Long Streaks and Corrections

Investors in US stocks have experienced one heckuva’ run since late 2012.  In fact, it ranks as one of the most magnificent runs of the past 50 years if viewed from the perspective of the 200-day moving average.  Through today, the S&P 500 has been above its 200-day moving average for 302 straight trading days.  Even with this months correction, the S&P 500 would have to fall another 4.5% to breach the 200-day MA, meaning that barring some crazy action in the next few days, the streak should continue for at least a little bit longer.   If the market did happen to fall 4.5% this afternoon and close below the 200-day, this still would rank as the 9^th longest streak over the past five decades.  Here is the Top-10 list:

Rank	Days above 200-day MA	Streak Ended
1	525	8/26/1998
2	394	7/12/1996
3	386	6/8/1965
4	363	3/24/1994
5	355	9/14/1959
6	333	12/13/1983
7	332	4/3/1962
8	313	7/15/2004
9	302 (and counting)	1/31/2014 (and counting)
10	293	1/19/1990

What’s happened performance-wise in the S&P 500 around the end of the other nine streaks on the list?  We looked at the peak to trough price performance for the index around these dates, with the peak representing the peak price during the streak above the moving average and the trough representing the low closing daily price in the weeks or months following the end of the streak.  As you can see below, the corrections associated with streaks of this magnitude aren’t pretty.  But, as we’ll point out in a bit, there are a few silver linings.  First, here is the above list with peak to trough declines added:

Rank	Days above 200-day MA	Streak Ended		Peak to Trough
1	525	8/26/1998		-19.15%
2	394	7/12/1996		-7.64%
3	386	6/8/1965		-9.61%
4	363	3/24/1994		-8.94%
5	355	9/14/1959		-13.85%
6	333	12/13/1983		-14.38%
7	332	4/3/1962		-27.97%
8	313	7/15/2004		-8.17%
9	302	1/31/2014		????????
10	293	1/19/1990		-17.88%

The average peak to trough decline during corrections associated with and following the end of long streaks above the 200-day is -14.18%.  The median is -13.85%.  

Currently, the S&P 500 at 1785 is a touch less than 3.5% below the peak of 1848 set last month.  If (and that’s a big “if”), this correction proves to be a streak buster, past episodes suggest that there’s more downside to this correction.  As seen above, the most benign correction associated with streaks of this magnitude was -7.64% in the 1996 timeframe.  If the current correction matched that correction, the S&P 500 would fall to a level of 1707.  If the correction matched the overall median for the other members of the top-10, the S&P 500 would fall to 1592.  

Now, it’s time for the silver linings.  

First, in a reflection of just how powerful the move upward was during the back half of 2013, a correction matching the median correction for the top-10 streaks would take the market back only to the levels observed in June 2013, a mere seven months ago.  While a nearly 10% correction would invite all sorts of hysteria in the echo chambers, in reality the hysteria would be a function of the magnitude of peoples’ anchored expectations.  When markets advance dramatically, investors extrapolate and expect those market returns to continue without hiccups.  When hiccups arrive, it’s hard for investors to put the correction in perspective.  They only think about the peak level.  A move back to the summer levels wouldn’t be that big a deal; most investors polled last summer would probably have taken a sideways market the rest of 2013 considering how negative many investors were at that point.  

Second, when looking at the top-10 list above, not one of those streak-ending corrections marked the beginning of a historic market calamity.  In each case, except the extreme decline of 1962, the S&P 500 achieved new all-time highs within a year of violating the 200-day moving average.  In the exceptional case of 1962, the S&P 500 achieved new highs within a year and a half.  Furthermore, in all cases except 1998, the long streaks and subsequent corrections occurred at the beginning of or in the middle of longer-term uptrends that continued for several years after the long 200-day MA streaks listed above concluded.  In the case of the end of the 1998 streak (the longest streak above the 200-day MA), the market ended up rallying sharply through 1999 and early 2000, only to collapse from 2000 to 2002.  The S&P 500 peaked just south of 1200 before the correction of late ‘98; the S&P 500 ended up shooting to a closing high of 1527 in March 2000, over 25% above the highest levels achieved during the streak.  

