Feeds:
Posts
Comments

Archive for the ‘Strategy’ Category

I received a number of emails asking me to revisit the backtests from last week’s post about using the Shiller PE to time the market (Worried about a Crash? Backtests Using Shiller PE to Time The Market (1926 to 2014)). The most common request was to separate the buy and sell rules such that if the strategy sold out at say one standard deviation above the mean, it didn’t buy back until the Shiller PE fell below its mean. The second most common request was to alter the strategy such that it hedged out the market rather than switching to cash.

The backtests use Fama and French data of the book value-to-market equity (the inverse of the PB ratio) data from 1926 to 2013. As at December 2013, there were 3,175 firms in the sample. The value decile contained the 459 stocks with the highest earnings yield, and the glamour decile contained the 404 stocks with the lowest earnings yield. The average size of the glamour stocks is $7.48 billion and the value stocks $2.54 billion. (Note that the average is heavily skewed up by the biggest companies. For context, the 3,175th company has a market capitalization today of $404 million, which is smaller than the average, but still investable for most investors). Portfolios are formed on June 30 and rebalanced annually.

The following backtests use the market’s state of knowledge at the time about the average, and standard deviations of the Shiller PE. The chart below shows how the Shiller PE’s average and standard deviations have varied over time.

Shiller PE, Average, and Plus/Minus Two Standard Deviations (1881 to Present)

Shliler PE mean and SDs

The mean now is 16.55, but was as low as 14.2 in the 1950s and, excluding the slightly higher reading at the start of the data, as high as 17.5 in the 1900s. The standard deviation has expanded over time. Until the 2000s two standard deviations above the average meant a Shiller PE of about 25, and now it means a Shiller PE of almost 30.

Performance of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (1926 to Present)

The following chart backtests three strategies. The first–“Sell at 1SD, Buy at -1SD”–buys the price-to-book value decile only if the Shiller PE is one standard deviation below its mean, sells into cash if the Shiller PE is more than one standard deviation above its mean, and holds cash until the market falls back below one standard deviation below the mean. The second–“Sell at 1SD, Buy at Mean”–buys the price-to-book value decile only if the Shiller PE is below its mean, sells into cash if the Shiller PE is more than one standard deviation above its mean, and holds cash until the market falls back below the mean. The third–“Sell at 2SD, Buy at Mean”–buys the value decile only if the Shiller PE is below its mean, sells into cash if the Shiller PE is more than two standard deviations above its mean, and holds cash until the market falls back below the mean.

Shiller Moving Average and Value Performance 1926 to 2014

All the strategies underperform the simple buy-and-hold strategy over the full period.

The market returned 13.94 percent compound and the fully invested PB value decile returned 20 percent compound over the full period. Sell at 1SD, Buy at -1SD returned 15.0 percent compound; Sell at 2SD, Buy at Mean returned 19.3 percent compound; and Sell at 1SD, Buy at Mean returned 15.9 percent compound.

Sell at 2SD, Buy at Mean had good lead until the 1990s, but has woefully underperformed since. Notably, it is still fully invested. Its sell rule won’t kick in until the Shiller PE hits 29.7.

Drawdowns of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (1926 to Present)

Shiller Moving Average and Value Drawdown Relative 1926 to 2014

The market has the worst maximum drawdown at 86 percent, and the fully invested PB value decile has a comparably bad maximum drawdown of 85 percent. Sell at 1SD, Buy at -1SD had the best maximum drawdown at 50 percent; Sell at 2SD, Buy at Mean had a maximum drawdown of 78 percent; and Sell at 1SD, Buy at Mean had a maximum drawdown of 60 percent.

Graham Rule: Performance of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (1926 to Present)

Benjamin Graham recommended maintaining a minimum portfolio exposure to stocks of 25 percent. Below we re-run the tests, but this time instead of kicking all of the portfolio into cash, we put only 75 of the portfolio in cash, and maintain 25 percent exposure to the value decile.

Shiller Moving Average and Value Performance Graham Rule 1926 to 2014

All the returns are improved, but the strategies continue to underperform the simple buy-and-hold strategy over the full period.

Sell at 1SD, Buy at -1SD now returns 16.8 percent compound; Sell at 2SD, Buy at Mean returned 19.6 percent compound versus 19.3 percent above; and Sell at 1SD, Buy at Mean returned 17.2 percent compound.

Graham Rule: Drawdowns of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (1926 to Present)

Shiller Moving Average and Value Drawdown Relative Graham Rule 1926 to 2014

The tradeoff for slightly improved returns is slightly worse drawdowns. Sell at 1SD, Buy at -1SD still has the lowest maximum drawdown, but now draws down 53 percent; Sell at 2SD, Buy at Mean has the same maximum drawdown of 78 percent; and Sell at 1SD, Buy at Mean had a maximum drawdown of 64 percent.

Market Hedge: Performance of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (1926 to Present)

In this set of backtests the strategy is levered and hedged. The ratios change depending on the level of the market. When the strategy deems the market cheap, it is 130 percent long the value decile, and 30 short the market. When the market is expensive, it doesn’t sell into cash, but reduces the long to 100 percent, and hedges out 75 percent of the portfolio using the market.

Shiller Moving Average and Value Performance Hedged 1926 to 2014

The Hedge at 2SD, Lever at Mean strategy outperforms the buy-and-hold strategy over the full period, returning 21.9 percent compound, versus 20 percent for the value decile.

Hedge at 1SD, Lever at -1SD now returns 19 percent compound; and Hedge at 1SD, Buy at Mean returned 18.9 percent compound.

Market Hedge: Drawdowns of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (1926 to Present)

Shiller Moving Average and Value Drawdown Relative Hedged 1926 to 2014
The tradeoff for the improved returns is bigger drawdowns.

