Just a Little Avalanche or an Implosion?

A few years ago, we briefly discussed the dynamics of sand piles in these pages, which are a special field of study in mathematics and physics (mathematically inclined readers can take a look at two papers on the subject here: ”Driving Sandpiles to Criticality and Beyond “ (PDF) and  ‘Games on Line Graphs and Sand Piles “(PDF) – unfortunately two other studies that used to be available have in the meantime disappeared from the inter-tubes).

Waiting to crumble: a giant sand dune in the Namib desert.

What makes sand piles interesting is the usually seamless transition from seeming stability to sudden collapse. Grains can be added to a sand pile one at a time for quite some time without disturbing its stability – it simply keeps growing. However, eventually a point is reached when one grain too many is added and an avalanche or even a complete collapse of the sand pile will ensue.

Despite their lack of purposive behavior, sand piles serve as a good metaphor for how asset bubbles usually end. Seemingly out of the blue, and with no clearly definable trigger event. All one can say with certainty is that instability within the pile (or bubble market) has increased over time, and that in the end, all it took was a superficially insignificant event or decision to trigger the reversal of the low-volatility growth everybody was becoming accustomed to.

In financial markets, this is usually preceded by a period of liquidity slowly but surely drying up as the monetary backdrop tightens and becomes more hostile. The worse the state of the economy’s pool of real funding is when this happens, the more profound the effects on financial markets and eventually the economy should be expected to be.

What concerns us here is the question of whether we have just seen a little “run-of-the-mill avalanche”, or if something more sinister lurks underneath the still well-entrenched complacency in the stock market.

Below is a chart comparing the DJIA, NDX and the NYA (roughly the “middling”, “strongest” and “weakest” performers). We have annotated a number of recent daily candles on the DJIA chart. With a hat tip to Robert Prechter of EWI, certain patterns and sequences of daily candles have been historically observed dead ahead of major panic selling episodes, and just such a sequence actually recurred before the beginning of this week.

Given the recovery over the past three trading days, the probability of a true panic sell-off has receded markedly, but it may not be completely off the table just yet. This is inter alia indicated by the continuing strong divergence between the broader market and “leading benchmarks” such as DJIA, SPX and NDX. Obviously, the more prominent averages and indexes are also the psychologically more important ones, so the focus should be on their performance.

Interestingly, the DJIA is currently in the technically strongest condition relative to the “pre-panic” pattern of daily candles, while NDX and NYA continue to look very vulnerable. Even the SPX looks not all that convincing, while the Russell 2000 Index (RUT) is an ongoing disaster area – albeit one that has received a strong boost over the past few days. The latter is usually a positive signal for the overall market, as RUT has been a leading indicator for both downside and upside moves for several years.

DJIA, NDX, and NYA: the annotations of daily candles in the DJIA show the developing potential panic cycle (h/t R. Prechter) and the subsequent deviation over the past three trading days. The action in the NDX looks not as good overall since its 200-day moving average remains an obstacle. The broad NYSE Index (NYA) not only continues to be the weakest performing major index, but its short term rebound attempt over recent days was quite lackluster as well.

The Monster Under the Bed

The next illustration shows the crashes of 1929 and 1987 next to each other, with the candle patterns alluded to above highlighted. As we always stress, the probability of such an enormous panic liquidation is extremely low. In fact, based only on the frequency of occurrences over the past 90 years in the US stock market, it stands at roughly 0.003%, which admittedly is somewhat less than peanuts.

Nevertheless, there are situations in which one has to at least consider the possibility, or rather, situations in which the re-emergence of known technical patterns demands some attention. Historically, crash-like moves – additional selling squalls that run opposite to the usual post-correction rebound – occur either within around 10% to 12% of an all-time high, or near the end of a bear market in a final “give up” move when the last hold-outs sell.

We very narrowly (and arbitrarily) define crashes from highs (relevant to the case) as declines in excess of 30% within 2 weeks or less from the market reaching an “oversold” level with RSI<30; this leaves us with just two examples in US equities since the 20^th century.

DJIA in 1929 and 1987 – these two panics were in many ways spitting images, and exhibited very similar pre-crash patterns involving a handful of crucial trading days and the associated short-term technical support violations.

It is of course not only the market’s position relative to its peak or the sequence of daily candles that provides a reason for pondering this. Absent the currently prevailing general fundamental and technical backdrop there would be nothing to discuss. Note that “technical” includes psychological data points, i.e., sentiment and positioning data.

The stock market is currently characterized by a high degree of overvaluation, an increasingly hostile monetary policy environment, a very long period of extremely one-sided bullish sentiment and positioning data readings, a broad range of weakening market internals with steadily widening deviations between sector returns (i.e., a breakdown in trend uniformity), large divergences across different markets (including foreign markets) and market segments, and a significant narrowing of market leadership with the bulk of index returns generated by an ever-shrinking  group of big caps. It is not one thing, it is actually hundreds of things (amusingly, it was recently reported that the public is finally completely convinced that stocks can only go higher).

