Desolated admirers of investment strategists

Shared by leading news prints such as The New York Times, as well as famous investors.  As a reminder, being among our 200 thousand Statistical Ideas followers is easy to do, through e-mail, @salilstatistics, facebook, LinkedIn, Instagram, YouTube. 

A year-end market ritual two decades in the making: celebrated Wall Street seers promote their “forward looking” stock prices.  But as you may know, this group (either individually or as a consensus average), we earlier this year proved to perform worse than a coin toss.  Now they have again obliged us today, by misguiding investors and ordinary Americans with more random information that they claim emulates the future.  Our prequel article, strategists: full of bull, was among of most popular blog articles ever (>1 million reads and a thousand shares) and went so far as being the cover and Most Popular story on Wall Street Journal’s MW, in addition to being written up in dozens of global news.  We expect this sequel article to complement that (already featured in this Sunday’s New York Times print and Forbes), covering needed new math ground concerning what we learn here at the end of the year looking back is just as valuable as what we do at the start of the year looking forward.  We also complete our 20^th year of forecasts in our database that we make free for the public to explore; it is the largest such data set in the world.  Last, we discuss the accuracy of this and any year’s forecast and the probability of these Wall Street strategists to have achieved their results by skill versus chance.  This in turns aids in our understanding of how valuable such forecasts are for those who take the time to immerse themselves in it.  Please read the prequel articles, as well at this sequel below.  And you’ll discover the robust probability insight into exactly how unwise these celebrity seers truly are.


2017: a continuation of failure
We start by gratuitously fulfilling the wholly unnecessary curiosity of those who want to know what the 2017 market pundits are up to, released today by Barron’s.

Forecaster	2017 target	% change from today’s level
Stephen Auth, Federated	2350	4.0%
Jon Glionna, Barclays	2400	6.1%
Jeff Knight, Columbia	2450	8.2%
David Kostin, Goldman Sachs	2300	1.8%
Dubravko L-B, JPMorgan	2400	6.1%
Tobias Levkovich, Citi	2325	2.9%
Adam Parker, MS	2300	1.8%
John Praveen, Prudential	2575	13.1%
Heidi Richardson, BlackRock	2400	6.1%
Savita Subramanian, BofA ML	2300	1.8%
average (consensus)	N.M.	5.2%

From a probability and statistics perspective, it is not interesting to explore the consensus average price level (so shown as Not Meaningful in the table above), but rather just the consensus % change.  And of course, on a continuous-geometric basis.  These predicted % changes shown in today’s Barron’s are likely a little less than what the strategists’ originally envisaged, given the quick upward bounce in prices, occurring in the background as these forecasts were developed. 

The other most important thing anyway for you to see here is -that among only 10 strategists and despite the lower bias described from fast moving prices during forecast formation- none of them still hold a negative view for 2017!  And therein lies a prodigious irony, since nearly 1/3 of the years -in the past two decades- have been negative (with average return in those years of -18%).  Think about that for a moment as you consider the risk of the stock market.  And that’s more negative years than most people, including these self-described “investment cognoscenti”, expect. 

What’s troubling in relation to that is this in those same nearly 2 decades of forecast data, only 8% of individual projections were negative.  This clearly shows how bullish-biased they are (as if a typical 9% annual prediction versus a 4.5% actual return wasn’t enough!), but the probability information given will still take this dismaying performance to the higher level.

We should appreciate that an investor is likely to be less concerned about Wall Street forecasters being conservative, as they will be concerned about being told that the markets will simply rise double digits’ percent (and then instead be treated to a decline of roughly that same amount!)  And yes, that happens. 

So we know there is some relevant asymmetry in what investors righteously care about.  Even without asking for this explicitly, investors are seeking to manage this frank risk with professional insights.  Though no insights are to be found.  As a case in point, in the 1/3 of all the years coinciding to when the markets dropped, take a guess at how many of those years did the consensus price estimate also show a drop? 

A spectacular ZERO. 

Demonstrating both no accuracy, and no value in the advice!  And yes, you would always be better off simply ignoring them, as you should ignore many unsolicited salespeople in your life.  Instead of the consensus, drilling in on the individual forecasters’ records -during those down years- we see that only 9% of those forecasts were negative.  So the equivalent rubbish as the 8% overall negative forecast probability shown in the entire data set (not necessarily conditional on just the 1/3 of years that happen to be negative) 

Now imagine having a coin calibrated to show “above average” 2/3 of the time, and “below average” 1/3 of the time.  Flipping this coin would therefore outperform a Wall Street strategist!  What’s worse is that simply guessing a positive or a negative for a year is an arbitrarily meaningless, severity test.  Sure it has some minor risk views we can examine. 