Hedge at 2SD, Lever at Mean strategy draws down 88 percent, worse than the market’s 85 percent; Hedge at 1SD, Lever at -1SD has a maximum draw down of 67 percent; and Hedge at 1SD, Buy at Mean has a maximum drawdown of 79 percent.

Changing the buy-and-sell rules, and hedging rather than running to cash alters the performance of the strategies. The levered and hedged strategy that maximizes exposure to the market–Hedge at 2SD, Lever at Mean–outperforms the simple buy-and-hold strategy, but does so with an enormous drawdown. Nothing else beats buy-and-hold. As we saw last week, the more conservative the Shiller PE ratio used, the lower the drawdown, but returns suffer. To generate the extraordinary returns of the value deciles I’ve examined over the last few weeks, it was necessary to remain fully invested in those value stocks through thick and thin. My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more. Click here if you’d like to read more on Quantitative Value, or connect with me on TwitterLinkedIn or Facebook.


About these ads

Read Full Post »

Charlie and I would much rather earn a lumpy 15% over time than a smooth 12%.

–Buffett, Chairman’s Letter (1996)

Can an investor concerned about a big crash use a systematic timing tool to exit the market before the crash without giving up too much return? One possible method for doing so is to use the Shiller PE as a valuation tool, and to move the portfolio into cash at some given level of overvaluation. The backtests below show the returns and drawdowns for exiting at four different levels of the Shiller PE ratio, from aggressive to conservative.

The first option is to simply always remain fully invested in the value decile (measured by price-to-book value). For our present purposes, this is the most aggressive. The three other timed strategies kick out of the value decile when the Shiller PE gets increasingly expensive. Mean, the most conservative kicks into cash when the Shiller PE gets just above its mean (17.6 for the data set). Slightly more aggressive is a strategy that kicks into cash at one standard deviation above the mean (a Shiller PE of 24.8) called 1 Std. Dev., and the next most aggressive kicks out at two standard deviations above the long-run average (a Shiller PE of 32.0) called 2 Std. Dev..

Edit: The backtests use Fama and French backtests of the book value-to-market equity (the inverse of the PB ratio) data from 1926 to 2013. As at December 2013, there were 3,175 firms in the sample. The value decile contained the 459 stocks with the highest earnings yield, and the glamour decile contained the 404 stocks with the lowest earnings yield. The average size of the glamour stocks is $7.48 billion and the value stocks $2.54 billion. (Note that the average is heavily skewed up by the biggest companies. For context, the 3,175th company has a market capitalization today of $404 million, which is smaller than the average, but still investable for most investors). Portfolios are formed on June 30 and rebalanced annually.

Performance of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies

Shiller and Value Performance 1926 to 2014

Over the period examined, the market (the equally weighted universe from which the portfolios were drawn) generated a compound average growth rate (CAGR) of 13.94 percent. Cash generated an average return of 5.17 percent over the same period. The fully invested value decile generated the best CAGR over the full period at 20.01 percent. The other strategies underperformed to the extent that they remained out of the market: The strategy that kicked into cash at the mean returned 13.4 percent yearly, the strategy that kicked into cash at one standard deviation above the mean returned 18.15 percent yearly, and the strategy that kicked into cash at two standard deviations above the mean returned 19.36 percent compound over the full period.

We expect underperformance for remaining out of the market. This is the tradeoff we make to avoid drawdowns. Is it worth it? Below we examine how much drawdown we avoid by getting out of the market at the different ratios.

Drawdowns to Value Decile (Price-to-Book Value), and 3 Shiller PE Timed Strategies

Shiller and Value Drawdown 1926 to 2014

The market had a maximum drawdown in 1929 of 86 percent, and has a Sharpe ratio over the full period of of 0.13. The fully invested strategy had a maximum drawdown of 85 percent, and generated a Sharpe ratio of 0.15. The other strategies generate lower maximum drawdowns, but do so for lower Sharpe ratios: The strategy that kicked into cash at the mean had a maximum drawdown of 69 percent, and the worst Sharpe ratio at 0.11;  the strategy that kicked into cash at one standard deviation above the mean had a maximum drawdown of 80 percent, and a Sharpe ratio of 0.14, and the strategy that kicked into cash at two standard deviations above the mean had a maximum drawdown of 84 percent, and a Sharpe ratio of 0.15.

Drawdowns Relative to the Market for Value Decile (Price-to-Book Value), and 3 Shiller PE Timed Strategies

Shiller and Value Drawdown Relative 1926 to 2014

This chart examines the drawdown to each strategy relative to the market. Where it is below the midline, the strategy has drawn down further than the market, and above the midline it is outperforming.

Benjamin Graham recommended maintaining a minimum portfolio exposure to stocks of 25 percent. Below we re-run the tests, but this time instead of kicking all of the portfolio into cash, we put only 75 of the portfolio in cash, and maintain 25 percent exposure to the value decile.

Performance of Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies (Graham Rule)

Shiller and Value Performance Graham Rule 1926 to 2014

The additional exposure to the market improves the returns for the three timed strategies. The strategy that kicked into cash at the mean now returns 15.23 percent yearly, the strategy that kicked into cash at one standard deviation above the mean returned 18.67 percent yearly, and the strategy that kicked into cash at two standard deviations above the mean returned 19.55 percent compound over the full period. All still underperform the fully invested strategy at 20.01 percent.

Drawdowns to Value Decile (Price-to-Book Value), Cash and 3 Shiller PE Timed Strategies

Shiller and Value Drawdown Graham Rule 1926 to 2014

The tradeoff for slightly improved returns is greater drawdowns and volatility. The strategy that kicked into cash at the mean had lower a maximum drawdown of 73 percent, but an improved Sharpe ratio at 0.12;  the strategy that kicked into cash at one standard deviation above the mean remained unchanged with a maximum drawdown of 80 percent, and a Sharpe ratio of 0.14, and the strategy that kicked into cash at two standard deviations above the mean had a maximum drawdown of 85 percent, and a Sharpe ratio of 0.15.

