When testing a model it is always difficult to determine if the results you are achieving are repeatable or not.  If you are testing a high relative strength model, for example, are the results coming from one or two stocks that make the whole test look fantastic?  If that is the case I would have my doubts about how that strategy would perform in real-time.  But if the results are truly from an underlying factor performance (regardless of the individual securities in the portfolio) then you have something you can work with.

The way we determine if a model is lucky or not is to run multiple simulations based on a random draw of securities.  In a relative strength model we might break our universe into ten different buckets.  Out of the highest bucket we might draw 50 stocks at random.  We hold those stocks until they are no longer classified as high relative strength securities.  Once they fall below a specific rank we sell the security and buy another one at random.  If we run 100 trials we get 100 different portfolios over time.  What we are trying to determine is if the individual securities in the test really matter, or is just the concept of buying high relative strength securities over time what causes the outperformance.

As it turns out, what stocks go in to the portfolio aren’t as important as exploiting the factor.  A disciplined approach is that consistently drives the portfolio to strength is what drives the returns over time.

(Click To Enlarge)

The table shows the results from one of the factors tested in the paper.  You can see the range of outcomes each year as well as how each model did over the 16 year test period.  Sometimes the models outperform, sometimes the underperform, and some years you have mixed results.  But over 16 years, all of the models outperformed!  All we did was pick stocks at random out of a high relative strength basket.  There is nothing complicated about it.  The main thing is that the process is systematic and extremely disciplined.

More details about the testing process and results can be found in the paper (click here).

Momentum works. Seminal research by Narasimhan Jegadeesh and Sheridan Titman in 1993 first identified momentum as a systematic source of risk for equity investors. Their research—corroborated by numerous subsequent academic studies—revealed that, historically, momentum investing had provided excess stock returns over a market index.

…..

Overall, momentum returns outperformed market returns by an average of 3.13 percentage points on an annualized basis.

You can see the results on a country-by-country basis below.

Source: Forbes/Gerstein Fisher

Leaving aside the ridiculous statement about momentum being first identified in 1993, their study is important because it shows that momentum returns are universal.  Based on their study and others, we agree.  Dorsey Wright has been running a Systematic RS International portfolio since 2006 and the results have been excellent.  Our net returns have exceeded the EAFE benchmark by a similar amount to what Gerstein Fisher found–not surprising given the substantial overlap in time frames.

To receive the brochure for our Separately Managed Accounts, click here.  

Click here and here for disclosures.  Past performance is no guarantee of future returns.

Numerous academic studies have confirmed that, when measured in periods of approximately three to 12 months, past investment winners tend to keep on outperforming while past losers tend to keep underperforming.

Momentum is not simply a US phenomenon. A recent study² covering equities in 23 countries from November 1989 to September 2010 found evidence of strong momentum returns in North America, Europe and Asia Pacific; only Japan was an exception. Another study tracking the largest 100 stocks in the British market from 1900 to 2009 found that a portfolio comprised of the 20 best performers over the prior 12 months outperformed the worst performers by 10.3% annually³.  The same authors found momentum in 18 out of 19 markets, dating back to 1975 in larger European markets and 1926 in the US.

Momentum is not confined to portfolios of individual stocks – it exists in a variety of asset classes. A recent study⁴ has found that momentum exists in government bonds, commodities and currencies as well as country equity indexes. Momentum has also been found in corporate bonds⁵ as well as the financial futures market⁶.

The article is well-footnoted.  I recommend you read the original, which I linked to above.  The article does a good job discussing both the pros and cons of relative strength.  For example, the author points out that:

…there are prolonged periods where stocks with positive momentum underperform the market.  Despite an overall annualized premium of 3.9%, there have 22 periods where stocks with positive momentum have underperformed the market by greater than 5%, with durations as long as several years.

