"Positive" Divergence of New Lows?

One statistical divergence I’ve seen some discussion of lately is the smaller number of new lows compared to the January bottom. Theory says that this is a positive breadth indication. Since less stocks are posting new lows, less stocks are in poor technical shape. Hence, although the price level of the observed index is near or below the previous swing low, the makeup of the market is improved. Supposedly this has bullish connotations looking forward.

Proponents of this kind of analysis can easily point to some instances where the divergence seemed to work beautifully. One nice looking example would be August 2004 bottom. The S&P 500 poked beneath the May lows but NYSE New Lows contracted. The market put in a nice rally after that.

I ran a test to see if a contraction of new lows on a swing lower for the S&P 500 was predictive of a rally. Basically I looked for the SPX to make a 100 day low while the highest number of new lows in the last 100 days was greater than the highest number of new lows in the last 10 days. The trade entry point for the study was above the prior days high and the exit was 20 days later. Going back to 1992 I found 10 instances. I’ve listed them below.

It appears to me this divergence worked well during bull markets (98, 99, 04, 06) and not well during the bear market of 2000 – 2002. Success would therefore seem to be attributable to factors other than the divergence.

October ’98 and October ’02 launched some very strong 1-month moves and it’s interesting that the divergence was in place at those times. Based on the magnitude of success of some of these rallies the divergence may therefore be notable. As a stand alone indicator I was unable to find predictive value using my fairly simple test.

CBI to 15 – Elite Territory Now

The Capitulative Breadth Indicator (CBI) spiked to 15 today. This is the highest reading it’s posted since I began tracking it live in 2005. Backtests to 1995 show only 3 other instances when the market was closing at new lows and the CBI was as high as 15. They occurred in August of 1998, September of 2001, and July of 2002. All three instances eventually shot up to mid-30’s or higher. The highest CBI reading was July 2002 at 52. Needles to say, we’re hitting some pretty elite levels here and those other selloffs were about as scary as they come.

The 2001 and 2002 charts are posted in my Jan 22nd post “How Bad Can It Get?”. The 1998 chart is below.

I’d suggest traders review their own charts of those periods as well to better formulate a game plan they’ll be comfortable executing. Those selloffs were incredibly scary. There was also incredible opportunity for short-term profit at those bottoms. The market is going to bounce. I believe it is going to bounce very sharply. It may sell off a significant amount price-wise before that bounce occurs. Time-wise I still believe we’re only days away from a multi-week bottom. I’m currently on the lookout for a washout day or a reversal day that could mark the low.

Does Friday’s Drop In Fear Matter?

The market continued to fall on Friday and the S&P 500 posted another new closing low. Several fear gauges showed less extreme readings than Thursday, though. Participants seemed to take the selloff in stride. For example, while still relatively high, the VXO, VIX, CBOE Total Put/Call and CBOE Equity Put/Call ratios all dropped.

I ran some tests to see whether the lower fear levels occurring on a day where the S&P 500 made a substantial low indicated more selling was likely to come. For my tests I used the VXO. The conditions I layed out were 1) The S&P 500 must close at a “x” day closing low. 2) The VXO must close lower than yesterday. 3) The VXO must close at least 10% above its 10-period moving average. The study seemed to indicate that extended VXO levels – even if they were less extended than the day before, had a positive impact on returns over the next week or two.

Below are the results when looking for a 20bar low in the SPX.

The fact that the VXO is stretched appears to be more significant than the fact that it pulled back.

Moving the bar out to a 100 bar low for the SPX gave these results.

In short, Friday’s slightly reduced VXO levels don’t seem to have any negative connotations.

In my post on Friday I discussed the high put/call ratios. For some other interesting perspective on these numbers, check out the links below:
Traderfeed – An interesting study and indicator to add to your p/c charts.
Daily Options Report – Some thoughtful musings on p/c related to volatility.
VIX and More – A long term view of the equity p/c ratio.

CBI = 11. How Long Before The Bounce Begins?

My Capitulative Breadth Indicator (CBI) moved up to 11 on Friday. In the past, readings above 10 have signified broad capitulative action in individual large-cap components. It has been a reliable indicator of exhaustive movement and has frequently been followed by a substantial short-term bounce in the S&P 500. Readings above 10 have been a fairly rare occurrence with only 16 instances in 13 years between 1995 and 2007. They’ve become more frequent recently, though. This is the third time the indicator has moved to 10 or higher in the last 5 months.