Drawdowns Relative to the Market for Value Decile (Price-to-Book Value), and 3 Shiller PE Timed Strategies

Shiller and Value Drawdown Relative Graham Rule 1926 to 2014

The Shiller PE is not a particularly useful timing mechanism. This is because valuation is not good at timing the market (really, nothing works–timing the market is a fool’s or genius’s game). Carrying cash does serve to reduce drawdowns. It also reduces returns. The more conservative the Shiller PE ratio used, the lower the drawdown, but returns suffer, and Sharpe ratios reduce. To generate the extraordinary returns of the value deciles I’ve examined over the last few weeks, it was necessary to remain fully invested in those value stocks through thick and thin. My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more. Click here if you’d like to read more on Quantitative Value, or connect with me on TwitterLinkedIn or Facebook.

 

 

Read Full Post »

The 10th Annual New York Value Investing Congress is just weeks away (Sept. 8-9), ​but there’s still time to ​get your seat and enjoy the many benefits of attending the Congress in-person:

  • Speaker ​Wisdom – Congress speakers explain their stock ideas with detailed, well-reasoned analysis.  But the presentations are not live-streamed or distributed afterwards,so only onsite attendees benefit from the ​brilliance of these ​successful money managers.
  • Audience Q&A – Congress attendees are savvy, professional investors and their insightful questions often clarify and expand on a speaker’s investment thesis.  As with speaker commentary, live Q&A is only available to onsite attendees.
  • Exclusive Information – The Congress is closed to the media and live bloggers– our attendees benefit first from actionable information.
  • Great Networking –  The VIC Cocktail Reception has become a favorite among attendees for sharing new ideas, creating new business opportunities and making friendships that ​can last a lifetime.

Register now and benefit from attending the Congress in-person.  But seats are strictly limited to 275, so we encourage Greenbackd readers to register now, before we sell out.

For a special Greenbackd Discount, register now with Offer Code: GREENBACKD  at Valueinvestingcongress.com/congress/register-now-partners/

Confirmed Speakers include:

  • Leon Cooperman, Omega Advisors
  • Sahm AdrangiKerrisdale Capital Management
  • Carson BlockMuddy Waters Research
  • Andrew LeftCitron Research
  • Alexander RoepersAtlantic Investment Management
  • Jeffrey SmithStarboard Value
  • Amitabh SinghiSurefin Investments
  • Guy SpierAquamarine Fund
  • David HurwitzSC Fundamental
  • Michael KaoAkanthos Capital
  • Guy GottfriedRational Capital Management
  • Adam CrockerMetropolitan Capital Advisors
  • Whitney TilsonKase Capital
  • John LewisOsmium Partners
  • Tim EriksenEriksen Capital Management
  • Cliff RemilyNorthwest Priority Capital
  • With more to come!

Read Full Post »

I’ve been digging back through some very old historical data to understand how unusual the 1929 crash was and found the site Measuring Worth, which has the interesting mission “to make available to the public the highest quality and most reliable historical data on important economic aggregates.” Measuring Worth has a data series combining the Dow Jones Industrial Average and its precursor, the Dow Jones Average, all the way back to May 1885. Measuring Worth provides this fascinating history of the index:

On July 3, 1884, Charles Henry Dow began publishing his Dow Jones Average. By the time it was published daily eight months later, the index was composed of 12 stocks, 10 of which were railroads. This index appeared in the Customer’s Afternoon Letter up until July 8, 1889 when the first issue of The Wall Street Journal was published. On October 7, 1896, Dow started publishing two “Daily Moving Averages,” 12 industrials and 20 railroads (that would later become the transportation index.) This first Dow Jones Industrial Average (DJIA) was published through September 29, 1916. This first DJIA closed at 71.42 on July 30, 1914 and so did the New York Stock Market for the next four months. Some historians believe the reason for this was worry that markets would plunge because of panic over the onset of the World War. An interesting book by William L. Silber titled When Washington Shut down Wall Street: The Great Financial Crisis of 1914 and the Origins of America’s Monetary Supremacy (2007) has a different explanation. He thinks that Secretary of the Treasury, William McAdoo closed the exchange because he wanted to conserve the US gold stock in order to launch the Federal Reserve System later that year with enough gold to keep the US on the gold standard. Whatever the reason, the first day it reopened on December 12, 1914, the index closed at 74.56, thus the War had not had the predicted impact. On October 4, 1916, the WSJ starts publishing a (new) DJIA of 20 stocks, 8 stocks from the old index and 12 new stocks. It was traced the index back to December 12, 1914 at that time. It is important to know that data for the first DJIA of 12 stocks and the second DJIA of 20 stocks are BOTH available for the 21 months and the first index was about 36% higher than the second and the data here are adjusted to make them a consistent time series. On October 1, 1928 the DJIA of 30 stocks was first quoted in the WSJ. It contained 14 stocks from the second list of 20 and 16 new stocks. Its level was consistent with the second list, so no adjustment was necessary.

In the chart below I’ve plotted the daily closing price for the series from May 1885 to Friday’s close (the blue line, log, right-hand side), along with the drawdowns (the red line, left-hand side).

DJIA Closing Price and Drawdowns 1885 to 2014

Source: Samuel H. Williamson, ‘Daily Closing Value of the Dow Jones Average, 1885 to Present,’ MeasuringWorth, 2012.

