Trader’s Narrative

On returning from hiatus last week, I promised to give you the broad strokes of the market: then and now. There are a lot of different technical measures I could point to but in trying to keep things simple and provide a context, I found myself returning to one of my favourite market internal measures: the percentage of stocks above their moving average.

The first chart shows the portion of stocks within the S&P 500 Index (SPX) that traded above their respective 50 day moving average. This internal market statistic has gyrated from extremes several times within the past two years. On several occasions it has hit extreme lows, which not by chance, coincide with intermediate swing lows in the market.

But while the market hit extreme oversold areas, the rally that followed was the important litmus test. And if we look closely, we see that each successive oversold level was followed by a weaker and weaker rally. The bears appear to be grinding the bulls down.

Take the first instance in early March of 2007. Only 25% of S&P 500 stocks were above their 50 day moving average when the market stopped going down. The resulting rally took that number to 85% by the end of the next month.

Over at the price chart, by the end of April, the S&P 500 had recovered all the ground it had lost. Then it continued to rally, gaining an equal amount on top of the recovery. This is normal for a healthy bull market:

Now compare this to the next time the market became oversold in late July and mid-August 2007. The percentage of stocks trading above their 50 day moving average temporarily spiked below 10%. A rare feat. And as it usually has historically, this caused a rally.

But the bulls were able to recover the lost ground and a smattering more - before being pushed back in October (green arrow above). This time, the rally didn’t continue.

You can go through the next 3 examples yourself and see how prices recovered less and less with each instance of extreme oversold readings.

Notice especially how in January 2008, although there were again less than 10% of stocks above their medium term moving average, the ensuing rally was so weak that it didn’t even go above the previous swing low. And so, the market had a lower low and a lower high (red dashed line).

Extreme oversold conditions are a great opportunity for the bulls to fight back. The “value” buyer steps in and creates a floor. The momentum trader then steps in and creates a virtuous buying cycle for others.

An oversold extreme is not automatically reason enough to buy. This depends on the tone of the market. Which is only truly revealed when you observe how the market behaves after it gets oversold.

So here we are, heading into the weakest month of the year, with lukewarm sentiment and a market that seems to be looking for any excuse to meander lower.

Barry Ritholtz at The Big Picture had a recent comment about never buying a 52 week low. As you might expect, such absolutes simply don’t exist in trading.

Here’s fellow RealMoney.com columnist James “quant-jock” Altucher’s take:

I took all Nasdaq 100 stocks since 1996, including stocks that have been deleted from the index (to avoid survivorship bias). What happens if you buy stocks hitting 52-week lows that are trading for greater than $5 (avoiding penny stocks) and sell them one quarter later?

The results actually demonstrate that, over this period, the odds were on your side to outperform the market if you bought stocks at 52-week lows. The average return per trade was 7.34% (over 662 trades), including wins and losses. This far outperforms the average return per quarter of the Nasdaq during this period of 2.6%.

Some 60% of the trades turned out favorably and 40% were failures.

eBay is another, less quantitative, example of this “never buy a 52 week low” rule breaking down. eBay is a very seasonal stock which tends to bottom in the summer and top at Christmas and New Year’s.

Such seasonality makes sense because as people go out and enjoy the warmer weather, they aren’t inside selling/buying on eBay’s platform. But when the year rolls on, activity picks up as people buy gifts for Christmas and New Year’s.

This cycle naturally gets reflected in eBay’s stock. A bit like this example from 2004:

A simple rule based on buying in the summer (lets say end of July) and selling at the end of the year between Christmas and New Year’s showed the following returns in previous years:

In 1999: buy $11, sell $18 beautiful (but wasn’t everything going up then?)

In 2000: buying in the summer ($11) and selling at Christmas would have lost money… but you had a chance to sell at a good profit when it lifted to $20 in autumn 2000… before the subsequent decline in price.

In 2001: buying in the summer at $15 gave you a drawdown as eBay fell to $11 before rising again to $15 at Christmas.

In 2002: buying in the summer at $13.50 - after a tiny drawdown - saw eBay reach $18 by Christmas.

In 2003: buying in the summer at $26 gave you $32 at Christmas.

In 2004: buying in the summer at $38 gave you $58 at year’s end (see above chart).