As I’ve written in the past, buying the S&P 500 on a close in the CBI of 10 or higher and then selling when the indicator closes back at 3 or lower would have been profitable 100% of the time. The indicator is currently 17 for 17. This does not mean it will work this time. It also does not mean there won’t be significant downside before the expected bounce ensues.

In a post on January 22nd, “CBI Spiking – How Bad Can It Get?” I looked at the two worst selloffs from a percentage standpoint that occurred after a 10+ CBI reading. I’d suggest reviewing it.

I ran a scan tonight to see what the longest period of time was before a bounce ensued under the following conditions: 1) The SPX is trading at a 50-day low. 2) The CBI moves up to 10 or higher. Seven of the past 17 CBI spikes have occurred when then SPX is at least at a 50-day low. Of these seven times, the longest the market has continued to drop before beginning a substantial bounce was 3 ½ days. The lowest close was 3 days later. The lowest low was 4 days later. Interestingly, these were the July 2002 and September 2001 instances listed in the “How Bad Can It Get?” study.

While the market could fall substantially over the next few days, the CBI is suggesting a multi-week low should be made by the end of the week.

CBI Hits 8 but I’m Taking It Slow

The CBI rose once again today from 6 to 8. It has now passed the threshold of 7 where I normally begin taking on index exposure. I’m wary at this point because although individual stocks are extended, the index itself isn’t. The SPY is just now starting to break down out of its recent range. It lies about 2.5% above the intraday January lows and there’s a chance of cascading action from here. Spikes to 7 or higher are more typically accompanied by a mature selloff – not a fresh breakdown, hence the reason for my caution. A reading of ten could easily occur as early as tomorrow. If it does I will then likely begin to scale into the index trades with higher allocation.

The CBOE put/call ratios gave some interesting readings today as well. The equity-only put/call ratio, which has been tracked since October of 2003, posted a reading of 1.12 on Thursday. There have only been two other days with reading this high – June 14, 2004 and August 6, 2004. The June reading led to a brief bounce that lasted 7 days before the market started its next leg lower. The August reading led to a two day bounce. The low was taken out 3 days later by a whisker. That then marked the intermediate-term low as the market went on to stage a nice rally from there.

The CBOE total put/call ratio came in at 1.27. While high, this isn’t nearly as extreme as the equity p/c discussed above. What is interesting here is that this marks the 7th day in a row where the total put/call ratio has closed at 1.10 or above. Going back to 1995 the only other time this has happened was August 3, 2007. That led to a 3-day rally before the final leg down of that selloff began.

While I’m seeing some measures of extremes like the CBI and Put/Call Ratios above, others seem lackluster. For instance, the VIX has yet to spike sharply. Also, volume today was light. Fear is normally accompanied by volume. Tough to call it panic selling with so little participation. In summary, small dabbling may be ok, but other than my trades which comprise the CBI, I’m not getting too aggressive on the long side just yet.

Large Gaps Higher In Uptrends vs. Downtrends

At the end of January I showed what happened when the market gapped substantially lower during uptrends vs. downtrends. Surprisingly to many, buying large gaps down in downtrends was far superior to uptrends. Tonight I thought I’d look at another side of this. How do large gaps up fare in uptrends vs. downtrends?

Once again I demanded a 0.75% gap for my trigger using the SPY. I then tabulated the results of buying the gap up at the open and selling it at the close. What I found may again surprise you.

When the market closes above the 200-day moving average, trying to buy a gap has been a losing strategy. Of the 123 instances I found looking back to 1994, only 57 had tacked on more gains by the end of the day. The rest finished down from their opening price. The net total movement of buying all these gaps up and selling them at the close would have been a loss of about 17.4% for the 123 trades.

Downtrends showed a different picture. Buying gaps up of 0.75% or more during downtrends was actually profitable. In this case, 58% of the 105 instances finished with the SPY closing higher than it opened. The net total of the movement from open to close was a gain of about 26% as opposed to the loss we saw above.

I suspect short-covering is a big reason that large gaps tend to spark additional buying in downtrends but not in uptrends. Stops get blown through overnight and when they see the market getting away from them, panic-covering ensues.

Regardless of the reason it appears when the market is below its 200ma the easy money is typically made not by fading the large gap up, but by looking to go long. Fading large gaps up appears to be more fruitful in uptrending markets than down. These results seem to go against conventional wisdom and provide another example of a lesson that many traders may need to “un-learn”.

Back to Back Reversals and a Growing CBI

Reversal Bars
The market gapped down this morning, sold off to put in a 10-day low and then reversed and closed above the open. Both the open and close were in the upper half of the day’s range. I ran a test too see possible significance of this type of bar. Below are the results:

What’s amazing about today is that it is the 2nd day in a row we’ve had this same formation. I ran a test to see how the market has reacted to back to back reversal bars like this in the past and came up empty. This was the first time it has occurred for the SPY. It has never occurred for the NDX.

The above reversal bar on its own has shown a mildly bullish tendency. We’ll see if the market can muster anything more than a late-day rumor-inspired pop.

Capitulative Breadth Indicator(CBI)
The CBI hit “5” today. This is the first level that I consider significant. Since 1995, buying the S&P 500 when the CBI closed at “5” or higher and selling when it was back to “3” would have results in 77% winning trades and gross profits outsizing gross losses by over 5:1. I generally don’t buy at this level but rather cease taking on new short positions. The cluster that is forming could easily rise to “7” or “10” in the next day or so if the market continues to drop. At those levels I may consider index longs.

Lagging Nasdaq Study Update

Last week I posted a study on the intermediate-term significance of a lagging Nasdaq. The results suggested more downside was likely in the weeks and months to come. There was 3 criteria for the setup. Surprisingly, the 3rd criteria was not met last week. “3) The difference between the current NYSE/Nasdaq ratio and the 10-week EMA must be at its widest point in the last 5 weeks. (The red line must be farther below the yellow line than it has been in at least 5 weeks.)” Updated chart as of 2/29/08 below:

Does the late-week change to the indicator benefit the market? It doesn’t appear so. I removed rule three and just tested on rules 1 and 2:

1) The NYSE must make a new 5-week high this week. 2) The current NYSE/Nasdaq ratio must be at least 3.0% below the 10-week EMA. (The red line must be 3.0% or more below the yellow line.)

The market was shorted if the above conditions were met and covered X weeks later. $100,000 per trade.

The results – listed below – changed very little. When the NYSE is hitting new short-term highs while the Nasdaq is lagging badly, it normally means the NYSE will succumb to the will of the Nasdaq. Like the original study, the expectation over the next 1-10 weeks is negative.

After 4 Months Lower Are We Due For An Up Month?

I checked to see how the S&P 500 has reacted past times it finished down 4 months in a row. Going back to 1960 there have been 10 other occurrences with exactly 4 down months in a row. Of those 10 times, 5 finished higher the following month and 5 finished lower. The average winning month was 3% and the average losing month was 2.5%.

Looking at any time the market has closed lower 4 or more times, I found 21 occurrences. Ten finished the next month higher and eleven finished it lower. Average gain was 5.1% and average loss 4.5%.

The longest losing streak was in 1974 when the S&P 500 finished lower 9 months in a row.

Looking at consecutive lows on a monthly chart does not seem to provide any quantifiable edge.

1st Trading Day of the Month – Is There A Bias?

Over the weekend I got an email from “Brian” – a regular reader. He wrote that he had read one time that when the S&P 500 closes down 1% or more on the last trading day of the month, the next day has had a strong positive bias. He went on to note that eight of the last ten times this has happened the month has started with an up day.

I’ve read other research in the past suggesting that the 1st day of the month had a strong seasonal bias and I decided this whole notion was worth exploring more.

Brian’s 8 for the last 10 result matched my data, and in fact the 1st day of the month has been positive 11 of the last 13 times going back to November of 1998. Looking back further, though, there was no apparent edge. Between 1960 and November of 1998 there was a 1%+ drop on the last day of the month 25 times. Thirteen times the market lost ground the next day and twelve times it gained.

If we eliminate the 1% requirement and look at instances when the last day of the month closed down at all I found the following:

Since 1960 there has been a 55% chance of the 1st day of the month then having an up-day. When you consider that the chance of ANY day over that period being an up day was about 53%, the edge doesn’t seem large.

I broke out the data looking at 1960-1995 and then 1996-present and found some interesting differences. From 1960-1995 the chance of a down last day being followed by an up first day was only 47%. Since the end of 1995, it has happened 74 times with 54, or 73% of them, showing a gain on the 1st day of the next month.

Looking at all 1st days of the month shows a similar picture. From 1960 through 1995 the 1st day of the month finished positive 53% of the time – perfectly in line with all days. Since 1996 the first day of the month has finished positive 66% of the time.

Studies of recent history would show an upside bias on the 1st day of the month. Long-term studies would not indicate a bias. I cannot explain what may have happened in the 90’s that would have caused this to occur. Therefore it is difficult for me to say whether the bias truly has changed or what might cause the bias to revert back to pre-’96 form. Traders that try and incorporate the so-called “1st day of month bias” into their trading should be aware that the bias did not always exist, and therefore at some point may cease to exist again.