For me there are two striking features on the chart. First, the 1929 crash is on a different scale to any other crash, dwarfing even the second biggest crash, which was the most recent one beginning in 2007. The 1929 crash began on September 4, 1929, and bottomed on July 8, 1932, down 89.2 percent. The market wouldn’t achieve its 1929 high again until November 23, 1954, more than 25 years from its last peak, and more than 22 years from the bottom. By way of contrast, the second deepest drawdown at -53.8 percent, the 2007 Credit Crisis began October 1, 2007, and ended March 4, 2013, a mere 5 years and 5 months later. The second longest drawdown began November 1905, bottomed down -48.5 percent, and ended almost 10 years later in July 1915.

The second striking feature of the chart is how much time the market spends in a drawdown. The market hits a new high about 3.8 percent of days, which means it’s in drawdown 96.2 percent of the time. In addition to the 1905 and 1929 crashes, there have been two other busts that lasted longer than 8 years, including one that ran from May 1890 and ended almost 9 years later in January 1899, and another that began January 1973, and ended November 2, 1982. And there were six more that lasted longer than four years. It makes me think we really should break out the Dow X,000 hats every time the market crosses a round number that’s a new all-time high.

It’s not all bad news. Since 1885, the Dow Jones is up 552 times, which equates to a compound annual growth rate of about 5 percent (this excludes the impact of dividends, which have been material, to the tune of around 4 percent more on average, and 1.9 percent today). From the 1932 bottom to the next all-time high in November 1954, the market returned more than double that rate at 10.6 percent. Drawdowns are the usual condition of investing, but, over the very long term, the market has continued to grow. To generate the extraordinary returns of the value deciles I’ve examined over the last few weeks, it was necessary to remain fully invested in those value stocks through thick and thin. My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more. Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

I’ve been reading John F. Wasik’s Keynes’s Way to Wealth: Timeless Investment Lessons from The Great Economist, a book about British economist John Maynard Keynes’s life as an investor. Keynes started out as a foreign currency and commodity speculator in 1919, was wiped out twice in 1922, and 1929, and went on to become a Buffett-style concentrated value investor by around 1932. In 1991, Buffett, who noted in an earlier Chairman’s letter that Keynes had “began as a market-timer (leaning on business and credit-cycle theory) and converted, after much thought, to value investing,” described Keynes’s end-point as an investor thus:

John Maynard Keynes, whose brilliance as a practicing investor matched his brilliance in thought, wrote a letter to a business associate, F. C. Scott, on August 15, 1934 that says it all: “As time goes on, I get more and more convinced that the right method in investment is to put fairly large sums into enterprises which one thinks one knows something about and in the management of which one thoroughly believes. It is a mistake to think that one limits one’s risk by spreading too much between enterprises about which one knows little and has no reason for special confidence. . . .

One’s knowledge and experience are definitely limited and there are seldom more than two or three enterprises at any given time in which I personally feel myself entitled to put full confidence.”

Wasik reports that Keynes died in 1946 with a net worth of about £500,000, equivalent to about $36 million today, not including his extensive collection of artwork and rare manuscripts.

What is most remarkable about Keynes is that he was a self-taught Buffett-style concentrated value investor before Buffett. He wrote the letter to Scott around the time that Benjamin Graham published Security Analysis, but there is little evidence that he read it. Like Graham, Keynes had to contend with perhaps the most difficult period in modern history to be an investor, one that encompassed the second World War, the 1929 stock market crash, and the Great Depression. Throughout that 18-year period, while the UK stock market fell 15 percent, and the US market fell 21 percent, the discretionary portfolio he managed on behalf of King’s College, Cambridge grew fivefold.

The charts below, which begin in 1926 and continue through to April 2014, show how unusually difficult the period was. The charts use Fama and French backtests of the book value-to-market equity (the inverse of the PB ratio) data from 1926 to 2014. As at April 2014, there were 3,175 firms in the sample. The value decile contained the 459 stocks with the highest earnings yield, and the glamour decile contained the 404 stocks with the lowest earnings yield. The average size of the glamour stocks is $7.48 billion and the value stocks $2.54 billion. (Note that the average is heavily skewed up by the biggest companies. For context, the 3,175th company has a market capitalization today of approximately $400 million, which is smaller than the average, but still investable for most investors). Portfolios are formed on June 30 and rebalanced annually.

Compound Returns 1926 to April 2014 (Market Capitalization Weight)

PB VW Returns 1926 to 2013

The 1938 crash also hit Keynes hard. At the end of the year, his King’s College discretionary portfolio was down 40.1 percent.

10-Year Rolling Returns (Market Capitalization Weight)


PB VW 10-yr Rolling Returns 1926 to 2013

Drawdowns (Market Capitalization Weight)

PB VW Drawdowns 1926 to 2013

Compound Returns 1926 to April 2014 (Equal Weight)

PB EW Returns 1926 to 2013
10-Year Rolling Returns (Equal Weight)

PB EW 10-yr Rolling Returns 1926 to 2013

Drawdowns (Equal Weight)

PB EW Drawdowns 1926 to 2013

As these charts demonstrate, volatility and drawdowns are part and parcel of investing, but, over the long term, value has comprehensively beaten out the market and glamour stocks. To generate the extraordinary returns of the value deciles, it was necessary to remain fully invested in those value stocks through thick and thin. My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more.

Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

As we’ve seen over the last few weeks, value has comprehensively outperformed glamour stocks since 1951. Value also outperforms the market. Whether we examine the simple fundamental metrics like the price-to-earnings (PE) ratio, the price-to-book value (PB) ratio, and the price-to-cashflow (PCF) ratio, the dividend yield or a compound model of all excluding the dividend yield, all outperform.

Set out below are the results of Fama and French backtests of the equal weight portfolios of each metric’s value decile using data from June 1951 to December 2013 and the market. As at December 2013, there were 2,406 firms in the sample. Stocks with negative earnings, cashflow, or book value were excluded. Portfolios are formed on June 30 and rebalanced annually.

