual funds and ETFs today, you can get access to an amazing market index of 500 companies in the USA with extremely low fees and transaction costs at a penny a share.  Market index investing is such an amazing deal, it has confused people for years into questioning why more people don’t do it.  This is the reason huge swaths of people swear by investing in the S&P 500 and only the S&P 500. 

What Type of Asset to Add Now?

Now what do you combine with potentially the greatest single investment ever to improve it?  Since our ultimate goal here is to increase our long term geometric return, we’re looking for an asset that meets two criteria:

  1. Still has a high enough return to warrant mixing with the market index
  2. Is un-correlated – preferably negatively correlated – with the market index

Number 1 is fairly straight forward.   You want an asset which has as high a return as possible.  But number 2 is a bit more nuanced.   Our aim is to create the portfolio with the highest geometric return possible, maximizing this equation:

Arithmetic Average – Volatility2 / 2 = Geometric Average

Number 1 maximizes the arithmetic average.  Number 2 aims to minimize the portfolio volatility.

Negatively correlated assets move opposite to each other. An asset that moves opposite of stocks will likely increase in price when stocks fall, and and fall in price when stocks rise.  While this seems somewhat counterproductive, these assets smooth out the returns over time, increasing the long term geometric average. 

Long Term Treasury Bonds

Values are for demonstration purposes only

Treasury Notes and Bonds are the most liquid, most highly traded asset negatively correlated to market indexes.  Treasuries are not always negatively correlated to stocks, but generally they gain when stocks lose.  And of all government debt, long term treasury bonds have the highest yield and the highest return.1

Furthermore, treasury bonds are one of the most liquid investment in the world.  Trading costs on treasuries are cheap (we will use the TLT ETF), and the market is as deep as any in the world. Long term treasury bonds perfectly match our criteria, therefore it joins the stock index in the portfolio.

Why not Corporate Bonds?

Corporate bonds do usually provide a larger return than treasury bonds. However, they are very often still correlated to the return of stocks. This makes sense – if a company is doing poorly, both its equity (stocks) and its debt (bonds) become riskier, and both fall in price. Therefore, treasury bonds actually provide a far stronger pairing with stocks than corporate bonds do.

What’s Next?

This two asset portfolio is actually quite strong.  There is a reason many advisors recommend a 60% stocks, 40% bonds portfolio.  As long as you re-balance it every now and then, your probably going to get solid returns. 

However, there are still plenty of times when bonds and stocks move in lock step with each other.  During those times, this portfolio will be prone to large drawdowns.  It needs another negatively correlated asset with both stocks and bonds. 

Sadly, I don’t think that asset exists.  I’m not even sure it’s possible for something to be negatively correlated with two items that are negatively correlated with each other.  Therefore, the next best asset doesn’t correlate with either asset at all, and that does exist.

Gold

Values are for demonstration purposes only

Historically, gold shows little correlation to either stocks or bonds.  Luckily for us, it also produced fairly strong returns over the last 50 years.  Certainly not as strong as stocks, but slightly better than treasury bonds.  And as with stocks and bond, gold is extremely liquid, and in its electronic form (ETFs and futures) very cheap to trade.  So gold will be our third asset. 

Cash

The last asset is cash.  Cash can cut a portfolio down to a sensible risk level when all assets start going crazy, and inversely provides leverage to increase our risk and return when the market warrants it.

Values are for demonstration purposes only

Anything else?

Not really.  Sure you can still improve upon this mix with more assets, but we’ve picked the low hanging fruit.  Other assets will likely be highly correlated to some part of our portfolio, with higher volatility and lower return, eliminating most of the benefit of its addition. At some point the benefit of more assets just increases both the trading costs and the complexity of the strategy.

Values are for demonstration purposes only

That said there are a few other assets which may warrant inclusion in the future.

Foreign Stocks

Historically, this would have been a great idea.  Foreign stocks used to be fairly uncorrelated to US stocks.  But they generally are highly correlated today, especially at times of high volatility (crashes).  That said, the largest risk to my 4 asset portfolio is probably it’s USA centric bias.  I fully expect to add foreign stocks to this strategy in the future (probably with the EFA ETF).

Foreign Bonds

I’m less sold on foreign treasury bonds, but they too may end up providing some use.  Diversifying bonds outside of one country certainly sounds like a good idea. However, bonds’ true strength in this strategy comes from their negative correlation in times of stress, and I’m not really sure foreign bonds serve this purpose as well as treasuries.  But it’s something to think about.

