Weekly Sentiment Brief
Using the word "brief" in the heading is false advertising - these will generally be about the same length as this version, with roughly the same amount of waffle.
Sentiment Analysis is something I consider vitally important, particularly when sentiment reaches significant extremes (which I think has already happened).
It has to be stated up-front that I've got little sympathy for "one liners" - those people who just look at one chart, one timeframe, with indicators-of-choice based thereon. The market behaves in highly predictable ways, but you have to look at more than just a 1-minute or 5-minute (or hourly) chart. For example my personal preference for intraday trading is an 85-tick chart, but if I relied on that timescale alone, I would take far too many counter-trend trades - and they are a recipe for heartburn.
I will only use the word "fractal" once in these missives (I just used it... never again) but it's clear that when the market is biased in a given direction on timescales one and two levels above the timescale that you trade, there is little benefit (and much cost) in trying to "fade" the trend. That is, until the market is overcooked.
The "overcooked" stuff is a joy to behold, but a pain in the bum to "time". As a result, I present the first of "GT's Sentiment Axioms" -
Axiom the First:
Never - ever - trade the first instance of a sentiment extreme on the same timescale.
Let's say that during an uptrend, a market hits overbought (and/or sentiment looks excessively bullish) on a daily chart. Traders who trade holding periods measured in days might be tempted to throw down short, trying to "pick a top". That's fine so long as you're prepared to withstand what might be a considerable period of discomfort.
Folks like me will use that daily condition to help form our intraday biases - looking for intraday shorts that may develop into holding-period shorts. So we will look for short entries on intraday overbought readings of some or other oscillator - my favourite is a smoothed version of WIlliams' %R, because I can understand what it means.
This works especially well if the overbought oscillator reading occurs at or near an intraday sentiment extreme - that is, excessively bullish market breadth (very low levels of TRIN, very high levels of TICK, and a very strong intraday Advance-Decline reading). On very, very rare occasions you will also get "mixed markets" at the same time, where one index is performing OK but a second index appears to be faltering (e.g., the S&P up with it's A/D line healthy, but the Nasdaq100 weak with its A/D line below the waterline).
If the ducks line up that well, thank your Deity of choice, because you've probably lucked into a top that will hold for a week or more.
OK... now onto the actual stuff for this week...
US Equities
Index review
All of the major US indices made new lows for the year last Thursday, after registering (in late December) overbought conditions on a smoothed %R (not shown - charts show native %R) and obvious daily divergences - like that shown in red below - on oscillators like the price Rate of Change (ROC). The Dow chart is reasonably informative, and on first blush it looks like the market may give back the bulk of the November spike above the downtrend channel shown on the chart.
That's all well and good, but the Dow chart is now oversold, as are all other index daily charts; weekly charts have only worked off part of an overbought condition - thus the most likely outcome on a daily basis is a short term bounce that works off part of the oversold condition then re-engages to the downside.
The daily chart of my "Dow Theory-ish" ratio indicator - which plots the ratio {Dow Jones Industrial Average/Dow Jones Transports} was at extreme Dow-overbought levels for most of 2004, as the Transports took off to make new highs (which were not confirmed by the Dow Industrials). More on that later.
Now as I've said before, genuine Dow Theory uses the Industrials as its "core" index and the Transports as a confirmation index. To speak of a move in the Transports as being "not confirmed" by the Industrials is not really a valid Dow Theory expression - it is postulating the behaviour of the horse based on movement in the cart. That said, if the cart is pushing the horse, the horse can't slow down too much without a requirement for a visit to a veterinary proctologist.
Suffice it to say that for most of 2004 the Transports were making higher highs and higher lows (an uptrend) while the Dow Insutrials was sideways at best (and sideways to down if truth be told). The ROC divergence that presents itself on the Dow chart actually started to form earlier on the Trannies - another example of the cart leading the horse. The Trannies are now more deeply oversold, too - whic his further justification for the "bounce then decline hard" hypothesis.
All the while, other ratio indicators showed that the S&P was outperforming the Dow, and from mid-Arpil; the Nasdaq100 was outperforming the S&P500; that's 2 out of 3 for a "flight to shite" indicator Although Meatloaf might say "Don't be sad, coz 2 out 3 ain't bad", that's not how this game works. One key plank was missing: Semiconductors were underperforming all other indices, and continue to do so.
