Summary

  • If you want long-term outperformance, you must be able to stomach short-term underperformance. As William Bernstein said, “The most important investment ability is an emotional discipline.”
  • Investing is a team sport that requires this discipline from both the investment manager (to stick to his investment process) and the end user (to stick with the manager).
  • Setting realistic expectations, grounded in data, can improve investor outcomes by reducing the likelihood of panicked, emotional decisions. These expectations should include a firm understanding of the types of environments that are favorable/unfavorable to the strategy (i.e. when we can expect it to underperform and when we can expect it to outperform).
  • Underperformance that occurs in environments where a strategy is expected to struggle may be good evidence of a manager staying true to his process.

In our opinion, this post from Alpha Architect, titled “Even God Would Get Fired as an Active Investor”, should be required reading for those looking to better understand active investing.  Quoting from the post:

“Our bottom line result is that perfect foresight has great returns, but gut-wrenching drawdowns. In other words, an active manager who was clairvoyant, and knew ahead of time exactly which stocks were going to be long-term winners and long-term losers, would likely get fired many times over if they were managing other people’s money.

Question: If God is omnipotent, could he create a hedge fund that was so good that he could never get fired? No. It turns out even God would most likely get fired as an active investor.”

We covered their post and extended the analysis here.

For us, the clear takeaway is that there is no holy grail to investing.  Asset classes and strategies will always ebb and flow through periods of out and underperformance.  No strategy will beat the market year in and year out.  It just won’t happen.

If we want to outperform standard benchmarks over the long-run, we must hold positions that are different from the benchmark.  Holding different securities creates tracking error. And tracking error inevitably means that short-term underperformance will occur.

Even Warren Buffett, by many measures the best investor in U.S. history, is not immune from this truth.

From March 1980 to October 2016, Berkshire Hathaway A Shares delivered an annualized total return of 20.2%, 9.7% more per year than the Vanguard S&P 500 Index Fund (ticker: VFINX) over the same period.  Berkshire’s risk-adjusted returns are more than 60% better than the Vanguard benchmark (Sharpe ratio of 0.74 vs. 0.46) and the stock’s alpha is an astonishing 1.0% per month.

While there are many traits that likely contribute to his success, the one that we find most awe-inspiring is his discipline.  The graph below plots Buffett’s one-year rolling relative performance vs. VFINX.  Positive (negative) numbers indicate that Berkshire beat (trailed) the index fund over the prior 12 months.

pic1

Data Source: Yahoo! Finance.  Calculations by Newfound Research.  Past performance does not guarantee future results. 

Even the great Warren Buffett has lagged the market one out of every three years.  And in many cases this underperformance was significant.  There have been ten separate episodes where Berkshire underperformed the index fund by more than 10% over a one-year period and many periods of underperformance lasted significantly longer.

PeakTroughRecoveryPeak-to-Trough UnderperformanceTime to Breakeven
October 1985January 1986September 198718.1%1.9 Years
October 1989September 1990February 199325.6%3.3 Years
February 1996November 1996March 199822.1%2.1 Years
June 1998February 2000July 200254.4%4.1 Years
September 2002September 2005February 200829.9%5.4 Years
October 2008April 2012Ongoing32.7%8.1 Years

Data Source: Yahoo! Finance.  Calculations by Newfound Research.  Past performance does not guarantee future results. 

Through these difficult times, Buffett and Berkshire have remained committed to their investment process.

However, a manager committed to his or her investment process is only one ingredient for client success.  Investing is a team sport where managers must be committed to their process and investors must be committed to their managers.  This is precisely why we believe setting appropriate expectations is so critical.  The Buffett case study points to a number of key lessons for expectations management.

  1. Underperformance is not bad. 

    In isolation, underperformance, while it may be frustrating, is not necessarily evidence that a strategy is broken or should be abandoned.  One way we can think of Berkshire is as a portfolio of two different assets.  The first asset is simply the S&P 500.  The second asset is a long/short stock picking strategy driven by Buffett.When the long/short stock picking strategy delivers a positive return, Berkshire will beat the market and vice versa.

