Can You Beat Warren Buffett's Investment Returns Using Data?

|
Includes: BRK.A, BRK.B, IJR, PIMIX, QQQ
by: Logan Kane
Summary

The ideal hedge fund portfolio would use leverage to take a variety of uncorrelated risks that all cancel each other out, leaving the fund with little to no net exposure.

For example, if you could invest in enough items with no correlation to the business cycle, like earthquake bonds and reinsurance, your return increases linearly, but your risk only increases logarithmically.

Investing in alternative sources of return is one reason that Berkshire Hathaway has been able to make so much money over the years.

You can't perfectly cancel out risk, but we'll take a look at combining an unconstrained PIMCO bond fund with a data-driven ETF strategy that uses leverage.

There are risks to the strategy. The question is whether you can manage them properly. If so, you can historically pocket about 20 percent annual returns with a 10 percent standard deviation.

How to invest like Warren Buffett

Warren Buffett and Berkshire Hathaway (NYSE:BRK.A) (NYSE:BRK.B) are famous for earning 20+ percent annual returns for over a 50-year period. However good Buffett may have been as a stock picker, Berkshire Hathaway applies some principles that fly under the radar, but that play just as big of a role in the success of the company over time. Fortunately for us, Berkshire Hathaway isn't a secret. By studying Buffett, we can learn how to build better portfolios for ourselves and get higher returns for less risk.

The question I'd like to ask is:

1. Is it possible to build an investment portfolio built on uncorrelated and negatively correlated risks, and

2. get a 20 percent or higher annual return from it as Buffett has?

I believe that the answer is likely yes. And if you can compound capital at or above 20 percent, you can get fantastically rich with just a little patience, as Mr. Buffett has shown.

How Does Warren Buffett Invest?

Berkshire Hathaway's success comes in large part due to its ability to build a large portfolio of uncorrelated bets. Now, Berkshire's investment portfolio dwarfs its insurance float, but there's a lesson to be learned in how it accumulated capital over time from the business of insurance, especially in its early years. Insurance companies rely on a few mathematical principles like diversification and the law of large numbers to allow them to profitably take risks. Berkshire has been quite good at diversifying risks, as it turned an underwriting profit in 14 out the last 15 years. It then invests the premiums it receives in a variety of securities with the aim of taking more uncorrelated risks (most of Berkshire's insurance premiums are invested in a variety of bonds, as required by regulators, but some of the premiums are allowed to be invested in stocks, as well as more exotic investments). The resulting underwriting profits can then be invested in whatever Buffett chooses. This process has provided much of the seed capital for Berkshire's extensive equity portfolio over time.

Source: Wikipedia Creative Commons

This above chart is what true diversification looks like. On the other hand, traditional 60 stock/40 bond portfolios are not diversified; they're only cushioned, at a fairly high cost to your total compounded returns over time.

To this point, many of the elements of taking uncorrelated risks that have served Berkshire Hathaway so well are probably lacking in your investment portfolio.

Why traditional investment portfolios are flawed (and what to do about it)

The majority of individuals invest in the same way, which is to have concentrated risk positions in equities, particularly large caps. So while the returns are good over time, everyone suffers from the same risk of equity bear markets, which is when the business cycle turns sour and stocks fall between 25 and 50 percent. You can't diversify this risk away by investing in hundreds of different stocks because they're all correlated with each other.

This risk has real consequences. You're most likely to lose your job or incur losses and/or have financing difficulties for your business during a recession. If you could, you'd like to be in the opposite position, sitting on cash and government debt when a crisis happens to buy stocks at low prices. Basic game theory shows that anyone who has cash during a financial crisis can buy assets at fire-sale prices.

Risk allocation of a classic 60 stock/40 bond portfolio

Source: Wikipedia Creative Commons

This is a big reason that pension funds are starting to get into alternative sources of risk that don't have anything to do with the business cycle, like reinsurance. The odds of a major hurricane hitting Florida are pretty much fixed whether the S&P 500 is up or down, unlike corporate bonds, which tend to get hit at the same time as equities. Taking uncorrelated risks like this helps pension funds improve their Sharpe ratios and overall returns and does a better job of matching capital willing to take risk with the needs of the marketplace. There are ways to screw this up, but in general, reducing the take of the middlemen (like big banks and insurance companies) in the economy helps improve the efficiency of society at large.