The “market” here is an equal weight, and total return average of all deciles, and includes stocks with negative earnings, cashflow, or book value.

Value Deciles and Market Compound Returns (June 1951 to December 2013)

Combo EW Mthly Returns 1951 to 2013

Value Deciles and Market Drawdowns (June 1951 to December 2013)

Most of this outperformance is to the upside. When the market turns down, value tends to drawdown along with the market. The following chart shows the drawdowns to each of the value deciles and the market.

Combo EW Drawdowns 1951 to 2013

The chart really shows the enormity of the 2007 to 2009 credit crisis. Outside the 1929 crash, which I’ll examine at a later date, every other drawdown pales beside the credit crisis.

Value Decile Drawdowns Relative to the Market (Portfolio Constituents Equally Weighted) June 1951 to December 2013

This chart shows the performance in a drawdown of each of the value deciles relative to the market. Market drawdowns greater than 10 percent are indicated by the grey bars on the chart. Where the lines for each value metric are below the 0 percent midline, the value metric has drawn down further than the market, and where the lines are above the 0 percent midline, value has outperformed in the drawdown.

Combo EW Relative Drawdowns 1951 to 2013

We can see in the chart that, when the market is in a drawdown, the value deciles tend to drawdown faster, and further, and then recover earlier. We can observe this in the  late 1990s-from 1998 until 2000–when value struggled, and then from the early 2000s when value outperformed. We can also see it in the last big drawdown from 2007 to late 2008 when value fell further than the market, and then recovered faster from early 2009.

In a drawdown, value is below the line about twice as often as it is above the line.

Value has comprehensively outperformed the market over a very long period of time. Most of this outperformance seems to the upside. Despite the received wisdom that value has protected portfolios in a drawdown, it seems that value has fallen along with the market. When it does enter a drawdown, value stocks seem to fall first, and at a faster rate than the market, but they also tend to recover earlier, and faster.

Volatility and drawdowns are part and parcel of investing. To generate the extraordinary returns of the value deciles, it was necessary to remain fully invested in those value stocks through thick and thin. My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more.

Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

The combo value metric is the final frontier in value investment. The dark side of the moon. The manned mission to Mars. Like off-label Spam, most combination models are curious, indelicate compounds of parts unknown. Ovine? Bovine? Porcine? We just don’t know.

To illustrate the benefits of combination, in this post I shine a light on a compound model of known parts–the simple fundamental metrics the subject of backtests on this site over the last four weeks: the price-to-earnings (PE) ratio, the price-to-book value (PB) ratio, the price-to-cashflow (PCF) ratio, and the dividend yield.

Set out below are the results of Fama and French backtests of the equal weight portfolios of each metric’s value decile using data from 1951 to 2013. As at December 2013, there were 2,406 firms in the sample. The value decile contained the 283 stocks with the lowest ratio of price to earnings, cashflow, book value or dividends. Stocks with negative earnings, cashflow, book value or dividends were excluded. Portfolios are formed on June 30 and rebalanced annually.

Value Deciles Annual and Compound Returns (Portfolio Constituents Equally Weighted)

Value EW Returns 1952 to 2013

The PE, PB and PCF ratios deliver comparably excellent returns over the full period examined. The dividend yield meanders around not doing much, and few outstanding years are located therein. Studies have shown that the dividend yield is most most useful in conjunction with other measures of shareholder yield including net buybacks and net debt reduction. Including all of shareholder yield’s components leads to a return comparable to the return of PE, PB or PCF. Absent those measures, the dividend yield is an incomplete metric, and underperforms. For that reason, I have excluded it from the combination.

Composite and Component Value Deciles Annual and Compound Returns (Portfolio Constituents Equally Weighted)

Combo EW Returns 1952 to 2013

The composite, which selects portfolios by equally weighting the PE, PB and PCF ratios, delivers a performance over the full period that beats out PE and PB, and slightly underperforms PCF on a compound basis.The composite ratio generates an average annual return that beats out PCF, and PE, but slightly underperforms PB. This is important to note because it demonstrates the advantage of a composite measure over any single measure–even the best single measure periodically underperformed the composite.

Rolling 10-year CAGRs for Composite and Component Value Deciles

This chart shows the rolling 10-year CAGRs to each of the measures. All are highly correlated, but they diverge materially from time-to-time.


Combo EW Rolling 10yr CAGR 1961 to 2013

The following chart takes the rolling 10-year data from this chart and compares it to the performance of the composite.

Rolling 10-year CAGRs for Components Relative to Composite

Combo EW Relative Rolling 10yr CAGR 1961 to 2013

This chart best illustrates the advantage of using a composite. While the PCF ratio delivered the better return over the full period, it underperformed the combo in 58 percent of rolling 10-year periods. PB was a notable laggard at the beginning of the data, and slightly underperformed over the full period, but the composite was the better bet only 36 percent of the time. PE outperformed strongly at the beginning of the data, but didn’t show thereafter, and so the combo was the better metric 75 percent of rolling ten-year periods.

The attraction of the combo is that it delivers returns comparable to the very best metric at any point in time. While PCF delivered better returns over the full period, most of the outperformance occurred at the beginning of the data, and it lagged thereafter. In contrast, PB underperformed at the start, and then outperformed thereafter. While the combo didn’t ever have its day in the sun as leader, it was the most reliable, never lagging far from the front-runner. Given that we can’t know which metric will be the best at any time, it makes sense to employ all of them if it doesn’t hurt returns too much.

As we’ve seen over the last few weeks, over the long run, and with some regularity, cheap stocks tend to outperform more expensive stocks. The PB, PE and PCF ratios are all very useful metrics for sorting cheap stocks from expensive stocks, but we can’t know which will be the better bet at any given point in time. The combo spreads the risk of underperformance relative to any single metric, and, in doing so, generates reliable investment performance over the full period without lagging far behind the front-runner at point.