    Would most investors abandon the first part of the portfolio (the S&P 500) just because it loses money over a given period of time?  Probably not.  Firing Buffett just because the long/short strategy loses money (i.e. Berkshire underperforms the market) would be no different.

  2. In fact, occasional underperformance is to be expected. 

    For a manager to beat the market, they must be different than the market.  Being different means taking on tracking error.  Buffett could reduce the risk of significantly lagging the market by by allocating less to his “long/short” portfolio.  (Note: In practice, this would mean replacing some of his holdings with an index-tracking strategy).However, doing so would come at the cost of less outperformance (assuming that Buffett isn’t able to further increase his investment skill).  As an example, for Buffett to cut the probability of lagging the market by 10%+ in half, he would have to sacrifice about half of his annualized outperformance.

    Furthermore, risk-free outperformance is impossible over the long-run.  To the extent that such opportunities actually exist, they would most likely be arbitraged away very quickly.  Even low risk outperformance would require unrealistic degrees of investing skill. As an example, assume that Buffett was able to improve the performance of his “long/short” portfolio to such as a degree that the probability of underperformance in any given year was just one in ten (instead of the one in three that we see in the actual data). This would imply such massive long-term outperformance that Berkshire’s market-cap would currently exceed the rest of the S&P 500 combined.

  3. Placing “stop-losses” on managers is counterproductive. 

    While formal stop-losses on active strategies/managers may be rare, many investors operate with them in the form of annual or semi-annual performance reviews, eliminating managers when they underperform by too wide a margin.These stop-losses are counterproductive because they are almost guaranteed to be tripped, even by strategies exhibiting statistically significant alpha.  Going back to the early 1980s, the “long/short” component of Buffett’s portfolio has had an annual return of 9.8% with volatility of 20.5%.  Using this data, we can compute the probability of hitting certain informal stop-loss triggers over various holding periods.

    How to read the table: The highlighted row (10% Loss Trigger) and column (5-Year Holding Period) means that the probability of seeing at least one year of 10%+ underperformance during a 5-year holding period is 60%. If an investor would fire a manager for this degree of underperformance, they would more likely than not fire Buffett within the next five years.

    pic2

    Data Source: Yahoo! Finance.  Calculations by Newfound Research.  Past performance does not guarantee future results.

    We see that the probability of hitting certain pain points is quite high, especially when the tolerance for relative underperformance is low and holding periods are long.  And remember, this data is calibrated to reflect perhaps the best equity investor in history.  A manager with similar tracking error to Buffett and a smaller, but still impressive, expected annual outperformance of 3% has roughly an 80% chance of realizing 10%+ underperformance at least once over a five-year period.

  4. Tracking error can be managed in the portfolio construction process. 

    Of course, none of this is meant to imply that investors have to just accept tracking error to the market.  Different investors will have different tolerances for tracking error.  In all likelihood, these tolerances will also change over time.  For example, when the outlook for the U.S. equity market is bleak, an investor may be more willing to bear tracking error vs. the S&P 500 as they search for ways to enhance returns.These preferences are best addressed in the portfolio construction process where asset classes and strategies are blended together.  In fact, we explicitly account for the pain of tracking error to popular benchmarks when we construct our own strategic portfolios.

  5. Understanding when a strategy may excel and when it may struggle should be a core part of the diligence process. 

    We believe that by better understanding the potential magnitude and duration of underperformance, as well as the types of markets in which it is most likely to occur, investors can start to exhibit Buffett-like discipline.To illustrate the initial steps of this type of analysis, let’s turn to an example that is near and dear to our hearts: momentum-based tactical equity.

    We built a simple momentum strategy with a methodology similar to that used in AQR’s article “Back in the Hunt.”  Using data from Fama and French, we measure the trailing 1-year return for large-cap U.S. equities.  The allocation to equities ranges from 0% to 200% (i.e. no exposure to a 2x levered position) depending on how the most recent 1-year return compares to past 1-year returns.  When the current return is above the historical median, the strategy will have more than 1x exposure to the equity market.  When the current return is below the historical median, the strategy will have less than 1x exposure to the market.  At each point in time, we only use data that would have been available to investors at that time (i.e. the median is computed on a rolling bias).  This methodology is used to avoid hindsight bias.