After finding uncorrelated risks to take, you could take it a step further by finding ways to take risks that cancel each other out to some extent. If you're able to take two opposing but profitable bets on economic growth rising on one hand and falling on the other, then your overall risk is reduced even further, like a sportsbook taking bets on both sides of a football game. This is what PIMCO funds are designed to do. If you look out across asset classes, you'll find that certain assets tend to do well when others do poorly.

This is the principle behind a popular portfolio strategy being used on Wall Street called risk parity. Bridgewater Associates, the world's largest and most successful hedge funds, is a big proponent of risk parity.

The classic example of this that I like to give is that government bonds tend to perform very strongly during equity bear markets.

Most people don't know this, but TLT, which tracks long-term US Treasury bonds, went up 28 percent in 2008.

Long-term bonds accomplish two goals. The first is that they tend to give you a bunch of liquidity when stocks fall, and the second is that investors with mortgages avoid borrowing at the expensive 30-year rate to invest in short-term bonds which pay far less than they could earn by paying down their mortgages. No one will give you a margin call on your mortgage, but long-term bonds (or short-term bonds leveraged to the duration risk of long-term bonds) are incredibly helpful for investors with debt on their balance sheet as they reduce drawdowns and help manage portfolio risk while matching the rate exposure of your liabilities with the rate exposure of your assets.

Chart TLT data by YCharts

The point of risk parity isn't that you should actually try to set your risk exposure equally to each asset class, which is the way Wall Street tries to market it. Instead, use the principles behind the theory. You want to design your portfolio to:

1. Reduce bear market equity risk

2. Invest in uncorrelated and/or negatively correlated assets

The Core-Satellite Portfolio With Moving Average Volatility Targeting

For those of you who read my leveraged/volatility targeting series (link to part 1, link to part 2), some of this will be familiar to you. If you read this article and you're still confused afterward, go to the leverage/volatility targeting series to get up to speed! This portfolio uses volatility targeting to increase returns and reduce risk. I'm not going to tell you volatility targeting is a free lunch, but I can also tell you that no one has been able to disprove it.

For the explanation of why I'm choosing the ETFs that I am (and on return factor strategies in general), see my data-driven ETF series (here's a link to original article, and a link to the current update).

Anyway, the core portfolio is designed around a core portfolio of a well-known PIMCO bond fund, value stocks, and long-term Treasuries. 80 percent of assets are invested in the core portfolio.

The satellite portfolio is either invested in Treasuries or a 50/50 combination of the NASDAQ (QQQ) and small-cap stocks (IJR), depending on whether the risk environment is favorable or not. This is determined by whether the S&P 500 is above or below its 200-day moving average. 20 percent of total assets are invested in the satellite portfolio. The portfolios are rebalanced quarterly between asset classes for purposes of backtesting.

It sounds weird, but there are mountains of research about how volatility can actually be forecasted. And if you use a leveraged strategy that rebalances within relatively short time frames, being able to predict volatility is akin to card counters knowing the count of the deck in blackjack. By applying more risk when the environment is favorable and applying less when the environment is unfavorable, you're able to beat the market.

Here's a link to the full model I'm using.

Here's the baseline performance of the non-leveraged version.

Source: Portfolio Visualizer

As you can see, since 2010, the portfolio has returned 11 percent per year with a standard deviation of 4.9 percent and a max drawdown of only 5 percent. This gives it a Sharpe ratio of 2.08. It's not included in the calculation, but adding revenue from securities lending should improve the Sharpe ratio to over 2.1. You could certainly run a strategy like this with no leverage, but I feel that if you can tolerate the risk of a 60/40 portfolio, then you should be able to reap a 20 percent annual return from leveraging the strategy 2 to 1 (multiply the unleveraged return by 2, minus estimated borrowing cost).