Are IBD Follow Through Days After Day 10 Less Reliable?

One of the interesting claims that William O’Neil make about Follow Through Days is that they are less likely to work if they come more than 10 days from the potential market bottom. As part of the study on Follow through Days, I decided to test this. Those who missed the first several installments of this study may want to click on the “IBD Follow Through Day” label lower down on the right hand side of the page. This will be the 9th installment in the series.

Using the original basic assumptions of an 8% decline needed and a 1% up move on the Follow Through Day (as opposed to the current 1.7% requirement that I found to be less effective), I reviewed all FTD’s listed in the study.

A Follow Through Day actually occurring after day 10 was a fairly unusual occurrence. Downtrends and bottom formations typically carry significant volatility, so a strong, high-volume move off a low normally occurs before day 10.

Of the 65 FTD’s listed in the study, only 8 of them occurred on day 10 or later. They are listed below along with the FTD Day # and whether they were “successful” or not. (Success was defined in Part 1 – Are They Predictive?)

Seven of the eight FTD’s that came after day 10 were successful according the study. While the sample size may be too small to claim significance, there certainly seems to be no credence to the claim that FTD’s after Day 10 are LESS reliable. In fact, the opposite appears true. Seven out of eight seems especially impressive considering the fact that only 55% of the FTD’s in the study were successful. I suspect one reason for this may be that the delayed FTD allows stocks more time to carve out proper basing formations before the market attempts to launch higher. In light of the facts, it seems a curious claim for IBD to make.

The takeaway here is: next time a follow through day doesn’t come immediately, traders shouldn’t fret. The chance of success is likely higher.

VIX System Discussion Follow-up

After my VIX post a couple of days ago, “Frank” made some interesting comments in the comment section. He duplicated part of the system I discussed, but his results were very different. Using a 15% VIX stretch (a close 15% above the 10ma) he found 17 of 18 short trades since 2004 to be winners if you wait for a reversion to the 10ma to exit.

The big differences sparked some good conversation over at the Daily Options Report, which has picked up on this study. So I felt I should clear up a few things.

1) I mistyped in my post. Options were not used in my testing – futures were.

2) My “system” looked at several measures of overbought/oversold – it generally entered if more than one triggered and exited when the triggers were removed. The 15% VIX stretch was one example.

3) I did not scale in. If a trigger occurred, the system was basically “all in”. I’m sure scaling as Frank did, would help.

4) I don’t think 1, 2, or 3 above really matter much. Looking back to the beginning of 2004 I did a quick count of the number of days that the VIX closed at least 15% above its 10ma. I counted about 70 times. Frank only found 18 times. I suspect Frank was looking for a 15% stretch in the futures. I was looking for a 15% stretch in the actual VIX.

My point was to show that the futures and options could not be easily traded based on the action of the VIX alone. A stretch in the futures successfully reverting to its mean doesn’t surprise me. I’ll bet in many of those 18 cases the VIX index would have put the system into the trade even earlier and at a worse price – turning several of Frank’s winners into losers.

To sum up – the VIX is not tradable. Buying calls or puts based on VIX action as I’ve seen suggested in the media looks good on the surface but is a horrible strategy. Frank has generously shown that astute traders CAN successfully trade VIX futures by focusing on VIX futures action.
Below is a chart of the VIX(green) and front month VIX futures (orange). The chart is a few months old because I don’t have my futures data fully updated (I don’t trade it). The takeaway from this chart is that sharp moves in the VIX “cash” index are dulled in the futures. Note how while they generally move in the same direction, the cash VIX will oscillate around the futures. If you look at futures further out than front month, this “dulling” of the VIX moves by the futures will be even more pronounced.

The Intermediate Term Significance Of A Lagging Nasdaq

One indicator I look at comes from Gerald Appel’s book “Technical Analysis – Power Tools For Active Investors”. It is a relative strength measure of the NYSE vs. the Nasdaq looked at on a weekly chart. Without going into great detail, the premise behind the indicator is that the market tends to perform better when the appetite for Nasdaq stocks is greater than the appetite for NYSE stocks.

Part of this is due to the higher volatility of the Nasdaq, and part of it is due to investors willingness to speculate more aggressively when their outlook is positive. Whatever the reasons behind it, the indicator has been a pretty good barometer over the years. Mr. Appel suggests using a 10-week relative strength indicator to measure this phenomenon. This is what I’ve done in the chart below. The way the indicator works is as follows: When the red line is above the yellow line, the Nasdaq is leading the NYSE. When the red line is below the yellow line, the Nasdaq is lagging the NYSE. (Click to enlarge).