    Note: We use a 0% to 200% range for equity exposure as this allows us to both under and overweight equities.  This mirrors the recommendation that clients use our tactical equity strategies – which can shift equity exposure between 0% and 100% - as a pivot between their stock and bond allocations.  This pivot usage also allows for both the under and overweighting of equities. 

    This simple, hypothetical strategy beats the market by 86bps per year with a 20% reduction in the duration and magnitude of drawdowns.  (All returns are backtested and hypothetical and this hypothetical strategy was created and performance was calculated in connection with the writing of this analysis.  Index returns include the reinvestment of dividends and are gross of all fees and expenses.  Past performance does not guarantee future results.)­

    Like with Buffett, we see that the long-term, risk-adjusted outperformance of the strategy does not eliminate the possibility of significant short-term underperformance.

    pic3

    Data Source: Yahoo! Finance, Fama/French data website.  Calculations by Newfound Research.  Past performance does not guarantee future results.  The momentum timing strategy is hypothetical and backtested.  It reflects the reinvestment of dividends and does not include the payment of any fees or expenses.  The index is not representative of any Newfound strategy or index.

     

    The recent underperformance of the momentum strategy (+1.8% YTD through 9/30 compared to +7.5% for the market) is well within historical bounds.

    Digging into the data a little deeper, we see that the tactical momentum strategy tends to outperform when the market is down and when the market is up big (30% or more).  In more mild up markets, the strategy tends to underperform.

    pic4

    Data Source: Yahoo! Finance, Fama/French data website.  Calculations by Newfound Research.  Past performance does not guarantee future results.  The momentum timing strategy is hypothetical and backtested.  It reflects the reinvestment of dividends and does not include the payment of any fees or expenses.  The index is not representative of any Newfound strategy or index.

     

    We can also use the data to set expectations around the risk-mitigating properties of momentum-driven tactical equity.  We see below that the hypothetical strategy generally does a solid job protecting against large losses.  In most cases, downside protection tends to kick in when the market is down between -10% and -20%.

    That being said, we also see that momentum is not a perfect risk management tool.  In 1987, the strategy actually would have been down more than the market.  This makes sense since momentum is by definition backward, not forward, looking, and the strategy could be up to 2x levered.  As a result, it is not equipped to deal with unexpected rapid sell-offs like what was seen in 1987.  This is exactly why we advocate for a holistic approach to risk management, incorporating complementary tools like high quality fixed income, tactical asset allocation, and alternative strategies (e.g. managed futures).

    pic5

    Data Source: Yahoo! Finance, Fama/French data website.  Calculations by Newfound Research.  Past performance does not guarantee future results.  The momentum timing strategy is hypothetical and backtested.  It reflects the reinvestment of dividends and does not include the payment of any fees or expenses.  The index is not representative of any Newfound strategy or index.

     

Conclusion

No investor, not even the great Warren Buffett, is immune from bouts of short-term underperformance.  Short-term underperformance is a necessary, if annoying, reality for investors seeking to outperform the market on a risk-adjusted basis.

Rather than engage in the fruitless pursuit of constant outperformance, investors should seek to fully understand how their asset classes, strategies, and managers may perform in different market regimes.  This analysis can help users distinguish between benign underperformance and underperformance that may be a symptom of a serious problem (i.e. a manager deviating from his stated investment process).

Underperformance that occurs in environments where a strategy is expected to struggle may be good evidence of a manager staying true to his process.

Client Talking Points

  • Investing is a team sport that requires discipline from both the investment manager (to stick to his investment process) and the user (to stick with the manager).
  • To do better than the market from either a risk or return perspective, you must be different than the market.
  • Being different, while creating the opportunity for long-term outperformance, virtually guarantees occasional short-term underperformance.
  • Setting realistic expectations, grounded in data, can improve investor outcomes by reducing the likelihood of panicked, emotional decisions.