My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more.

Hat tip to Jim O’Shaughnessy whose What Works on Wall Street first discussed the combination metric, and his AAII article “What Works”: Key New Findings on Stock Selection for the idea for the relative performance chart.

Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

The dividend yield is a popular value metric for investors for two reasons. First, it’s the obvious metric for investors favoring income over capital gains. Second, unlike earnings or cashflow, dividends are actually paid out to shareholders, and therefore independently verifiable. Where other metrics like price-to-book value, earnings or cashflow rely on management providing a true accounting of a company’s performance, the dividend is tangible proof of excess free cash flow. Thus, goes the argument, the dividend yield therefore provides the most reliable picture of a company’s business performance, and prospects, which in turn leads to better investment performance.

Set out below are the results of two Fama and French backtests of the dividend yield data from 1926 to 2013. As at December 2013, there were 3,393 firms in the sample. The value decile contained the 198 stocks with the highest earnings yield, and the glamour decile contained the 137 stocks with the lowest earnings yield (the deciles are smaller than 1/10th of the stocks in the sample because fully 1,894 stocks pay no dividend at all). The average size of the glamour stocks is $8.60 billion and the value stocks $3.06 billion. Portfolios are formed on June 30 and rebalanced annually.

Dividend Yield Annual and Compound Returns (Portfolio Constituents Weighted by Market Capitalization)

In this backtest, the two portfolios are weighted by market capitalization, which means that bigger firms contribute more to the performance of the portfolio, and smaller firms contribute less. Here we can see that the value decile has outperformed the glamour decile, returning 10.3 percent compound (13.4 percent in the average year) over the full period versus 8.3 percent for the glamour decile (11.3 percent in the average year).

Dividend Yield VW 1926 to 2013

These returns are considerably lower than the returns found for the price-to-earnings and cashflow ratios over the last few weeks (see Investing Using Price-to-Book Value Ratio or Book Equity-to-Market Equity Multiple (Backtests 1926 to 2013)Investing Using the Price-to-Earnings Ratio and Earnings Yield (Backtests 1951 to 2013) and Investing Using Price-to-Cashflow Ratio and Cashflow Yield (Backtests 1951 to 2013)). The reason is that the earnings and cashflow backtests ran back to only 1951, and the dividend yield data, like the book value return data last week, begins in 1926. The difference is partly due to the 1929 crash, which had an oversized impact on returns. The crash is visible on the chart, and striking–it took almost twenty years for the value decile to fully recover.

To make a comparison possible of dividend yield’s performance to the performance of book, earnings and cashflow over the same period, I also measured the returns beginning in 1951. Since 1951 the high dividend yield value decile has generated a compound annual growth rate (CAGR) of 11.4 percent and an average annual return (AAR) of 13.6 percent. Over the same period the glamour decile returned a CAGR of 9.6 percent and an AAR of 12.9 percent. These returns are still considerably lower than the returns generated by PB, PCF, and PE over the same period.

Dividend Yield Value Premium (Market Capitalization Weight)

The value premium is the outperformance of the value decile over the glamour decile. This chart shows the yearly returns to each of the value and glamour deciles, the value premium (value-glamour) in each year, and the rolling average from the start of the data in 1926:

Dividend Yield VW Value Premium 1926 to 2013

The rolling average tells a sad story for value relative to glamour: The value premium has gradually disappeared over time. Over the 73 years of data to 2000 it was actually zero, but it has slightly recovered since to be 1.8 percent compound over the full period.

Decile Performance (Market Capitalization Weight)

The following chart shows the returns to each of the deciles sorted by dividend yield (Updated to include non-dividend payers).

Dividend Yield and No Div VW Decile CAGR 1926 to 2013

This chart shows that the dividend yield is a fair, but not great, metric for sorting stocks into value and glamour portfolios. This is due to the fact that less than half of all stocks pay dividends (only 44 percent pay dividends). A better comparison might be the dividend payers to the 1,894 stocks in the non-dividend paying cohort. The non-dividend payers (No Div) underperformed all the dividend payers except for the glamour decile, generating a compound annual growth rate (CAGR) of 8.4 percent and average annual return (AAR) of 13.2 percent over the full period (and, since 1951, a CAGR of 9.0 percent and an AAR of 13.5 percent).

Recent Performance (Market Capitalization Weight)

As we’ve seen over the last few weeks, value’s outperformance over glamour is not a historical artifact. If we examine just the period since 1999, we find that value has been the better bet.

Div Yield VW Returns 1999 to 2013

Though it started out almost 40 percent behind in 1999, value outperformed glamour over the period since 1999, beating it by 5.2 percent compound, and 6.5 percent in the average year–about the same differential for PB last week.

Market capitalization-weighted returns are useful for demonstrating that the outperformance of value over glamour is not due to the value portfolios containing smaller stocks. Unless you’re running an index (or hugging an index), they’re not really meaningful. The easiest portfolio weighting scheme is to simply equally weight each position. (If we’re prepared to put up with a little extra volatility for a little extra return, we can also Kelly weight our best ideas). Here are the equal weight return statistics for dividend yield.

Dividend Yield Annual and Compound Returns (Portfolio Constituents Equally Weighted)

Dividend Yield EW 1926 to 2013

In the equal weight backtest value generated 12.7 percent compound (16.1 percent on average), beating out glamour’s 11.6 percent compound return (15.5 percent on average).

Since 1951 the equally weighted high dividend yield value decile has generated a compound annual growth rate (CAGR) of 13.5 percent and an average annual return (AAR) of 15.7 percent. Over the same period the glamour decile returned a CAGR of 12.5 percent and an AAR of 15.5 percent. These returns are still slightly lower than the returns generated by PB, PCF and PE over the same period.