Since 1971, close to 100% of the market’s gains have occurred when the Nasdaq is leading rather than lagging. As you can see from the chart above, the Nasdaq has begun to lag badly. I decided to look and see how the market has performed under similar conditions in the past.

Using the NYSE composite as the “tradable” vehicle, I set up the following rules: 1) The NYSE must make a new 5-week high this week. 2) The current NYSE/Nasdaq ratio must be at least 3.0% below the 10-week EMA. (The red line must be 3.0% or more below the yellow line.) 3) The difference between the current NYSE/Nasdaq ratio and the 10-week EMA must be at it’s widest point in the last 5 weeks. (The red line must be farther below the yellow line than it has been in at least 5 weeks.)

If all three conditions are met, sell the market short on the close. Cover X weeks later. Results below:

As you can see, a Nasdaq lagging as badly as it is right now has been quite bearish historically. The bearish tendency carries through over a significant period of time as well (10 weeks.) If the Nasdaq could begin to assert a leadership role, that could help the current rally attempt greatly. If not, bulls better hope it’s different this time.

Do Reliable Oscillations In The VIX Make VIX Options An Easy Profit Vehicle?

I’ve seen some articles in the press over the last few months suggesting that one way to profit in volatile markets is by trading VIX options. They typically make it sound easy. “You don’t even need to know the direction of the market. You just need to determine whether volatility is likely to rise or fall. If you think volatility is going higher, you can buy VIX call options. If you think its going lower you can buy VIX put options.” The problem with this logic is that VIX option prices do not follow the VIX index. They follow VIX futures prices. A couple of months ago I decided to quantify how much this really matters.

It is well known by traders that the VIX has a strong tendency to oscillate. Therefore, when people consider trading the VIX, they many times think mean-reverting strategies will work best. I took some simple mean-reverting strategies and applied them to the index to see what kind of returns I would get. Two examples were: 1) Short the VIX if it closes 15% or more above its 10-day moving average. Cover when it closes below its 10-day moving average. 2) Short the VIX if it closes at a 10-day high. Cover when it closes below its 10-day moving average. In both cases the opposite stretch would apply for purchases. Not surprisingly, they worked. What was intriguing was HOW WELL they worked. I then combined these strategies with a few others to create an indicator which would signal to me when the VIX was stretched and due for a reversal.

Assuming you treated the VIX as a security and allocated a certain dollar amount whenever you bought/shorted it, over the last 3 years my simple system would have returned about 170% per year based on raw returns (no commissions or slippage).

Now, to see the effect that trading futures would have on the system, I downloaded all the historical data from CBOE and ran the trades through using front month options. I performed rollovers those times when the future expired before the trade closed. Note that the entry and exit triggers were based on the action in the VIX – not in the futures. The purpose of the study was to see whether someone could trade futures/options based on the action in the VIX index. The results? Instead of returning 170%/yr over the last 3 years, the system now returned 5% total!! Factor in some commissions and slippage and my incredible system is now a money-loser.

Moral of the story: Be careful when trading VIX options / futures. Simple systems which look spectacular on the VIX cash index simply do not translate.

Below are some informative links which also discuss this issue.



Surge Leads To Breakout – Is That Good?

More bailout talk today led to another surge of buying. The major averages closed strongly positive on good breadth and increased volume. Technically notable is the fact that the triangle pattern in the S&P was broken to the upside. An argument could be made that the triangle broke to the downside on Friday and that break quickly led to an upside reversal. Based on the definitions in the studies I laid out last week, there was no break on Friday. The first breakout came today. In this case I’ll stick with my definitions because those are the ones that were used to develop the statistics. The minimum target based on those studies’ triangle measurements would be 1446.57. Of note is that over 70% of these patterns have failed to reach their target before dropping below the lower triangle line. In this case a failure would mean a move below 1327.04. It appears looking for an entry to fade this breakout may provide the greatest edge.

The surge the last two days which broke this market out has certainly been impressive. I looked back to see how the market had reacted following similar boughts of strong buying. Below is a table displaying how the market has performed near-term following two strong days of buying during a long-term downtrend.

Historically, fading these surges has provided a decent edge. As can be seen in the picture above, not all breakouts are good news. Between the 2-day surge and the triangle breakout study it appears the edge over the next few days is to the downside.