One of the most common and over-used expressions in finance or investing is 'diversification.' To some it means spreading out your capital among different asset classes and involves buying gold, bonds, stocks and the money market. For others it involves buying a mix of international stocks while having a variety of sectors, industries and market capitalizations. Another investor might incorporate all of the above in addition to other methods and techniques.
One of the goals of diversification is to create a portfolio of holdings that trade independently of each another. Gold goes up and down with little relationship to fluctuating bond yields and the price of corn or the share price of Apple (AAPL). If each holding moves independently, then the combination of your holdings should create a 'diversified portfolio' that has a fairly smooth upwards slanting equity curve.
Simple Asset Class Diversification Works
Creating a simple version of the Permanent Portfolio using the ETFs S&P 500 (SPY), iShares Barclays 20+ Year Treasury Bond (TLT), SPDR Gold Shares (GLD) and the iShares Barclays 1-3 Year Treasury Bond Fund (SHY) creates a decent historical equity curve like this (red is the 4 ETF portfolio and the blue is the S&P 500):
Broad diversification between asset classes seems to provide a smooth ride. Less can be said about diversification in just the stock market. Why does sector, industry and country diversification in stocks fail to provide the same downward protection? The S&P 500 is already quite diversified, yet as the chart above shows (blue line), this group of stocks fell hard. Yet look at how spread out the sector weighting is (this diagram represents the equal-weighted S&P 500 index - not the cap-weighted).
Is there a method to lower downside risk to bear markets beyond industry and sector diversification?
Diversification in Bull and Bear Markets
The paper, Are the Gains from International Portfolio Diversification Exaggerated? The Influence of Downside Risk in Bear Markets (Butler, Joaquin, 2001), highlights a noteworthy point: bear markets are highly correlated. So in calm markets you can buy stocks from various countries and these may move independently giving you a nice steady equity curve. A bull market will have the international markets weakly correlated but the bad news is that bear markets are like a contagion that spreads and pulls everything down at the same time. When you need your diversification the most, it fails.
The same is true between sectors and industries. While these might move up and down with less correlation in calm markets, when the economy is tanking and everyone is losing money - and when everyone is afraid and pulling money out of the market uniformly - most stocks drop regardless of the industry and sector.
If you are only using diversification on the buy-side, this is bad news. But can we use this anomaly to signal when to get out of the market and start raising cash when downside risk is high? Can we merely screen for conditions with high price correlation and start reducing our exposure to the market? Indeed we can.
Lack of Diversification is a Sell Signal
A simple shortcut to pick up diversified holdings is to screen for stocks that trade with a low price correlation to each other... and assume that this is because there is some underlying difference that spells diversification. But this only helps us buy diversified stocks, we want to sell when diversification fails us.
If markets become highly correlated during bear markets - much more so than bull markets and even more so than calm markets - could we not monitor our holdings and sell when price performance of each holding is extremely similar to the other holdings in our portfolio? Wouldn't this give us a signal that we are in danger of a highly correlated drop?
Using Price Correlation in My Portfolios
Take for example the Piotroski-inspired model I have made. It is a variation of the F-Score system plus my own take on the system.
Chart 1 (All charts, screening and back-testing is compliments of Portfolio123)
This has a decent annualized return of 28%, but the drawdown of 57.87% is not much better than the broad market (the benchmark is the S&P 600). In this simulation I have 0.25% slippage factored in. Average daily turnover needs to be a minimum of $175,000 at time of purchase - so this isn't a system for big time investors.
Anyway, the point is that this system has a nice upside but a big downside to match. But what happens when I restrict new purchases based on price correlation? I compare the price performance of the aggregate portfolio to the price performance of each holding. When any one holding is too highly correlated to the portfolio - it is sold. So in a highly correlated price drop (a.k.a. bear market), the holdings will be progressively sold and cash will be raised.
Of course, our diversification (maximum price correlation) rules increase the portfolio turnover and slightly raise net return after accounting for slippage, but the big win comes through our reduced drawdown. The chart below shows how the model sells off holdings and raises cash in highly correlated conditions.
This can be added to virtually any strategy or model to lower maximum drawdown and raise cash during times of highly probable market crashes. The chart below is a large cap high-yielding strategy with the first diagram on the left having no price correlation rules and each successive chart having an increasingly restrictive set of price correlation rules set on the buy and sell side.
While there is slightly more total return in the left diagram, many investors would be willing to trade some of that return for a shallower bear market loss such as in the 2nd or 3rd diagram.
So does diversification fail us? Hardly. But diversification is not a passive management tool. Diversification is a great tool for active management. Make sure your stocks have independent price performance when you buy and every day thereafter. When all your stocks start behaving the same, it is time to find new different stocks or begin selling off positions to lower your risk of a highly correlated portfolio price drop.