Bitcoin

Hear me out.  In some ways, this seams silly.  But on paper, it’s nearly perfect.  Bitcoin does have great returns over its short life.  And it seems totally uncorrelated to gold, bonds, and stocks.  Currencies in general are very uncorrelated to other assets, but their return and volatility are usually so low, there isn’t really a point in owning them.  But bitcoin flips this with high volatility and return.  The one negative, it is expensive to trade. But it definitely needs studying.

Keep it Simple

So there you have it. An extremely robust portfolio with just 4 assets. Extra complexity is just not needed. Simple as possible, but not simpler.

1-When treasuries are negatively correlated to stocks, the yield doesn’t matter as much, and the true benefit to the asset to the portfolio comes with the larger volatility paired with the negative correlation.  A fully negatively correlated 30 year t-bond would nearly eliminate all systematic risk in a standard stock market index because of its higher volatility.

2- Astute readers will notice the strong similarity with the permanent portfolio developed by Harry Browne. This portfolio invests 25% each in Stocks, Bonds, Gold, and Cash, and holds those same percentages overt time. Surprisingly, something that simple actually does very well compared to many of the best investment strategies. My mindsets for choosing these 4 assets is very similar. Together, they are really all you need as they all complement each other. My main difference lies with the proportion of the 4 assets in application, and in the dynamic vs. static nature of our portfolios.

10 Replies on “Stocks, Treasuries, and Gold

  1. You can have a third asset negatively correlated to both Asset 1 and Asset 2. Suppose
    Asset1 = 2 * F1 + F2 + E1
    Asset2 = -2 * F1 + F2 + E2
    And construct
    Asset3 = -F2 + E3.

  2. Great post. I’m confused about the artitmetic return of cash being 0.05? Could you elaborate a bit..?
    Thanks
    E

  3. Why not IEF instead of TLT? Less volatile and also with negative correlation to SPY. The last 7 years of better performance of TLT is directly related to dropping interest rates – a scenario that will not repeat itself in the next decade. From 2002 – 2013 return was almost identical. Or maybe a bond aggregate would be better? The less volatility the better, not? Or is more negative correlation to other asset more important?

    1. With the negative correlation, I want the higher volatility from TLT. It means I need less of it to balance out the portfolio and it acts almost like a levered IEF, without the leverage.

  4. You used gold data which dates back 50 years. That is problematic because of golds surge in the 70’s.
    First, prior to the year 1970 golds price was fixated by the government for several decades which means it didn’t keep up with inflation for that time period. From the year 1971 golds price was again set by the free market, and because of the decades long fixed price there was an “inflationary vacuum” which pumped up the price of gold severely. Also, during the 70’s we had a high inflation due to the oil crisis which also made golds price explode (gold and oil prices are closely related).
    To put it short: the 70’s are not a good proxy for determining the expected return associated with gold and should be left out of the analysis.

    Also, why have you only chosen gold as a commodity, why not commodities as a whole for further diversification? It’s like going for the tech sector for the stock market part instead of the total stock market, as technology has been outperforming any other sector in the last several decades. That’s recency bias. Just as we shouldn’t rely on tech outperforming other sectors for the future and should look to hold the whole stock market, we also shouldn’t rely on gold to outperform other commodities for the future aswell.

    1. Feel free to leave the 70s out. Doing so doesn’t reduce gold’s return enough to warrant leaving it out of the portfolio.

      Commodities, returns are often correlated to stocks in downturns since both of thier demand is related to to the outlook of the economy. This makes them much less useful to the portfolio. Gold, because it isn’t consumed, doesn’t folow this pattern. Gold is half money half commodity, making it much more useful.

  5. Golds CAGR jan 1972 – jul 2020: 8,02 %
    Golds CAGR jan 1980 – jul 2020: 3,21%

    5 year rolling return average 1972 – 2020: 6,45%
    5 year rolling return average 1980 – 2020: 3,04%
    * I used portfoliovisualiser, 1972 is the furthest data that is avaible. If it was 1970, the difference would be even bigger.

    I think those are significant differences in returns.

  6. I am curious about how you would form your expected return of gold as input to your portfolio optimization process. For nominal return, it is straight forward, expected inflation. In real terms, it would be 0% geometric return.

    1. I think that’s a reasonable method for finding the geometric return of gold. You would need to add half the variance to that though to get the arithmetic return. Ultimately though I think that formula is likely short by a population growth rate.

Leave a Reply

Your email address will not be published.