Part of the elegance of strategies like these is that I've optimized them to deliver the maximum return for each unit of risk taken. All you have to do is decide how much risk you're comfortable taking and/or how much return you need. You can run the strategy at 1-1, 1.25-1, 1.5-1, or 2-1 leverage, and get the roughly corresponding risk and return profile of the underlying strategy, scaled to each level of risk tolerance.

I also ran a stress test going back to 2008, and the Sharpe ratio including that time period was 1.44, with roughly a 10 percent annualized return and a 6.5 percent standard deviation.

The portfolio is designed to be comfortably leveraged as far as 2-1 and releveraged at your convenience (asset prices go up, you reset the leverage to 2-1, exponentially increasing the size of your total holdings over time). Leverage is allowed to float as long as the S&P 500 is above its moving average. When the market starts to do poorly and the S&P 500 is below its 200-day moving average at the end of the month, the risk management system kicks in and dials back the risk.

Is the stock market predictable?

Is the bond market predictable?

I've written before on here that I feel that investors are biased towards concentrating their risk in large-cap stocks. I cover mostly large-cap stocks on here because that's where the demand for information is greatest, but if you notice the disclosures, I never have individual stock positions. That's because using leverage on the ETFs of dividend stocks and small-caps is a better way to make money than picking stocks, for the most part. It's easier to manage the risk of using leverage (using techniques like volatility targeting, rebalancing, and asset class diversification) than it is to manage the risk of being 100 percent in equities in a time like 2008. Leverage absolutely terrifies people, but if you look at the numbers the drawdowns aren't nearly as bad as going 100 percent equity with no leverage if you know what you're doing.

PIMCO is one of the companies that takes this philosophy to heart. The idea behind the PIMCO income fund (PIMIX) is to take a variety of uncorrelated risks, some of which cancel each other out. For example, if the economy slows, government bonds tend to do well. If the economy strengthens, credit spreads tend to tighten. If you're able to balance your book with respect to credit and interest rate risk, the idea is that the risk exposures will cancel each other out, leaving you to simply pocket the risk premium. This approach allows PIMIX to yield about 5-6 percent currently. The fund carries one of the highest five and 10-year Sharpe ratios for a publicly traded product that I've ever seen.

The fund is typically about 300 percent long and 200 percent short at any given time. The managers tend to short sell cash instruments and bonds they believe are overvalued, and go long bonds they believe are undervalued. Additionally, to manage interest rate risk, they use interest rate swaps to hedge their interest rate exposure. To manage credit risk, the fund often buys credit default swaps.

If you look at the holdings of the fund you quickly realize that if you decide to invest you're going to have to trust some people that you don't know in Newport Beach to run this multibillion-dollar bond fund. You're also going to be paying an active management fee on top of it. Not everyone reading this is going to be able to stomach this, and that's okay. However, I'm comforted by the theoretical and historical indications that the bond market can be beaten.

The bond market is weird, but it's also predictable in some ways.

Seeking Alpha contributor Ploutos (Ploutos is the Greek god of wealth) has covered some of the more interesting anomalies in the bond market, such as the momentum anomaly in bonds, and the tendency of high-yield bonds to post strong returns in January when issuance seasonally dries up.

Some of the stuff PIMIX invests in includes non-agency MBS, carry trades, and convertible arbitrage trades. Improperly managed risk in MBS and carry trades by banks and broker-dealers using leverage and the repurchase market basically caused the 2008 financial crisis, but PIMCO managed the risk extremely well. PIMIX only suffered a 10 percent drawdown in 2008.

Here's a quick backtest of the strategy to 2008 leveraged 2-1 with no risk management (volatility targeting) applied whatsoever on our end. If I had better software, I'd show you the full breakdown, as I am unable to apply short-selling and leverage to the core-satellite model with the current software I have. I did add a 10 percent allocation position to gold, however. Even with no risk management on our end, the strategy recovered the 2008 losses fully by September 2009. Since then, there hasn't even been a 10 percent drawdown. Note that we were able to calculate the returns of the core satellite strategy by hand above to be roughly 20-21 percent.

