Contradictions Do Not Exist (Was May-15 Really a “Major” Top?)
“Contradictions do not exist. Whenever you think that you are facing a contradiction, check your premises. You will find that one of them is wrong.”
— Ayn Rand, Atlas Shrugged
As intensely volatile – and often times scary – as the past year has been, punctuated most recently by Friday’s shocking reverberations from the UK’s exit from the EU, US equities have remained remarkably resilient. Indeed, even with Friday’s white-knuckled thrill ride through this, the DJI and SPX are only ~400-500 bps off their all-time highs which were printed nearly 14 months ago in May-15.
That proximity to May-15’s all-time high is spectacular considering some of the events that have been sandwiched in b/t then and now including the aforementioned BREXIT on Friday, a collapse in crude and EMs driven by Obama’s financial warfare against Russia, Aug-15’s CNH deval and capital flow issues, all of which many have tried to represent at one time or another over the past year as the First Angel of Seven summoning its trumpet with a calamitous GFC-redux sure to follow.
And yet the flip side of this resiliency coin is the reality that US equities have also been mired in a stubbornly narrow trading range that has flirted with and failed to break above that aforementioned May-15 high on numerous occasions over the period in question.
Ice cream cones for both bulls and bears have abounded, but sprinkles have been far tougher to come by.
And though we now live in a world where one has to his or her avail a nearly infinite cornucopia of easily accessible data points that they can use as inputs to their speculative mosaic-building process, for me, as I try to objectively handicap the likely directional bias of the next fat pitch served up by the market, nothing has been more revealing than the chart pattern the DJI and SPX put in place as they were peaking in May-15.
In simply subjectively eye-balling the appearance of over 100 years worth of major (defined below) tops in US equities, not a single one has ever resembled the pattern put in place of the alleged (if one is a bear) “major top” from May-15. Can anybody find me a major top in 100 years of trading data that looks anything like the wimpy, rounded one the DJI and SPX put in place last year, especially coming off a nearly unprecedented six-year sprint of gains such as the one that culminated in May-15?
Based on simply appearance alone, aren’t major tops supposed to convey ineluctable signs of speculative excess with major blow-off peaks?
But as I said above, we’re trying to objectively assess whether or not May-15 represented a major top. My attempt to answer that question in persuasive fashion by presenting two simple charts is not only subjective, but unconvincingly so at that.
However, the beauty of markets is that virtually every question can be answered in such fashion through the back-testing of data.
So, if major market tops are supposed to convey exhaustion points of speculative excess and momentum, it would make sense that all of the previous ones in history that went on to generate material draw-downs – let’s say a minimum of 20% on a weekly closing basis – would share the same type of finger-prints highlighting as much.
For instance, a major blow-off top should theoretically trade a good deal of %-points higher than the average price of the past one year/ 52 weeks, correct?
Thus, let’s look at every single instance of the DJI (more trading history than SPX) registering at least a 20% peak-trough decline on a weekly closing basis off a major high and find out how far in %-terms the peak price that began the decline was above the market’s trailing 52-week average price.
I count 22 such declines of at least 20% – or, differently, 22 major tops – dating back to the late 1800s. Here’s a collection of charts that depict every single one across multiple generations for those that care to continue eye-balling how most historical tops appear vs. 2015’s appearance.
1960s-1970s (none in 1950s)
Importantly, across 100+ years of data in the charts above, the market stood, on average, nearly 15% above its trailing 52-week average price as it put a major top in place.
Even more importantly, across 22 major bear markets, 7% was the single-lowest percentage by which the price at a major top exceeded its trailing 52-week average price.
By contrast, the DJI’s 5/15/15 weekly closing price of 18,273 stood a mere 5% above its trailing 52-week average price.
In other words, in confirming my subjective eye-balling of 100+ years of topping patterns, if the May-15 high for US equities was indeed a major top, it will have marked the single least speculative one in history when assessing all previous ones under this objective standard.
Let’s run another cross-check against everything above by comparing the paths the DJI took off its 1961, 1966 and 1976 highs in the table above relative to the one it’s taken off its May-15 high so far. These are the only three major tops in history that were even remotely as “non-speculative” into those highs as May-15 was (i.e., the 1961, 1966 and 1976 tops were 7.4%, 7.1% and 7.0%, respectively, above their trailing 52-week average price vs. the 5.0% registered at 2015’s high).