Dividend Yield Value Premium (Equal Weight)

Dividend Yield EW Value Premium 1926 to 2013

Again, the value premium was never very large for the equal weight portfolios, and has gradually diminished to 1.1 percent compound over the full period.

Decile Performance (Equal Weight)

Dividend Yield and No Div EW Decile CAGR 1926 to 2013

In the equally weighted portfolios, dividend yield does an even poorer job sorting glamour and value portfolios. The dividend payers don’t even reliably outperform the non-dividend paying cohort. The No Div decile, which returned a CAGR of 13.4 percent and an AAR of 21.2 percent over the full period (and, since 1951, a CAGR of 12.4 percent and an AAR or 18.3 percent), beat out the return on the value decile.

Recent Performance (Equal Weight)

Div Yield EW Returns 1999 to 2013

In the equal weight portfolios, value has slightly outperformed glamour since 1999, beating it by a 3.9 percent compound, and 2.8 percent in the average year.

As we’ve seen over the last few weeks, over the long run, and with some regularity, cheap stocks tend to outperform more expensive stocks. Unlike the PB, PE and PCF ratios, which are all very useful metrics for sorting cheap stocks from expensive stocks, the dividend yield is less useful. This is likely because only around 44 percent of all stocks pay dividends. The message seems to be that even expensive dividend paying stocks outperform all non-dividend payers in the market capitalization weighted portfolios, but not in the equal weighted portfolios. Dividend yield doesn’t seem to be a particularly reliable metric for sorting value and glamour.

My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more.

Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

The price-to-book value ratio (PB) is the granddaddy of the value metrics. Book value is preferred by many value investors to cashflow and earnings metrics because it is stable year-to-year where cashflow and earnings are variable. This is an important property for the following reason: Where a business at a cyclical trough with diminished cashflow and earnings might look expensive on the basis of price-to-cashflow or price-to-earnings, that same business may appear cheap on the basis of price-to-book value because book value won’t fall much or at all in a downturn, and vice versa. Thus, goes the argument, price-to-book value gives a more reliable picture of a company’s usual business performance, which in turn leads to better investment performance.

Set out below are the results of two Fama and French backtests of the book value-to-market equity (the inverse of the PB ratio) data from 1926 to 2013. As at December 2013, there were 3,175 firms in the sample. The value decile contained the 459 stocks with the highest earnings yield, and the glamour decile contained the 404 stocks with the lowest earnings yield. The average size of the glamour stocks is $7.48 billion and the value stocks $2.54 billion. (Note that the average is heavily skewed up by the biggest companies. For context, the 3,175th company has a market capitalization today of $404 million, which is smaller than the average, but still investable for most investors). Portfolios are formed on June 30 and rebalanced annually.

Annual and Compound Returns (Portfolio Constituents Weighted by Market Capitalization)

In this backtest, the two portfolios are weighted by market capitalization, which means that bigger firms contribute more to the performance of the portfolio, and smaller firms contribute less. Here we can see that the value decile has comprehensively outperformed the glamour decile, returning 12.6 percent compound (17.7 percent in the average year) over the full period versus 8.6 percent for the glamour decile (10.9 percent in the average year).

Book Value VW 1926 to 2013

These returns are considerably lower than the returns found for the price-to-earnings and cashflow ratios over the last few weeks (see Investing Using the Price-to-Earnings Ratio and Earnings Yield (Backtests 1951 to 2013) and Investing Using Price-to-Cashflow Ratio and Cashflow Yield (Backtests 1951 to 2013)). The reason is that the earnings and cashflow backtests ran back to only 1951, and the book value return data begins in 1926. The difference is due to the 1929 crash, which had an oversized impact on returns. The impact of the crash is visible on the chart, and striking–it took twenty years for the value decile to fully recover. Something similar has happened to the glamour decile since 2000–it hasn’t grown in 13 years.

To make a comparison possible of book value’s performance to the performance of earnings and cashflow over the same period, I also measured the returns beginning in 1951. Since 1951 the low PB value decile has generated a compound annual growth rate (CAGR) of 15.0 percent and an average annual return (AAR) of 17.9 percent. Over the same period the glamour decile returned a CAGR of 9.6 percent and an AAR of 12.6 percent. These returns are approximately the same as the returns generated by PCF and PE over the same period.

BE/ME (Market Capitalization Weight)

The reason for value’s outperformance is simply due to the fact that the value portfolios bought more book value per dollar invested: 4.3x versus 0.25x for the glamour portfolio. (I used a rolling average. The “average” I’ve quoted is for the full period. The rolling average has been higher, but it’s rarely been lower.) This chart shows the extraordinary bargains available for twenty years following the 1929 crash:

Book Value to Market VW 1926 to 2013

Recent Performance (Market Capitalization Weight)

As we’ve seen over the last few weeks, value’s outperformance over glamour is not a historical artifact. If we examine just the period since 1999, we find that the return is higher than the long term average to 1926, and value has continued to be the better bet.

VW PB Returns 1999 to 2013

Though it started out almost 30 percent behind in 1999, value outperformed glamour over the period since 1999, beating it by 5.2 percent compound, and 6.4 percent in the average year.

As I noted last week, market capitalization-weighted returns are useful for demonstrating that the outperformance of value over glamour is not due to the value portfolios containing smaller stocks. Unless you’re running an index (or hugging an index), they’re not really meaningful. The easiest portfolio weighting scheme is to simply equally weight each position. (If we’re prepared to put up with a little extra volatility for a little extra return, we can also Kelly weight our best ideas). Here are the equal weight return statistics for book value.

Annual and Compound Returns (Portfolio Constituents Equally Weighted)

Book Value EW 1926 to 2013

In the equal weight backtest value generated 20.2 percent compound (27.3 percent on average), beating out glamour’s 6.3 percent compound return (10.4 percent on average).