Third, in six of the nine other streaks in the top-10 listed above, market price exceeded 20x normalized earnings (Shiller CAPE P/E) and in two more, valuation was just south of 20x, but above the long-term median.  1983 was the only such extraordinary streak that occurred with normalized valuations below historical median.  As such, arguments that the current streak will prove “different” or “more dangerous” because valuations are stretched don’t hold water. As we discussed a few weeks ago, valuation indicators, even long-term indicators, are terrible short to intermediate term timing tools.

Thus, while there are always exceptions to a rule and firsts for everything, back and fill corrections associated with past extraordinary streaks above the 200-day moving average have proved to be speed bumps on the way to bigger gains in relatively short order.  The historic market declines, such as those experienced in 2008, 2001 to 2002, and 1973 to 1974 have generally followed periods of market churning that provided warnings signs that trouble was on the way.  

It’s no fun to see red on the screens and to hear the negative banter on TVs and in the press, especially to kick off a New Year.  There’s nothing at this point, however, to suggest that this correction marks the beginnings of a 2008 repeat.  

Keep Arms and Legs Inside the Ride at All Times

As we write this note today, the market is having its first hiccup of the New Year, not to mention its first hiccup in what feels like many weeks.  The S&P is down nearly 3% for the month, which of course invites incredible amount of teeth-gnashing.  This weekend, we’ll hear from everyone that world markets are going to Hades and back and that social unrest is right around the corner.  Ok, that’s a bit of an exaggeration, but suffice to say there are a number of people running around with the heads on fire and their rear-ends catching because there’s red on the screens.  Considering investors experienced a somewhat unusual year last year with 30%+ returns and amazingly low volatility, it comes as no surprise that investors are jolted by some cold water to the face.

We planned on undertaking this exercise anyway, but it becomes more interesting on a down nearly 2% day for sure.  Using simple statistics, it’s interesting to look at yearly and monthly S&P 500 returns and get a glimpse of how much randomness and variance/volatility actually occur through time.  Our minds tend to conveniently ignore these facts, especially after outstanding years like 2013.  Over time, though, we all need reminders that outcomes like the start to this month and year are actually relatively high probability events.  The current month isn’t exceptional (at least not yet).  Let’s get to the wonkish work.

We looked at yearly S&P 500 total returns (price change + dividends) going back to 1931 as well as monthly returns (just price change) over that time period.  The simple statistics are quite interesting for sure.

Let’s start with the yearly numbers.

Average total return for the S&P 500 over that 83-year period is 11.81%.  Standard deviation for that set of returns is 19.31%.  Under a normal bell-curve, approximately 50% of outcomes should fall between +0.674 and -0.674 standard deviations from the average.  95% of outcomes should occur between -1.96 and +1.96 standard deviations from the average.  We know that markets have slightly “fatter tails” than typically expected under a normal bell curve, so we made a few minor adjustments to account for this fact.  After those adjustments, we find that the 50% range for yearly market performance outcomes is between 24.9% and -1.28%.  60% of outcomes should fall between 31.94% and -8.33%.  Therefore, there’s a decent probability in any given year that S&P 500 returns can fall in a pretty wide range.  Let’s say the market doesn’t move much from today’s close and finishes down 3%.  Statistically, that wouldn’t be that phenomenal of an event, though pundits and analysts would tell you it feels like the end of civilization.  Returns like that would be frustrating after such a strong run of performance years, but nothing that shouldn’t be expected having participated in the markets for a while.  And remember, 50% of yearly returns should fall outside of 24.9% on the upside or -1.28% on the downside.  Let’s expand the probability range, if you will, out to 95%.  That range is +50.23% to -26.03%.  Beyond those values, we see the monster tail events that are talked about for eons and eons.  Sure enough, there are 5 years total that fall in the extreme tails (6% of outcomes, close to the 5% that should occur in the extreme tails), 3 down, 2 up.  The three extremely negative years are 1931, 2008, and 1937.  The two monster up years were 1933 and 1954.  Of note, 1974 came very close to joining the others, down 26.00%.  Basically, years like this should, up or down, should occur once every 20 years on average.  