In doing so in the analog below we’ll find that even these tops aren’t very comparable to the path the DJI has taken off the May-15 high so far. For instance, the paths off the 1961 (blue) and 1966 highs were much weaker at the outset of the decline as both fell ~25% peak-trough through the first 175 trading days post-high; by contrast, off the May-15 high into Aug-15’s lows the market only fell ~15%. Further, the other high from 1976 (green), while it was more resilient early on vs. 1961 and 1966, was much weaker at the current juncture than we are now, as it was down 20%+ at this point vs. ~5% this time.
Here’s what the path off the May-15 high looks like vs. the average of the 1961, 1966 and 1976 post-high paths in the analog above. Again, the post-May-15 high path looks nothing like the average as it’s displayed far too much strength at nearly every juncture.
We can also run another cross-check on a time-related basis. Meaning, the analysis above only looks at how tops have historically formed on a price-related basis.
If we think about those tops differently, on average, how long did it take each of them to register their first 20% decline?
Bear markets typically take little time to get rolling, so presumably history will reveal that the 22 bears in the table above all managed to register a 20% decline fairly quickly.
On average, historical bear markets take only 33 weeks to register their first 20% decline on a weekly closing basis. Moreover, only two out of 22 bears in history have taken longer to register a 20% decline than the current cycle, assuming it is indeed a bear, as 58 weeks have elapsed off the May-15 high without such a decline.
As it stands, only the 1906 top, which ultimately turned into the 1907 Banker’s Crisis, and the 2000 top, which turned into the tech bear (and both of which were peak-trough 45%-50% bears!), took longer to register their first 20% declines relative to the current move, at 60 and 62 weeks, respectively.
Thus, the current 58 week period without a 20% decline is a major historical outlier (just as the % its peak price was above the trailing 52-week average price which had zero precedents) with only two precedents that exceed it. However, despite their being two precedents in this analysis compared to zero in the first, as the table shows, both the 1906 and 2000 tops saw their peak price vastly exceed their trailing 52-week averages, by 26% and 11%, respectively, well above the 5% level at the May-15 high.
In addition, if we look at what the SPX was doing at its 2000 high instead of the DJI, we’d find that its peak weekly close of 1,527 during the week of 3/24/00 was nearly 12% > its trailing 52-week average price and it took the index only 51 weeks to register its first 20% decline (the SPX peaked nearly two months after the DJI in 2000); thus, the SPX’s action was very different at the 2000 top on both a price and time basis compared to the May-15 high, if even the DJI in 2000 shares a time precedent with the May-15 high.
Regardless, let’s analog how the move off the May-15 high looks compared to the paths the DJI took off its 1906 and 2000 tops to see if we can find any similarities.
These are actually fairly similar in my mind; much more similar than the way the post-2015 path shapes up relative to the earlier analogs in this post. That said, the move off the May-15 high was much stronger off the top than both 1906 and 2000, was stronger at the highs of the initial rally and was much stronger throughout the second rally that we apparently just exited.
However, as the table suggests, both of these moves finally registered a 20% decline heading into weeks 60-62, or, as the analog reveals in the gray shade, the precise juncture off the high where the the current cycle is just now entering.
Therefore, if one believes this analysis more so than the one that showed the May-15 high was barely above its 52-week average price, the market is entering a critical juncture that could spell doom without immediate and renewed buying interest. In fact, this may be one of the juiciest parts to begin shorting equities if this analysis is accurate.
In pulling it all together, here is a scatter plot that shows each of history’s 22 previous bears with the x-axis representing the % by which each cycle’s peak price exceeded its trailing 52-week average price and the y-axis representing the number of weeks it took off the cycle peak to generate a 20% decline. The green marker is the 5% and 58 weeks that correspond to the May-15 high. It is important to emphasize how much of a historical outlier this top would be across both the price and time frames if it were a major top.
As Rand suggested, contradictions do not exist in the world.
As such, the idea that May-15 marked a major top and the chart pattern into and after that top would have failed to produce any resemblance to other major tops in history – both on subjective appearance and objective statistical grounds – is a contradiction in terms whose basic premises fail to pass the scrutiny of 100+ years of data.
Accordingly, if the market cannot register a material decline in the next two weeks (the 1906 and 2000 examples above are arguably the best and only precedents for this to occur), I believe the likelihood of a 20%+ draw-down occurring on a weekly closing basis thereafter will become the lowest-probability outcome, and our separate work calling for a pivot into fresh, all-time highs and a sustained rally thereafter the highest-probability outcome.
If such a decline begins to hasten itself in the near-term it likely becomes a reasonable bet that the May-15 highs ultimately morph into a 45%-50% bear market, with the 1906’s Banker’s Panic likely playing the muse for a similar panic in Euro financials this time around.