Leveraged version (without the risk-management strategy)

Source: Portfolio Visualizer

While this is still a pretty nice graph, comparing the returns of the first and second portfolio shows that applying the volatility targeting and rebalancing cuts the max drawdown by about 40 percent and boosts the returns by 3-4 percent per year on the upside. This is possible because you're continually using your profits as collateral to buy exponentially more stocks and bonds, but shrinking your risk exposure in times of volatility. The strategy is designed to automatically adjust to the target leverage when the volatility environment changes, and to reallocate assets based on moves in prices.

By setting a leverage target, you're adjusting your position size and managing risk better than the other gamblers who double down or weak hands who sell at the bottom. While it's possible to lose money by using a strategy like this, using risk management ensures that your drawdowns rarely are as bad as the portfolios of typical investors. On the return side, the roughly 4 percent difference in annual returns shows that there's a real value add from the volatility management strategy.

If you really want to go down the rabbit hole on the art and science of volatility targeting, try reading these two papers. Volatility targeting works well for equities, but it's absurdly effective for junk bonds.

1. Intertemporal Risk Parity

2. Leverage for the Long Run

Implementation

The target allocation for the portfolio is as follows. This is for ease of backtesting, but you can do better by using futures where possible to take advantage of lower margin requirements and cheaper implied interest rates.

Long:

  • 100 percent of net asset value in PIMIX
  • 40 percent of net asset value in VYM
  • 20 percent of net asset value in TLT
  • 15 percent of NAV in IJR (goes to TLT when S&P is below 200-day moving average at end of the month for a minimum 60 days)
  • 15 percent of NAV in QQQ (also goes to TLT when S&P is below 200-day moving average at end of the month for a minimum 60 days)
  • 10 percent of net asset value in IAU (optional, alternately you can put this portion of the portfolio in IJR/QQQ)
  • Total - 200 percent long

Short:

  • 100 percent cash

Margin loans are always going to be more expensive than the implied repo rate you can get in the futures market. For example, with Interactive Brokers, you might be able to borrow $1,000,000 for under 3.5 percent, but in the futures market, you can borrow the same amount of money off your balance sheet for 2.5-2.6 percent. Also, note that you may be forced to wait 30 days to get your full desired leverage ratio (SEC rules make you wait 30 days to borrow against mutual funds).

You can work around this by buying the equities first, then borrowing against them to buy the mutual funds using portfolio margin. Alternately, use derivatives to get your desired leverage exposure.

For the Nasdaq and TLT positions, you can use futures instead of buying the ETFs to get an equivalent position instead of paying cash. You could also do this for gold, but might want to stick with IAU it if the gold market is too far in contango. Using futures keeps a lot of the leverage off of your personal balance sheet and lowers your transaction costs. Then, when you're adjusting to your target leverage, you can buy and sell QQQ and TLT in small quantities.

Another thing to note is that by using futures you can go long the 5- and 10-year Treasuries with a little leverage rather than invest in the 30-year bonds (statistical evidence for why leveraging Treasuries works so well here). The margin requirements for Treasury futures are tiny, so you don't have to tie up a lot of your liquidity with them (plus you double your carry by borrowing cheaper). Doing this would increase the notional amount of leverage you're using above 2-1 but would actually reduce your risk.

And yes, if you use this strategy, you are using leverage to acquire assets that also are using leverage. This is the case whether you buy stocks (almost all companies have debt) or bonds, or funds that deal in both. That's part of what allows you to get such high returns. As such, risk management is of paramount importance. You can choose between leverage or concentrated risk, but you can't have both.

Conclusion

The biggest risk in this portfolio is that PIMCO is a fluke. I personally don't think that it is. I like this portfolio because the market risk in the portfolio is managed by scaling back positions in volatile markets. Much of Warren Buffett's success comes from good stock picking, but I believe his real talent has been pricing risk well (in both the insurance and stock markets). I don't have a good way for you to access the reinsurance market, but this 2.1+ Sharpe portfolio is a good first step towards redefining risk and reward.

Feel free to add your own opinions, criticisms, and questions in the comments!

Good luck to all!

Disclosure: I am/we are long TLT, VYM, IJR, QQQ. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.