Now there are two sets of monstrously overpriced stocks in the US markets; Semiconductors and Internets. Nobody in their right mind even looks at a chart of the internet stock index (although it went nutso post-November). The Semiconductors - although monstrously overpriced - still get valued (in part) based on sensible evaluation of their actual business (albeit with ludicrous permissible multiples) rather than stupid "click and eyeball" valuations of internet stocks, which persist to this day. So although I keep a weather eye on Internets, the SOX is my "flight to shite" indicator. It's been languishing for most of the recent advance - evidence that although dumb money was being thrown at internets, the big money was not chasing b as much as the market advance seemed to suggest.
With all of the foregoing in mind, the very basic technical structure appears to favour a bounce this week, that should exhaust before it hits overbought. Trading days are likely to exhibit early strength that fades toward the close (especially if significant repurchase dough is not forthcoming from Uncle Easy). Earnings announcement spikes (i.e., when bellwethers release "better than expected" earnings) should be opportunities to sell intraday on post-announcement overbought indicators.
And now the Extended Remix...
Sentiment Extensions...
Although the markets have been weak recently, there is still no shortage of stocks on both exchanges that show bullish point-and-figure configurations. The NYSE Bullish Percent Index (BPI) is at levels that one would expect at interim market tops, as is the Nasdaq100 BPI.
Likewise, the Short Interest Ratio for the Nasdaq (the proportion of all Nasdaq free float that is sold short) fell to its lowest level in 6 months in December04, down to 2.6% (the level at the beginning of 2004). The SRI rose consistently during the first half of 2004 (as the markets declined) reaching a high of 4% at the end of August ... which pretty much coincided with the low of the year (now do you see why I'm a contrarian???). By the end of 2004 the short interest ratio had fallen back to its level at the beginning of the year even as the market rose to exceed its January 2004 level.
What does that say about the extent to which short covering was driving the post-election rally? What does it say about the bearishness of short sellers? (Hint: it says that most shorts were late and wrong, and that their bearishness evaporated the moment the torch was applied.)
Commitment of Traders - Dumb Money Still At Work
While we're talking about dumb money, let's take a sneak peek at what small specs ("non-reportable" positions) are doing in the latest Commitment of Traders report from the Commodity Futures Trading Commission (CFTC).
I have a rule for evaluating data - if a number is produced in the "headline table", then every dolt on the planet can see it, so it's useless.
If a number is in a "non-headline" table, only 5% of people (usually folks who get paid to be right, and ZERO percent of people who write for CNBC) will read it. If any further computation is required (say, the use of a "divided by" key), you can throw out another 4.5% of your original sample.
What I'm getting at is that just as I don't look at headline numbers when it comes to Industrial Production, CPI or Durable Goods data, nor do I just look at dopey numbers like the level of net short or long open interest held by Non-Reportables. It's not really even adequate as a "first cut", and two things are far better as an indicator of "Dumb Money at Work":
- the proportion of Open Interest held Long by Non-Reportables; and
- the Ratio of Non-Reportable Long Positions to Non-Reportable Short positions (I call this the "Dumb Bull Ratio").
So when you get more-than-usual amounts of OI being held by dummies, and the dummies are overwhelmingly bullish, you know that the smart money (both commercials and Large Specs) are positioning for a decline.
At swing lows - lows that are properly tradeable from the long side - the Dumb Bull Ratio (Non-reportable Longs divided by Non-reportable shorts) will be a number less than one. At swing highs, it will be a number like 3 (or higher - the higher the better, for shorts).
So what do we see at present?
At present it's just a tad under 4 - down from a high of 6.6 in mid-December 04 - which was the first decent sign to start looking for the short side.
To give a better historical perspective on jsut how wrong the nuffies are at turning points, the Dumb Bull Ratio hit a low of 0.33 in March of 2003, which was the low of the year. It (the ratio) then re-tested that low in August - thus making the nuffnuffs net short right at the bottom of the sole pullback for 2003, which ended in August - after they had been getting more and more net long during the rally).
By the next US Indices instalment of this newsletter-ish-thingo, I will have set up RantPro's own charting appplications so that I don't have to buggerise around in Excel trying to get something presentable (that's why I haven't shown anything related to the Bull Ratio stuff today).