But what’s more interesting is to know that if we parameterize this by simply looking at the actual correlation between the strategists’ predicted level and the actual level, that relationship is negative!  This distressing result makes the variance in forecast error of 21%, which is improperly higher than the raw market’s standard deviation of 19%!

The probability portrait we have here is very depraved, and actually tends to get worse during the traumatic times when you actually require the strategists to provide some insight (for example before a large market swing!)  It would be bad enough if all strategists just put out a constant 9% forecast each year, which of course is twice the annual return over the past 2 decades. 

Instead what we showed in the prequel article is that the Wall Street strategists tend to gyrate about this 9% typical forecast in the least helpful ways (e.g., stating +11% in 2008 when it was instead -49%).  See forecasts in sky blue above, and the actual returns in pink.  To be clear, and as we’ll show again below, it is far better to be within 10% of a -49% market year, then it is to be bulls-eye when the market happens to be a more normal +9%.

Busted clock, true once daily

We all know that a “stopped” clock is right twice daily.  What’s lesser known is that a clock, showing a random time each minute, is also typically correct twice daily!  In that sense only, both these clocks types appear to offer the same value. 

But the latter clock is actually a little worse than the former clock, when we consider the uncertainty in how much error there can be (between the clock’s result and the actual time).  The first clock in this analogy is a perceived “dumb” passive investor.  And the second clock here would be a speculating “monkey”.  And we’re about to see an important life lesson:

That it is better to be smart while being perceived as dumb, than it is to be dumb while being perceived to be smart.

Now mapping this random clock analogy to our investors, we can think of the clock as being an unfair coin that results in “up” 2/3 of the time and results in “down” 1/3 of the time.  On the other hand, the professional forecaster only predicts the market results far less than half of this (8% individually, and 0% on a consensus-basis.)  So these Wall Street soothsayers are the equivalent of a clock that is -by chance- only right once daily!  This is appalling performance, and statistically significant in being worse than a coin flip or a speculating monkey.  Certainly worse than a passive investor who can merely read the financial data, in an unbiased way.

Of course these active scammers below don’t want to hear such aggressive nonsense so they offer you this scripted rebuttal to justify their fees: “we might not get the actual S&P price target right, but we offer clients better value in getting the risk factors right”.  They confuse a confused and fee-paying client, with their ability to add any value. 

They cloak their terrible audited calls with other smoke.  Such as randomly pitching you on the idea of large versus small companies, which industries, or which risk factors.  They’re playing roulette with your money, their investment returns prove it, and The House wins all the time by draining fees. 

The defenses of Wall Street strategists are the equivalent of stating one can’t offer you a great box of chocolates, but they can offer you individual chocolates in a box, which are all great.  Remember the market is a complete fusion of all factors (such as all industries and company sizes), so it’s impossible to know a subset’s relative performance well if you don’t know how it will perform in relation to all the others. 

And strategists collectively don’t know this either, if they can’t in parallel get the overall market call right.  See these smug zoots acting sage, but with just a fantasy 9%, sitting between their ears.

Now with over 20 years of forecasts (19 years of actual performance) we are better able to judge the quality of the forecasters and apply a sense for, probabilistically speaking, how do they contrast with a “random clock”.  Using the following confidence interval formula estimate here, we can see that the standard deviation of about the actual 1/3 market probability drop.  The formula works even if the markets are randomly bouncing around each year, and strategist guesses don’t fall into up or down portions so easily.

So when they project negative years only 8% of the time instead of 1/3, they are demonstrating that they are greater than a standard deviations worse than a monkey, or a coin-flipper, or whatever mindless analogy one wants to use.  While this alone shows that strategists have been definitely offering skill in losing money, as we’ll explore below, there is another similar math technique covering this same point.

The Emperor hopes you forget…

… that you saw him wear no clothes.  Wall street strategists happily talk about “price targets” as though this concept has any value.  They falsely make the process appear scientific by providing a dash of “fundamental” economic data to help substantiate their work.  But then they grotesquely gravy over this with a “price multiple” swag that entirely invalidates any underlying reasoning.  For this useless guess, that hope you’re imprudent enough to pay them large sums of your hard-earned money to be told simply what you want.

Let’s take 2016 for example, seeing the chart below.  Once again we had the typical 9% forecast, and the forecasts were relatively tight among strategists.  And none of the 10 had a negative forecast for 2016.  But within just weeks of this year they were dumbfounded by the deep market crash that happened as they were putting together their imaginary forecasts.  No sooner did they return from their island vacations, were they confused how to once again burned clients since they were boxed in with their typical 2200 S&P forecast (the 10 forecasters ranged from 2100, to 2500).  Yet the market fell below 1830 (or less than everyone’s 2016, 2015, and 2014 forecasts!)