We should also reiterate that the fact that the economy is doing well by most measures and that its performance is widely considered sound is not really relevant to the probability of a market panic taking hold. It may become relevant thereafter, depending on how sound economic conditions actually are. Since macro-economic aggregates in a credit expansion-driven bubble economy reveal nothing about the creeping capital consumption enabling the boom, it is a good bet that it is not as sound as most people seem to think.

The “Gruenderkrach” – a.k.a. the crash of 1873 at the Vienna Stock Exchange. This was the event that gave birth to the term “crash” to describe a stock market panic. “Krach” literally means “noise”  – evidently the normally staid and well-mannered gentlemen’s club where stocks were traded with the decorum appropriate to high finance suddenly erupted in a lot of unseemly noise. Good manners where quickly discarded when everybody tried to sell into a bidless abyss at the same time.

Panic Averted? Maybe Not.

Obviously, even if there is an increased probability of a panic sell-off, it remains very low indeed. As a rule, an initial decline to oversold levels accompanied by a growing number of divergences (in this case, positive ones) will lead to a sizable rebound. The rebound over recent days may well be sufficient to invalidate the panic sell-off scenario – even if one casts the “panic net” wider to include several of the less spectacular crashes and mini-crashes from the 1998 wobble to the heavy beatings administered in 2008 (2008 was a fairly unique sequence in any case).

A quick checklist says that time is, in fact, an essential element (when events become compressed in time, people don’t have enough time to think, but they will become fearful):  emerging panics are usually not interrupted by more than two or three days of counter-trend moves ending in positive closes in a row – which means that as soon as three or four clearly positive closes in succession are established, the immediate danger has most likely passed.

The fact that divergences warning of an upcoming low often present themselves at certain points is by the way not relevant to a panic selling wave per se. However, if a rebound that eats up some time ensues after they appear, a different scenario is very likely emerging (note that it could still be the beginning of a bear market).

Might there be reasons why the “three day threshold” may no longer be as meaningful as it once was? A few things do come to mind:

1. the global turn-of-the-month effect, which has become extremely entrenched as a persistent seasonal pattern over the past 20 years or so, and particularly over the past 10 years.
2. The stock with the largest market cap (AAPL), hitherto completely impervious to market weakness, held by practically everyone and avoided by shorts (its short interest ratio stands routinely below 1), is reporting on Nov 01.
3. Friday’s payrolls report is still widely held to be the most important economic report in terms of short term market-moving power, although this is actually no longer true.
4. The mid-term elections early next week represent an extra wrinkle: normally a seasonal “mid-term rally” would already be well underway by now. In fact, it should have begun almost exactly on the same day on which the recent sell-off started.

So far this year an almost perfect inversion of normal mid-term seasonality has been in play, but it is impossible to tell whether it can or will continue. The market may well re-synchronize with “normal” mid-term seasonality and rally sharply into the end of the year, which is presumably partly depending on expectations regarding the election outcome.

In 2018 to date, the market’s usual seasonal mid-term election pattern was almost perfectly inverted.

Complacency Remains Rife For No Good Reason

The next chart shows SPX and RUT and the RUT-SPX ratio in order to provide color to what was mentioned about them further above. We see a mild positive in the RUT’s tentative relative strength rebound, but not much more. Both indexes continue to look weak overall.

SPX, RUT and the RUT-SPX ratio: tentatively encouraging signs for an upturn, but ultimately not really much to write home about yet.

Lastly, here are a few technical indicators illustrating current market sentiment, as well as our trusted SPX New High-New Low Percentage Index (NHNLP). Neither can be called particularly comforting for the bullish case. It seems complacency remained rife throughout the sell-off and traders have quite eagerly and quickly embraced the rebound.

CBOE equity put-call volume ratio, the TRIN, and the VIX – this is the “you’ve got to be kidding” chart.

NHNLP – after the most recent sell signal, the target range indicating a solid oversold situation was once again not reached. The index has already returned back above the zero line – normally a buy signal. However, a level of 2 is not really convincing in that regard as yet. It continues to indicate an unhealthy new highs/ new lows dispersion and a single down day in the near term is bound to trigger the next sell signal.

Conclusion

There may not be a crash-like panic sell-off, given how rare these occurrences are – but we would continue to err on the side of caution. It just doesn’t sit well with us that there is so much nonchalance and complacency in evidence amid a technical picture that just doesn’t look very good at the moment. Better avoid getting buried if the sand pile does decide to crumble.

Charts by: StockCharts, ShareLynx, Seasonax