Now, let’s look at the monthly numbers.  

Average monthly return over the past 83 years is 0.59%, with a standard deviation of 5.43%.  Like the yearly numbers above, we can quickly calculate the probability ranges for monthly outcomes.  50% of outcomes should fall between 4.25% and -3.08%, again a pretty wide range, at least relative to many investors’ expectations.  Accordingly, 50% should fall outside of that range, up and down, over time.  As of today’s close, the S&P 500 is down 3.04%, which actually falls within the 50% range.  It’s painful, yes, but nothing to write home about in the long history of US markets.  Think about this.  If we’d written at the end of 2013 that half of the months during 2014 could come in either above 4.25% or below -3.08%, many off the top of their heads would call us crazy.  Yet, statistically, this isn’t necessarily a crazy notion at all.  Certainly, markets and data sets aren’t that clean.  The point remains, however: market movements are more random and violent than our heads perceive them to be.  By the way, what is extreme on a monthly basis?  The 95% range is 11.23% to -10.06%.  Out of 1007 months, 45 months fall outside of this range, 4.5% of the outcomes.  Again, this is close to what we should expect in the extreme tails.  A negative 10% month or worse is something to write home about.  Fortunately, we haven’t witnessed one since early 2009 when markets were tanking and scaring the wits out of everyone.  

Moving beyond the numbers, there are several takeaways we can pull from this analysis.

First, equity markets deliver solid returns over long periods of time, but deliver a decent amount of volatility in return.  As hard as it is, expectations for institutional and individual investors must be set accordingly; otherwise poor decisions detrimental to performance are made.  

Speaking of too much noise and poor decisions, much of this industry, especially the financial media, is geared towards encouraging professionals and individuals to trade or react around market moves that in retrospect are rather unexceptional and “to be expected”.  On a down 2% day, there’ll be no shortage of serious banter on CNBC about the end of markets.  Stock selection “lightning rounds” will be particularly electric.  Study after study shows that investors are terrible short to intermediate-term market timers, even those with teams of analysts and expensive market analysis tools at their disposal.  For nearly all investors, spinning one’s wheels to turn, let’s say, a 3% monthly decline or an 8% yearly decline into a positive month or year, will probably do much more harm than good in the end.  

Instead of worrying about the 2% or 3% or 5% down months, or the 0% to -10% down years, it behooves all to think more about protecting against the yearly massive tail events, i.e. the 1974s, the 2008s, and the 1931s of the world.  Those provide the most devastating shocks to long-term portfolio performance.  This, of course, is easier said than done though, believe it or not, chopping off a portion of these tail events can be accomplished through many different methodologies.  Many institutions have produced models successful in identifying situations during which probabilities are high for serious, bone-crushing bear markets.  Many market professionals, for instance, anticipated the events of 2008/2009.  For individuals, perhaps taming the volatility dragon involves employing a strategy like a 60/40 equity/bond model or employing some of the simple hedging ETFs out there as a small portion of a portfolio.  

In any case, this brings us to our final point, one mentioned last week as well.  Institutional and individual investors alike should have a process or plan and stick to it, as hard as that is under many circumstances.  Whether a technical trading risk or rebalancing or dollar cost averaging system or long-term investor plan/statement, having a solid plan in place allows one to ignore noise during months like these that fall in the typical range and avoid execution errors that occur time and time again.  There’s no shortage of people that bought the hype and bought “growth” stocks and funds at the 2000 internet peak, nor is there a shortage of investors that dumped every asset under the sun at or near the very lows in March 2009, only to keep piles of cash on hand as markets doubled or more around the globe.  Shooting from the hip and making decisions based upon news mentioned in the financial press, or on TV, or in analyst research reports is a recipe for poor long-term outcomes.  

The statistics above show us that performance noise is a part of the month-to-month and year-to-year existence of an equity market investor.  It’s always surprising to see how noisy “normal” truly is.  Letting noise affect decision-making, a common problem for us humans, is the biggest problem in the markets.  If investors spent more time creating objective intermediate to long-term investment plans and processes, and less worrying about what is happening in X or Y country around the world or in Washington DC, outcomes would improve significantly.

Fundamental Free Lunch: Revisited

Several months ago, we back tested S&P 500 performance using a very simple fundamental model to show that fundamental ratios applied consistently can produce outsized returns.  In our original example, we rebalanced a model portfolio annually with the cheapest 100 stocks in the S&P 500 based up trailing EV/EBITDA (Enterprise Value to Earnings before Interest, Taxes, Depreciation, and Amortization).  The portfolio is equal weighted.  This top quintile valuation approach applied each year produced annualized returns that significantly outperformed the annualized returns of the S&P 500 from 1994 through 2012.  

Obviously, 2013 was a big year for the S&P 500, with the Index’s total return finishing at 32.38%.  How did the simple/passive EV/EBITDA strategy work out?

Quite well, actually.  Buying the 100 names in the S&P 500 with the lowest trailing EV/EBITDA ratios (as gauged by Bloomberg) on 12/31/12 would have returned 46.79% this past year.  Therefore, since the end of 1994, the simple portfolio has returned 14.57% per annum vs. 9.62% for the S&P 500.  Keep in mind that the model portfolio numbers from the back test aren’t inclusive of dividends unlike the S&P 500 returns.  

Like last time, there are a few caveats.

First, while the numbers from the simple back test significantly outperform over time, just because the names come from the “value” category doesn’t mean they offer shelter in the storm when markets become volatile.  Annual standard deviation for the model is higher at 22.6% versus 20.1% for the index.  The largest yearly loss for the passive model is -40.97% in 2008, almost 4% below the -37% number posted for the S&P 500.  

Second, as we pointed out last time, fundamental models like this can be very streaky.  The simple back tested model underperformed the S&P 500 for the first five years of the time period, 1995 through 1999, admittedly a tough time for value type names relative to their cousins in the growth category.  Since the end of 1999, the simple model has only underperformed in two years: 2008 and 2012.  No doubt, as with any strategy, there will be another sustained underperformance period for this simple model.  John Maynard Keynes (may have) famously said, “Markets can remain irrational longer than you can remain solvent.”  A twist on this might be, “Fundamental market models can underperform longer than you can maintain your scant patience and sanity!”  The abandonment issue can become particularly acute when you have clients leaving in droves because the portfolio has underperformed.  Ask many value types that managed portfolios and managed to survive through the late 1990s.  Though proven right over time, investors left in droves.  It’s easy to talk about patience and long-term success, but it’s hard to stick to ones guns in the short to intermediate term when career pressure builds.  Of course, it’s the greed/fear career preservation psychology that keeps many from outperforming over time.

Third, and perhaps a corollary to the second point, the above performance shows how much of an impact simplicity and process discipline make in portfolio success over time.  Often, investors are attracted to the notion of using complex ideas and theses to achieve success.  We all love intricate tales.  This example shows that simple reversion to the mean fundamental techniques work when selecting individual names.  A macro portfolio example might be the simple 60/40 equity/bond passive portfolio which has outperformed many indices over time with better risk metrics.  Again, having the patience and confidence to execute a simple process through good times and bad is trickier than most people understand.  

While this simple test focused on EV/EBITDA, other studies produced over time with different fundamental ratios such as price to book and trailing P/E ratios also show statistically significant outperformance over time.  Reversion to the mean gives us exploitable “free lunches.”  Unfortunately, the gulp factor intervenes when its time to pull the trigger on undervalued investments.  In any given year, many of the names on the buy lists that produced the above results would have been the names showing up in the financial press with negative stories attached or with a string of negative sell-side analyst quotes in the margins.  At the end of the day, the itch to chase what’s working and to shun what hasn’t been working is far too strong for many humans.

January Effect:

Each year, investors and pundits consistently dole out a bread and butter piece of market wisdom, something along the lines of, “So goes January, so goes the year.”  This ranks up there with “Sell in May, Go Away,” and the “Santa Claus Rally” among the popular nuggets to guide investors and traders through the minefield.  Though January is still young, we thought it might be interesting to look at some of the data for ourselves and see if January performance offers any clues for the rest of the year.  Is there such thing as a January effect?  

To look at the January effect, we used simple S&P 500 yearly and monthly returns going back to January 1950 providing 64 January and yearly performance data points (Side note: we tend to use the S&P 500 because the historical record is much longer than the international indices).  Over that 64-year period, 23 Januaries posted negative price performance.  Granted, it’s not a massive sample size, but it’s something to work with.  

Now, let’s get into the data.  

Looking at overall yearly results lined up with positive or negative Januaries is somewhat supportive.  Out of the 23 years with negative Januaries, 12 posted negative overall years and 11 ended in positive territory.  Of the 41 years with positive Januaries, 36 were positive, nearly 90%.  Surely, the record following negative Januaries is a coin toss, but keep in mind that only 17 out of the total 64 years overall were negative (26.5%), and 47 were positive; there’s a positive bias to the yearly numbers to begin with.  The odds for a negative or positive year following a negative or positive January are higher than the odds for a positive or negative year coming into the year in general.  

The actual performance data from the 23 years with negative Januaries is interesting.  Since 1950, the average annual return for the S&P 500 is 9.1% and the median is 12.04%.  For the 23 years experiencing negative Januaries, the average comes in at -3.76% and the median -2.97%, a very large difference.  Ultimately, we found the source of that gap interesting.  Overall, the 47 positive years since 1950 averaged returns of 17.1%, with a median of 15.63%.  However, the 11 years that turned out positive following a negative January only posted average returns of 8.74%, with a median of 4.46%, far below their positive year compatriots.  The 17 negative years since 1950 posted average returns of -13.02%, with a median of -11.50%.  The 12 negative years with negative Januaries posted average returns of -15.22%, with a median -11.66%, not much different than the numbers for all negative years.  Therefore, for whatever reason, a negative January really seems to constrain the potential for upside during positive years.  

Going to the positive side, a positive January seems to take the edge off an overall negative year.  Years that turned out negative after posting positive Januaries show average returns of -7.76% and a median of -9.30%, less than the overall average and median for negative years.  Conversely, positive years with positive Januaries come in better than the overall average and median.  The average performance is 19.65% and the median is 19.02%. 

To keep the numbers straight, we’ve included a table below:

S&P 500 PERFORMANCE, 1950-2013
	Average		Median
Overall	9.10%		12.04%
Positive Years	17.10%		15.63%
Negative Years	-13.02%		-11.50%
Yearly Performance after a Negative Jan.	-3.76%		-2.97%
Negative Jan., Positive Year	8.74%		4.46%
Negative Jan., Negative Year	-15.22%		-11.66%
Positive Jan., Positive Year	19.65%		19.02%
Positive Jan., Negative Year	-7.76%		-9.30%

We’ve thrown a lot of numbers out there.  What’s the quick takeaway for us?  Obviously, if January 2014 closes negative (the S&P 500 is down about 0.3%), it doesn’t mean it’s time to close shop for the year and run for the hills.  Exercises like this can be opportunities to exercise our creative gene and have some fun with market data, but shouldn’t be taken as a catalyst to make wholesale, major investment changes.   However, a negative close would be another bit of evidence to sock away that 2014 might be much more of a wrestling act than the past few years.  As we’ve seen above, the performance record for years with negative Januaries, whether they turn out positive or negative, is much lower than the overall performance record.  For some reason, the animal spirits of January, or lack thereof, seem to spill over.  There are a few weeks left, though, plenty of time for markets to move up, up and away (hopefully).  

Sentiment Update: New Year's Edition

Markets closed out 2013 with a dash to new highs, further amplifying talk of bubbles.  As we like to do every several weeks, let’s review some sentiment indicators to see whether or not the underlying data is indicating frothy behavior.

Unlike our last few sentiment reviews, the picture has started to lean more to the “overly-optimistic” side, perhaps giving credence to those looking for a short-term correction.  Encouraging news, though: across the board, the indicators we watch haven’t even reached levels observed in early 2011 following the 2010 market recovery and preceding the rough summer and fall of 2011.  Sentiment-wise, markets are far from reaching extreme levels.

First, let’s look at the CBOE Put/Call ratio.  We use the 10-day moving average.  High values indicate more puts trading relative to calls, hence more bearish trading activity.  Keep in mind these indicators are contrarian.  Too many bulls and the market may be due for a correction.  Too many bears, and the market may be due for upside.  The Put/Call ratio’s 10-day average at 0.715 has reached the lowest levels since late 2010/early 2011.  This is also close to some of the lowest values observed over the past nine or ten years.  Thus, CBOE Put/Call trading is showing that complacency has crept into the marketplace, though complacency is nowhere close to the complacency of the very late 1990s.  The chart follows, going back to 1995:

Next we’ll look at a separate index covering options market activity, the ISEE All-Equities Sentiment Index.  According to the International Securities Exchange, the “ISE Sentiment Index is a unique put/call value that only uses long customer transactions to calculate bullish/bearish market direction.”  In this case, higher values indicate more bullish activity and lower levels indicate more bearishness.  Like the CBOE Put/Call, we’ll take the 10-day moving average to smooth out results.  This indicator provides a more neutral sentiment picture.  The current value of 166.20 is slightly below the average of 169.74 going back to 2006.  As you can see in the chart below, extremely high values in late 2007 and in late 2010/early 2011 preceded significant equity market volatility.  So far, this indicator shows that sentiment has yet to come close to reaching extremely positive readings.

Finally, we’ll shift more towards sentiment among individual investors.  In this case we use the Farrell Individual Sentiment Indicator, which normalizes bull, bear, and neutral market readings from the weekly AAII Bull/Bear data to better capture long term sentiment trends.  In this case, we rely on the 10-week moving average of the indicator.  Values above 1.50 indicate extreme bullishness, while values below 0.50 indicate extreme bearishness among individual investors.  At 1.11, the indicator sits above its long-term average of 0.93 going back to 1987, but well below values associated with past extreme bullishness.  Interestingly, as we’ve observed in past sentiment reviews, individual investor sentiment has actually spent a good portion of the past two years probing the lower end of the range.  Mutual fund cash flow and other indicators show that many investors are only just now beginning to re-dip their toes back in the water.  This indicator provides a decent visual representation of investor reticence to get involved in equity markets since the financial crisis.

Add it all up, and we get a picture of market sentiment that is slightly above longer-term averages, but a good bit away levels associated with major market cliff-dives.  How does this fit into a potential market performance picture for 2013?  As we observed a few weeks ago, years following 20%+ up years are usually not as robust in terms of upside gains, but usually up nonetheless, with returns tracking around historic averages.  Sentiment indicators currently show that markets don’t quite have the energy-source in terms of high levels of bearishness that precede sharp, sustained, record up moves.  However, there’s still fuel left in the tank before markets become overly optimistic.  This is a sentiment picture that fits well into a view that there is a decent probability for “good” not “great” market performance this year.  We’d expect a few challenging episodes this year in terms of corrections in contrast to 2013 where the biggest peak to trough correction in the S&P 500 totaled 7.5%, benign by historical standards.