In typical knee-jerk reaction, most of these strategists impulsively just dropped their year-end “forecast” in early 2016, as shown as you follow the red line above.  Sure enough though, when the market then rebounded and advanced through the summer, most of them again were disbelieving it and revised this same year-end “forecast” again (this time up by just less than 40 points, to 2138).  The market is currently at 2258; 5% higher than their most recent call of 2138 (or 3% higher than the original 2200 call).  Even the very reasons they gave for being concerned in 2016 turned out to be false: no one mentioned the January crash, nor Brexit, and they certainly got every aspect of Donald Trump wrong. 

Someone needs to give the Editor at Barron’s a lecture that the term “forecast” starts with the prefix “fore”, suggesting something in advance.  This should not be turned into a courteous second-guessing backcast, which means look at the random stock pattern and state that you brilliantly knew this would happen with your money.  As a reminder these cumulative changes over time can be quite alarming, and of course generally uselessly bullish even to the most timorous marketers (as shown here through 2015). 

But because we see these wild gyrations in their forecast, and over such a short period of time, this shows the Emperor was wearing no clothes, at all.  And we also notice something disheartening or comical - depending on your perspective.  The times they tend to make a judgement about the relative change in the markets (outside their typical 9%), the market generally then moves in the opposite direction, and sometimes on bullish calls the market can instead drop significantly (we noted the 2008 example earlier, but there are a handful of such examples). 

Strategists are the vilest types of fortune tellers, since they completely miss the necessary downturns in the market, which can wipe out years of your hard-earned savings.  And that’s what some experienced so easily, earlier this year.

During such downturns we see in the chart above that they also didn’t advise clients to stay the course, since they themselves dropped their forecast and led people to believe the markets would then only return a lower 8% return (projected in late February), when it would instead rise 15% since then.  A couple mathematical issues make Wall Street forecasts terrible, as we can so finely see this year.  In addition to the actual difference between the predicted level and the actual level (this is the focus of many simple critiques on strategists). 

The couple ideas here are that there is still wild dispersion in the analyst outcomes such that even if the consensus is correct, the larger dispersion in forecasts are less attractive in judging the analysts’ accuracy.  And their actual dispersion has zero relationship to anything meaningful.  For example, say we have two back-to-back years where the market returned 4%:

Year	Strategist 1	Strategist 2	Consensus error	Dispersion of error
1	4%	14%	-5%	5%
2	9%	9%	-5%	0%

Even though both years show -5% forecasts from a consensus average perspective, the 2^nd year is clearly the more accurate year for the typical strategists.  The second issue is say we have a 3^rd year where the market crashes to -25%.

Year	Strategist 1	Strategist 2	Consensus error	Dispersion of error
1	4%	14%	-5%	5%
2	9%	9%	-5%	0%
3	-15%	-15%	-10%	0%

So while year 3 has a worse consensus error versus year 2, the fact is it is far more important to have strategists who are closer when the market moves violently, rather than be more close when the market barely budges from its predicted value.  This year is reflective of what we typically see from strategists, who provided insincerely “safe” bullish calls, which appear more correct on a consensus basis, but isn’t necessarily tight nor useful for when markets eventually crash badly.  Imagine the 4% and 15% strategists staying at those levels in each of the 3 years above.

Using this discussion, we should note that none of the strategists rise to the level of providing value, skilled or luck based, though some are quite skilled at detracting value.  Take the Bears Stearns’ strategist whose terrifying forecast accuracy is shown below and are the epitome of the revolting surprises that await investors who use Wall Street “insight”.

For the overall industry of strategists, see this chart below that shows how each year’s market returns contrasted against that year’s consensus forecast.  And the size of the data shows the dispersion in errors (also colored red for large and blue for small). 

What we see from this is that 2016 (in dark green) was an ok year relative to the original (2200 S&P) forecasts Wall Street firms put out 1 year ago.  Though one can see in the red composite that this year was already a typical consensus forecast year of about 9%.  And of course -in typical depressing fashion- these forecasters then lowered those estimates at the wrong time and in a sense shot themselves in the foot again.  These are lessons you should only be reading about in a web log, and not experience with your nest egg. 

Be careful in 2017.  With reduced yet positive Wall Street forecasts, you might neglect the exceptional false sense of accuracy that comes with it, particularly this year as the bull rally has extended without a meaningful pullback in some time.  Implying room for unusually great uncertainty, in the early part of the year.