Although investors have a marked tendency to abandon strategies when they underperform for a period of time, that might not be a good idea with relative strength.  Despite periods of underperformance, long-term results have been remarkable:

The $1.00 investment in momentum stocks grew to $67,309, nearly 30-times larger than the $2,321 earned in the S&P 500. [August 1927 to July 2011]  For long-term investors, this outperformance has been remarkably enduring. In 99.6% of the 10-year rolling periods since July 1937, momentum stocks have outperformed the S&P 500. [my emphasis]

Investors have a lot of choices when it comes to selecting an investment strategy, but not many have been as well validated over as long a period of time in multiple markets as relative strength.

It’s not you, it’s me….  I think everyone has used that line at some point, but nobody does it better than George Costanza!

I have been putting data together to update our white papers.  It’s no secret that running a Global Tactical Asset Allocation (TAA) strategy was difficult last year.  But when I looked at the data it was very clear that the problem wasn’t the strategies.  The real problem was how the market behaved during 2011.  It’s not you, it’s me.  It’s not your trend following strategy, it’s what you’re trying to follow.  The market was essentially a psycho, stage 5 clinger last year!

The data I will reference in this post is an extension of the data we published last year in two white papers.  If you haven’t read them you can find them here.  Our research process for this dataset takes a diverse universe of ETF’s and creates 100 different equity curves for a number of different momentum factors.  The universe has a number of different asset classes represented including Equities (Domestic & Foreign), Bonds, Commodities, Currencies, and Real Estate.  The results provide a good idea about how a momentum-based, global TAA strategy would have performed.  By creating 100 different equity curves we are taking luck out of the equation and showing a realistic range of outcomes from buying high relative strength securities out of our universe.

Most of the momentum factors we follow underperformed last year.  The factors we are showing refer to the lookback period to do our rankings.  The 1MORET factor (1-month return) means we used 1 month of data to calculate our momentum ranks (all securities are held until they fall out of the top of the ranks, which might be as short as one week or as long as a couple of years).  The 12MORET factor uses the prior 12 months of price data to rank the securities.  The 3-month factor actually performed the best in 2011, but only 40 out of the 100 trials outperformed the S&P 500, so you needed some luck to outperform.  The 6-month factor was the next best, but only 1 trial outperformed so you needed to be really lucky.  All the other trials were very poor.  There was so much short-term volatility back and forth last year that the very short 1-month formulation period was deadly.  It paid not to be too quick on the trigger last year!

Full Year 2011

(Click To Enlarge)

But looking at 2011 in aggregate doesn’t really tell the whole story.  The beginning of the year was good for these strategies.  That person you were dating held it together pretty well for the first couple of dates!  Through the end of April, most of the strategies were outperforming the S&P 500 on average.  The 6-month factor was doing great as all 100 trials were outperforming.  Ironically, the factor doing the worst was the 3-month factor.

2011 Through April

(Click To Enlarge)

The problems for trend following strategies began in May.  There were a series of sharp trend reversals in a number of different assets: Bonds, Stocks, Precious Metals, Currencies (Yen & Swiss Franc).  No matter what factor you were using from May to the end of the year it was difficult.  It was tough to get traction anywhere.  The only factor that did even so-so was the 3-month factor, and that was the worst factor through April.  That’s just one of many examples of how crazy the 2011 market was!

2nd Half 2011

(Click To Enlarge)

So where do we go from here?  Well, the, “It’s not you, it’s me…” line always leads to a breakup.  That’s probably not a bad idea when dealing with something that doesn’t change.  Does that psycho, stage 5 clinger ever get any better?  Nope.  It only gets worse.

But markets change, and TAA based on momentum is very adaptive.  We will not be in a choppy, range bound environment forever.  Trends will emerge. (If they don’t, it will be the first time in history.)

Investors were euphoric about momentum-based TAA strategies in the first part of the year.  Looking at the data you can see why – they were working exceptionally well.  After the last few months, people are certainly not as excited.  In reality, now is the time to be really excited about relative strength strategies, not back in April.  Now is the time you want to be adding money.