Since 1951 the equally weighted PB value decile has generated a compound annual growth rate (CAGR) of 20.0 percent and an average annual return (AAR) of 25.4 percent. Over the same period the glamour decile returned a CAGR of 6.4 percent and an AAR of 10.8 percent. These returns are approximately the same as the returns generated by PCF and PE over the same period.

Book Value-to-Market Equity (Equal Weight)

Book Value to Market EW 1926 to 2013

Again, the value portfolios outperformed because they bought more book value per dollar invested than the glamour portfolios: 4.57x on average versus 0.25x  in the glamour portfolios.

Recent Performance (Equal Weight)

EW PB Returns 1999 to 2013

In the equal weight portfolios, value has really outperformed glamour since 1999, beating it by an extraordinary 15.9 percent compound, and 16.1 percent in the average year.

As we’ve seen over the last few weeks, over the long run, and with some regularity, cheap stocks tend to outperform more expensive stocks. Like the PE and PCF ratios, the PB ratio is a very useful metric for sorting cheap stocks from expensive stocks.

My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more.

Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

The price-to-cashflow ratio (PCF) is a popular metric among value investors. Many believe that using cashflow, rather than accounting earnings, delivers a truer picture of a company’s business performance, which in turn leads to better investment performance.

Set out below are the results of two Fama and French backtests of the cashflow yield (the inverse of the PCF ratio) data from 1951 to 2013. As at December 2013, there were 2,526 firms in the sample. The value decile contained the 269 stocks with the highest earnings yield, and the glamour decile contained the 311 stocks with the lowest earnings yield. The average size of the glamour stocks is $4.74 billion and the value stocks $4.80 billion. (Note that the average is heavily skewed up by the biggest companies. For context, the 2,526th company has a market capitalization today of $272 million, which is much smaller than the average, but still investable for most investors). Stocks with negative cashflow were excluded. Portfolios are formed on June 30 and rebalanced annually.

Annual and Compound Returns (Portfolio Constituents Weighted by Market Capitalization)

In this backtest, the two portfolios are weighted by market capitalization, which means that bigger firms contribute more to the performance of the portfolio, and smaller firms contribute less. Here we can see that the value decile has comprehensively outperformed the glamour decile, returning 16.7 percent compound (18.6 percent in the average year) over the full period versus 9.3 percent for the glamour decile (11.5 percent in the average year).

VW PCF Returns 1951 to 2013 v2

These returns are practically identical to the returns found for the price-to-earnings ratio in last week’s post (Investing Using the Price-to-Earnings Ratio and Earnings Yield (Backtests 1951 to 2013)).

Cashflow Yield (Market Capitalization Weight)

The reason for value’s outperformance is simply due to the fact that the value portfolios generated more cashflow per dollar invested; 27.2 percent versus 4.3 percent for the glamour portfolio. (I used a rolling average this week. The “average” I’ve quoted is for the full period. The rolling average has been higher, but it’s rarely been lower.):

Cashflow Yield VW 1951 to 2013

Recent Performance (Market Capitalization Weight)

As we saw last week, value’s outperformance over glamour is not a historical artifact. If we examine just the period since 1999, we find that, though the return is lower than the long term average, value has continued to be the better bet.

VW PCF Returns 1999 to 2013

Value has continued to outperform glamour since 1999, beating it by 8.7 percent compound, and 6.2 percent in the average year. The reason for lower returns recently may be due to the popularization of simple value strategies, but I think it’s more because the market is still working off the massive overvaluation of the late 1990s Dot Com boom.

As I noted last week, market capitalization-weighted returns are useful for demonstrating that the outperformance of value over glamour is not due to the value portfolios containing smaller stocks. Unless you’re running an index (or hugging an index), they’re not really meaningful. The easiest portfolio weighting scheme is to simply equally weight each position. (If we’re prepared to put up with a little extra volatility for a little extra return, we can also Kelly weight our best ideas). Here are the equal weight return statistics for the cashflow yield.

Annual and Compound Returns (Portfolio Constituents Equally Weighted)

EW PCF Returns 1951 to 2013 v2

In the equal weight backtest value generated 20.7 percent compound (23.8 percent on average), beating out glamour’s 9.3 percent compound return (12.5 percent on average). Folks who saw last week’s post might note the small advantage for the cashflow yield’s value decile over the earnings yield’s value decile, 20.7 percent versus 20.1 percent. We’ll examine the significance of this small win by cashflow in the coming weeks.

Cashflow Yield (Equal Weight)

Cashflow Yield EW 1951 to 2013

Again, the value portfolios generate more cashflow than the glamour portfolios, generating 24.6 percent on average versus 4.1 percent in the glamour portfolios. As we saw last week, the average cashflow yield for the equally weighed value portfolio is slightly lower than the average cashflow yield for the market capitalization-weighted portfolios, which indicates that, over the full period, bigger stocks tended to be a cheaper method for buying cashflow than smaller stocks. That won’t always be the case, but it’s interesting nonetheless.

Recent Performance (Equal Weight)

EW PCF Returns 1999 to 2013

In the equal weight portfolios, value has really outperformed glamour since 1999, beating it by 11.1 percent compound, and 10.0 percent in the average year.

As we saw last week, over the long run, cheap stocks tend to outperform more expensive stocks. Like the PE ratio, the PCF ratio is a very useful metric for sorting cheap stocks from expensive stocks.

My firm, Eyquem, offers low cost, fee-only managed accounts that implement a systematic deep value investment strategy. Please contact me by email at toby@eyquem.net or call me by telephone on (646) 535 8629 to learn more.

Click here if you’d like to read more on Quantitative Value, or connect with me on LinkedIn.

Read Full Post »

Older Posts »

Follow

Get every new post delivered to your Inbox.

Join 4,429 other followers

%d bloggers like this: