A thousand false summits

Mountain climbers know the feeling.  Working one's way up, many thousands of feet to a peak, only to then discover that it is just a base and not the ultimate summit.  The actual summit, and the long road in-between, is still much further ahead and well beyond one's visibility - even when standing near a lower peak.  This process, of being misled through foggy illusion, is similar to the folly-filled process we have with picking market summits, either before or many years after the fact.  We show here that this is the main essence of market timing. 

Equity market participants assumed that the crash, which occurred after the technology bubble of the late 1990s, would have been the last that they would see those levels at, ever again.  During this market crash, the S&P fell from the 1500s, down to the 700s in 2002.  And in 2007, a very insignificant few thought, when the market barely peeked above those memorable 1500s again, that this would just usher in a market fall that would take us to levels below even the 700s that came after the technology bubble crash. 

But that's what exactly what happened, and in early 2009 the S&P bottomed in the 600s.  For Americans being told to reinvest from after when the S&P bottomed in late 2002, only to lose it all again in 2008, this has given them broad psychological wear and tear.  It's left them with a general mistrust of markets and their promoters.  Suddenly some are now super rich (unlike economist Piketty's book, we'll argue below that some of these will lose their share within a generation), and many others feel something in the system smells "rigged".  This also does not take into account the economic double-whammy of the Administration having to get involved as millions of Americans completely lost their home value, in addition to their jobs.  A number of important industries also teetered on insolvency.  Having led TARP's analytics team during this time gave a first row perspective on analyzing the underlying complexities linking these different parts of the economy, strained in a historically unprecedented time.  The lowest level of the market since the technology bubble, came two recessions later in 2009.  Could people have been misled into just seeing two cycles (the first one being a false summit), when it was really one super cycle?

What happened to employment during this time?  The unemployment rate (UR) in 2007 had bottomed at a higher level then the UR bottom from the technology bubble.  Things never really fully recovered!  In 2009, the UR shot up even higher than what we saw from the technology bubble recession (today we're still just at this early millennium UR high).  We can still see the psychological scars today on the faces of those who falsely believed the worst was over in 2002, and these people hold high-level jobs in many influential fields.  Having been misled by official experts on TV, who barely know better, the general public's opinion of the economy and the markets are now appropriately jaded.

The markets are in fact very difficult to time.  It’s of course difficult to know too where any bottom is.  But this latter point we show is secondary to the more important difficulty of knowing where the market tops are (we will interchangeably here use the term "top" and "summit").  Put a different way, if one were to have sold off their equity holdings just once during the past 15 years, it would have been better to have done so in 2008, as opposed to in 2000-2001.

There are many statements from famous businesspeople, which unfortunately conflate these ideas of market timing and finding many false bottoms.  Vanguard founder, John Bogle, has often correctly warned of the general risks of market timing.  But recently he mostly focuses on just a false reason that even if one can correctly call the market summit, then one still shouldn’t get out of the market because he or she wouldn’t know when to get back in.  It is on this sub-logic where we argue something different.

We know it is clearly, highly difficult to skillfully outperform the market by a significant amount for a lengthy time period, through ongoing stock selection.  But we have a system where there are also many people who have their bread and butter dependent on many other people believing the opposite.  But there are a number of studies that employ different empirical, and econometric techniques, and they mostly come to the same broad conclusion even as each of the decent study's details will differ slightly.  Warren Buffett himself directing in his will that his wife's funds be 90% in the S&P 500 is a clear signal of how he feels about active management through BRK, or anywhere else.  The recent research on this blog that has received popular attention is somewhat unique in that it further clarifies for different time horizons, the outperformance level needed to establish skill and the bifurcation (of this rare portion of the population who have) into skill versus by luck. 

It also makes sense that that most people shouldn’t attempt these other active management strategies either, such as market timing.  It is theoretically reasonable, similar to the dense probability analysis we have on stock selection, to assume that if most investors are trying to time the market, then more than half of them will be less successful at it.  Hence those would be better off not trying.  Put differently, if most were successful at market timing, then the markets would never have a chance to get so out of equilibrium simply for everyone free joint benefit throughout time.  Thus defeating anyone’s ability to market time at all.

There are a few popular quotes, which have also expressed similar concerns on this narrow topic of market timing difficulty. 

“I can't recall ever once having seen the name of a market timer on Forbes' annual list of the richest people in the world. If it were truly possible to predict corrections, you'd think somebody would have made billions by doing it.”

This quote is from the iconic American businessman Peter Lynch, who himself had an exciting decade-plus run versus the market.  His quote points to a couple things that we’ll discuss below.  Before we do that though, let's take note that others in the industry, including David Babson, who have echoed similar comments.  He noted that market timers only appear intelligent but if they actually knew what they were doing, “they would become a billionaire so quickly he would not find it necessary to sell his stock market guesses to the general public.” 

It is difficult to market time successfully, but is being a billionaire the proper test for whether anyone can do it?  While we're at it, why do we expect billionaires to have extraordinarily more insight versus anyone else on technical matters outside of the narrow things that once got them to be wealthy?  A higher fraction than most would assume, of these Forbes 400 billionaires, will see their family's name dropped off that list within a generation.  A higher fraction than most would assume, of alternative investment funds, run into trouble (see here, and here for some example stories).  Today the reference to a billion dollars seems to get tossed around in popular commentary, as though it were throwaway change, but it's not.  Sure, today having just over a billion dollars is the required pass into the club of the 400 wealthiest Americans.  We can also state that in the summit year of the technology bubble, half the amount was necessary to join this club, or hundreds of millions of dollars in net worth.  Both inflation context and some degree of rationality should be used when whether what a billionaire thinks about any sort of technical issue outside their narrow expertise (and these hundreds of billionaires are quite heterogenous in skill set), is the correct test to disprove that idea. 

We also need a proper definition of what market timing is, so that we can discuss ideas amongst ourselves using a reasonable reference.  Some may consider market timing be something we perform annually, some consider it daily, and yet others consider it on an intra-day tick by tick basis.  It is from the vantage point of the individual performing the analysis, combined with the computational tools of the day, which influences what the market timing definition is.  In this article we use a scale of looking at weekly market data here.  From a rich probabilistic stand-point, it is likely that successful market timing would not carry more alpha versus someone who successfully performs ongoing stock-selection.  And anyway, both are highly difficult for most people to do consistently well, and of course including those in the financial industry.

There are also the societal costs of the wear and tear associated with misidentifying the historic inflection points of the market and of the economy.  This has been true across the centuries, and around the globe.  The 20^th century British economist John Maynard Keynes wrote the following in his magnum opus book.

"We have not proved able to take much advantage of a general systematic movement out of and into ordinary shares as a whole at different phases of the trade cycle....As a result of those experiences I am clear that the idea of wholesale shifts is for various reasons impracticable and indeed undesirable. Most of those who attempt it sell too late and buy too late, and do both too often, incurring heavy expenses and developing too unsettled and speculative a state of mind, which, if it is widespread, has besides the grave social disadvantages of aggravating the scale of fluctuations.”

Even the brightest advisers fall victim to the exquisite charm of market timing.  From Keynes, to modern Nobel laureates in economics, we see famously regrettable acts on peaks, which turned out to be false summits.  But anyway given the various backdrop provided by other businesspeople and economists, let’s now mathematically step into why it is difficult to market time.  Again it is argued here that identifying the actual summit is the more relevant component, and not necessarily the ability for people to identify an actual bottom.  It is more difficult to collect robust data to analyze what fraction of investors market time, at any point in time.  This is similar to when we discussed at the early Volcker Rule making analysis and knowing it is difficult in a complex pile of transactions to systematically identify what's risk taking versus hedging.  A significant portion of market participants market time however.  During the recent financial crisis, for example, there was obviously a large fraction of investors selling stocks from late 2007, through early 2009.  Market exchanges violently heated up.  Some were also "selectively" buying along this downturn, trying hard to outsmart one another by picking the bottom.  Yet others continued to sell early rallies from 2009, amazingly misidentifying the market as somehow master-minding a head-fake rally (i.e., a “bear trap”). 

For mathematical and risk purposes though, what we know better about is how variable the market fluctuation are, as it grows over time.  This stochastic process can be modeled over a richer history of the U.S. markets, and we can see this interesting mathematical convoluted outcome for both single-day and major multi-day drops in the market.  Using both of these empirical stories, in combination, provides enough of a sample framework from which to establish a framework over the merits of market timing.

It makes sense for the rare and highly skilled people who can spot a market top, with the defined multi-percentage market bottom somewhere ahead of it, to remove themselves from the market then for a while.  This is contrary to Bogle’s self-defeating opinion, which is mostly that market tops just can’t be every be identified by anyone in a way that can be acted upon.  Now we have established that the market top definition here is one that has a statistically significant enough of a subsequent market drop, and thus if in fact discovered it would warrant market timing.  In theory, these opportunities can be resolved so quickly that getting back into the market could be elusive.  An example of this is, say, if the moments prior to the Flash Crash had qualified. 

There is a significant body of economic and investment data showing that this is usually not the case.  For substantial pullbacks in the market, one can safely hold out just a couple weeks for the risk patterns to subside.  This is all again keyed off the premise that one knows where the correct market summit is to begin with.  What we can not do (and is elaborated on below) is blanketly advise all people that no one can employ such a strategy, even for just a basic period of a couple weeks. 

We also see that rarely does market timing become so tricky, within just a couple week period, that one’s entire strategy is completely beholden to knowing when to return to the markets.  Actually instead, the counter-example is stronger that missing one of these bottom chance(s) to return, only means that one can always repurchase the recovered market back at the same level as the prior top.  Thereby giving the individual neither a gain nor loss on the trade.

Now let’s see these ideas with the past 12 months of empirical data, using a broad set of different equity markets around the world.  Here we look at a varied sample including the S&P, FTSE, Nikkei, and Bovespa.  In the case of the FTSE and Nikkei, the market shape forms the start of an concave arc, which makes the results exciting and valid.  Though we should continue to re-evaluate for some time still (the 2009 nadir was not known until 7 years after experts assumed the 2002 bottom was the all-clear).  Also this strategy doesn’t work as well right now for Bovespa, but only given the choppy slide down in the past year for the index.  It's fine to include it still in the analysis, given it is an emerging market, and simply visually identify the top to aid our analysis.  Altogether we can see the results below, and you can play with the data yourself.  It is freely available on the accompanying public web portal.

S&P, plus investment proxy if one removed themselves from market for a couple weeks at top(s)

FTSE, plus investment proxy if one removed themselves from market for a couple weeks at top(s)

Nikkei, plus investment proxy if one removed themselves from market for a couple weeks at top(s)

Bovespa, plus investment proxy if one removed themselves from market for a couple weeks at top(s)

We know collectively that these indices give a sense for different market ranges participants might expect, across different market periods.  While it's not required to focus only on the U.S., here it best represents the typical scenario for theoretical modeling.  There is also flexibility to assume even finer time frequencies (versus weekly), which would only enhance the results. 

In all of the markets shown, sitting out a couple straight weeks would have been fine generally at least once annually, and this is normal.  And this also would have overall led to a small amount of alpha for the investor.  In none of the markets did we see a nadir in just a single-week’s timeframe from the summit, hence we have alleviated the bulk of the unsubstantiated risk of subsequently missing out on the market when it hiked back up. 

Also, there were roughly 2 to 3 market timing opportunities per market index, during the 12 months period, which also is roughly normal particularly with the general uptilt in the markets then.  Each of these mathematical opportunities was worth about 3% or more, relative gain versus the market.  To repeat an important point here, all of these statistics above are in-line with long-run normal market gyrations.

In summary, it is not a good idea for the vast majority of individuals to market time, for any reason.  The process will be riddled with many hundreds of false summits over your lifetime.  Each one subjecting the individual to unneeded psychological and financial, wear and tear.  There will in fact be maybe a hundred or so correct market summits (a small fraction of the many exciting peaks that turn out to be disappointing false summits) that may acted upon.  Finding false bottoms is not the main concern, as the reason most people shouldn’t be market timing is only on the low probability associated with just correctly knowing where the market summit is.  The reason against market timing is not, as for example what Bogle sometimes only suggests, that a correct market-top picker would just not know when to get back into the market.  For the rare climbers who correctly know that they are at a mountain summit, we shouldn't go so far as to deny that it's there simply for fear they might now know where the bases are below them, before they get to zero altitude (i.e., ground level).  Market speculation is a treacherous and uncertain science, and there are more important societal issues to worry about.  Bottom picking should be correctly couched as a secondary issue, at best.

Now for unrelated updates, this is just to inform you that earlier this week published an Opinion article in the prestigious Royal Statistical Society.  Additionally am in the process of recruiting volunteer, community leaders to help advise on this year's forthcoming children’s statistics playsite.  Also this week have published the paper book, Statistics Topics, which was many years in the making and half the proceeds going to charity.  It is available on Amazon.  Details on all of these projects can be found on the main tabs of the Statistics Ideas blog.  Also elsewhere on the public web portal, we have some raw absolute and relative BRK performance data.  This just some low-level raw data has always been there for educational purposes, but it now being publicized in response to the recent new crush of interest, from major media outlets covering four continents, for any "latest" analysis and thinking on Buffett's record (particularly given his annual earnings conference in Omaha).  There is no probability analysis here, which are related to my recent advanced research that serves a much broader educational purpose (see the blog articles for that).  On the web portal above is just presentation of some raw data so that everyone can narrowly see the statistically growing pattern in BRK's underperformance severity, if they fairly consider all six charts shown.  The vast majority of the media have been great in their presentation of the information.  A tiny fraction have instead taken one or two small points on here, in order to advance a flawed position that BRK (still) greatly outperforms the benchmark through any accounting basis over any recent time frame- including the five-years as first noted on the blog back in February.  Lastly, exciting forthcoming research will be published on the mathematical analysis of liquidity risk, and health and traffic data, though the mediums for those reports may not be on this blog.