DAL, which manages $1.3 billion, has been using data like this to make its investment decisions. It calls this strategy “upgrading,” an approach that it has been advocating in its newsletter since the 1970s.

The strategy looks at one-, three-, six- and 12-month fund returns. The belief, which began with Burt Berry, DAL’s founder, in the 1970s, was that market trends could not be forecast and were clear only in retrospect. But the trends lasted long enough so that you could capitalize on it. Call it the hot hand of investing.

DAL applied its strategy to the funds in this study, starting with the top 15 in 1989 and tracking their one-, three-, six- and 12-month returns. When a fund dropped out of the top third — below 99 — it sold the fund and reinvested the money in the top-ranked fund that it did not own.

DAL calls it “upgrading;” the academic literature typically calls it “casting out.”  Whatever you call it, our Systematic RS portfolios follow exactly the same process.  Sell it when the rank drops and replace it with the best-ranked item you don’t already own.  DAL used a composite measure of relative strength—as we’ve mentioned before, many methods will work as long as the portfolio gets exposed to the strong performers.

The article actually had a little broader mandate.  DAL looked at a range of funds for the New York Times to try to determine what worked and what didn’t.  Here’s what they found:

The best-performing funds over time were not necessarily the ones with the lowest fees, run by the best-known managers or focused on any particular strategy, according to more than 20 years of data examined by DAL Investments, an investment adviser and publisher of the NoLoad FundX newsletter in San Francisco. DAL analyzed the returns on 306 mutual funds for The New York Times.

Janet M. Brown, president of DAL Investments, said the deep dive was motivated as much by trying to figure out what worked as by testing the effectiveness of the firm’s own unconventional strategy. (More about that later.)

“The overall challenge of mutual fund investing is selecting funds in advance that people think will do well in the future,” Ms. Brown said. “The easiest thing would be to buy and hold or to select a manager with a good long-term track record and buy it and forget it. That was not an effective way of selecting funds.”

I added the bold.  It’s important because most investors select funds using exactly the process that DAL found ineffective!  (See Andy’s note on this same problem!)

What DAL found is that the funds that beat the benchmark changed over time.  As Ms. Brown said, “In my view, it has less to do with the brilliance of the portfolio manager as when their styles are in sync with the market,” she said.  This makes perfect sense.  Not one of the funds outperformed all the time.  In fact, the average fund lagged the benchmark one-third of the time.  Some funds lagged more years than they outperformed, but still had market-beating returns.  Other funds outperformed two-thirds of the time, but still fell behind the benchmark.

Because the styles that work keep changing, good portfolio management adapts.  That is one of the essential traits of relative strength: it does not discriminate and adapts to whatever is working.

Professional pundits are not usually paid to make correct forecasts. They are paid to sound convincing, whether they are columnists or figureheads for asset managers.

I agree that sounding convincing is essential in order to win and keep business.  However, I also believe that intellectual honesty is an important part of the long-term success of a business.  Forecasting is not just challenging…it’s impossible.  We have chosen to do our homework, gain a deep understanding of relative strength, and make every effort to convincingly make the case to investors for using relative strength as part of their asset allocation.  It’s also a nice benefit that the data is on our side.

HT: Abnormal Returns

Please read the white paper to see the details of the Monte Carlo testing process used for this study.  This simulation was done on a universe of ETFs from a variety of asset classes over an 11-year period of time (2000-2010).  However, the cumulative mean results are as follows:

What conclusion can be drawn?  The best results over the 11-year period of time came from using a 6-month look-back relative strength factor.  For the purposes of the white paper, percentile ranks were used to determine when to get out of a a given trade.  Getting out of a position every time its relative strength has a bad month or even bad three months of price performance did not lead to as good of performance over time as using an intermediate relative strength ranking stop.

This underscores the importance of understanding how to thoroughly test a set of decision rules so that you can be best prepared to run those decision rules in real time and with real money!  It is human nature to want to bail out of a position after it looks like the trend may be turning.  In hindsight, we may well see that indeed we would have been better to get out much earlier.  Of course, we may also see that the relative strength stabilized and then moved higher.

The goal with any trend following system is to capitalize on long-term trends.  Relative strength is ideally suited to help you accomplish that goal.  However, it won’t come without some discomfort along the way.  Today, the question may be about gold, but tomorrow the same question will be asked about a different asset class.  Relative strength provides a clear process for determining when to get in and when to get out.

O’Shaughnessy details the results of relative strength from 1926-2009:

A $10,000 investment on December 31, 1926, in the decile of stocks from All Stocks with the best six-month price appreciation is worth $572,831,563 at the end of 2009, a compound return of 14.11 percent a year. This return dwarfed an investment in the All Stock universe, which turned $10,000 into $38,542,780 over the same period, an average annual compound return of 10.46 percent.

It is also important to point out that O’Shaughnessy found that this relative strength portfolio outperformed the benchmark in 68% of single-year returns, 79% of rolling 3-year returns, 87% of rolling 5-year returns, 95% of rolling 7-year returns, and 98% of rolling 10-year returns.

Also, keep in mind that this is just a generic relative strength strategy based on a 6-month return factor with annual rebalances.  His book also showed that using a 12-month relative strength factor also outperformed the benchmark with an compound return of 12.34 percent.

In case you were wondering, a strategy based on buying stocks with the worst 6-month returns and then holding for a year had an annualized return of 4.15 percent!  As stated in the book, “If you’re looking for a great way to underperform the market, look no further [than buying relative strength laggards].”

O’Shaughnessy on the symmetry for the relative strength deciles:

The decile returns for All Stocks by six-month price appreciation reveal a perfect staircase, with the performance of decile 1–containing the best six-month price performing stocks–at the top and returns of the other deciles descending in step to the tenth decile, which contains the worst six-month price performers.  The top six deciles all beat All Stocks, with the bottom four all underperforming the universe.

Source: What Works On Wall Street

The results are in and they are highly favorable for relative strength investing.

Has 14 years of out-of-sample performance of these two mutual funds confirmed the motivating backtests?

HFCVX [Hennessy Cornerstone Value] underperforms both its benchmark Russell 1000 Value Index and the S&P 500 Index. The fund underperforms the S&P 500 Index by about 0.5% per year, compared to the backtested average annual outperformance of about 7%. Also, its standard deviation of annual returns (20.1%) is higher than that for the benchmark Russell 1000 Value Index (18.7%). Backtested outperformance has not persisted over a 14-year out-of-sample implementation.

HFCGX [Hennessy Cornerstone Growth] outperforms both its benchmark Russell 2000 Index and the S&P 500 Index. The fund outperforms the S&P 500 Index by about 2.5% per year, compared to the backtested average annual outperformance of about 10%. Its standard deviation of annual returns (21.2%) is about the same as that for the benchmark Russell 2000 Index (21.1%). Backtested outperformance has persisted at a subdued level over a 14-year out-of-sample implementation.

Relative Strength still works on Wall Street

Source: CXO Advisory

In other words, the dividend strategy has not been able to beat the market over the last 14 years, while the relative strength strategy has outperformed in real life.  This mirrors CXO’s findings earlier.  I might note that the outperformance of the Cornerstone Growth strategy comes despite the Q3-Q4 2008 – Q1-Q2 2009 performance of relative strength, which was a big historical outlier.  The underperformance of relative strength was epic during that brief period—and Cornerstone Growth outperformed anyway.  I would further note that the 2.5% annual outperformance is after fees.

Evidence suggests that relative strength is a strategy worth implementing.

This index has dropped to the middle of the distribution over the last couple of weeks.  The 10-day moving average of this indicator is 62% and the one-day reading is 58%.