In 2005: buying in the summer at $42 - traversing a slight drawdown - gave you $46 at Christmas.

What about this year? eBay has been getting pummelled this summer along with the general market and technology stocks in particular. Will the seasonal tendency repeat? Stay tuned and I’ll tell you in about 5 months

In the meantime, I hope it’s clear that such ‘rules’ are nonsense. A 52 week low has no real significance, taken out of context and certainly should not determine whether you buy or sell a stock in isolation.

And in case neither James Altucher’s Nasdaq 100 study nor the eBay seasonality I pointed out has persuaded you, here is another example of buying a 52 week low that worked out nicely.

John, from Tale of the Tape, asks a very interesting question regarding the nature of sentiment during a bear market. He wonders whether high bearish sentiment is a bullish signal during both bear and bull markets.

Well, lets examine the bear market that began in early 2000 by looking at the S&P 500 index and the absolute readings of AAII bearish sentiment that accompanied it:

Now, be mindful that the dates for the survey and the weekly candles don’t match exactly. What I’ve done is to round ahead to the next candle/week. So for example, if the survey date is May 12th and there are two candles with start dates May 8th and May 15th, I put the “sentiment” dot on May 15th’s candle. Capisce? The result is that in some instances it may appear that sentiment was ‘off’ or ‘late’ in signalling a rally… where it wasn’t really.

Some observations from this first phase of the bear market:

• first heavy bearish sentiment was waay off for the S&P (much better for Nasdaq)
• ditto for the lighter one that came right after it
• sentiment didn’t get bearish again for a long time
• but when it did, it was a great signal for a rally in March 2001
• in early stages of a cyclical bear market, sentiment can be a great tell
• but only after the market has seen a significant decline and/or correction

And the next phase of the bear from 2002 onward:

Wow! The graph lights up like a Christmas tree:

bearish sentiment went haywire approaching 1000 - right at previous support
• naturally, a lot of people became bearish as the index revisited its lows
• bearish sentiment accompanied the index’s plunge!
• clusters of major bearish sentiment can actually be dangerous!
• it is much better to get a lone ‘dot’ than a crowd of them (see first graph)
• you need severe and sustained bearish sentiment for a significant bottom
• bearish sentiment can ‘wash out’ - no real bearish reading after March 13, 2003

But remember that sentiment, especially only one measure of it such as AAII, should never be used as a tell in isolation. Sentiment as a contrarian measure is useful, but it must be combined with other tools to round out a picture.

Today the members of AAII responded to their weekly sentiment survey with a very lopsided expectation for the next 6 months: 58% bears, 24% bulls and the rest neutral. Interestingly enough, this level of bearishness was registered after the strong close on Wednesday. So maybe people are wising up to these one day spikes, or else, they are truly very pessimistic.

The bull ratio (calculated by dividing the bulls by the sum of the bears and bulls) was last seen at these levels in early April 2005 and in mid February 2003. To find this same level of absolute bearishness (58%), you would have to go back to early 2003 :

In mid February of 2003, we saw a reading of almost 58% bears but it wasn’t until later in March that a definitive bottom was put in and a new cyclical bull born. The March 2003 bottom was put in with a 51.4% bear reading and the bull ratio was at 0.40 - higher than it is now.

The next time that AAII bearishness approached 58%, believe it or not, was way back in October 1990 :

Unlike the 2003 scenario, from August 1990 to the end of October 1990, there was an almost uninterrupted string of bearish readings above 50%. But similar to 2003, there was a dip back to retest the bottom (purple circle) in early 1991 and then the market zoomed ahead.

After an unforgettable debut last year, Baidu.com spent the next few months deflating but really nowhere in particular. Recently though, it gapped up on a massive volume spike and it has surprisingly stayed aloft while the general market has gone to hell in a handbasket.

And this week it actually broke above its multi-month channel. Although it does have some resistance at these and higher levels, they are not that recent so I don’t think they will have a large effect. If BIDU can continue to act with the relative strength that it has shown up till now, I think we can expect some great things in the weeks and months to come.

Yet, I can’t deny that the chart of Baidu reminds me of Sears Holdings. If you recall, SHLD was acting very strong as well… until it broke support last week and tumbled below $150. Obviously the large instutions that were dumping shares into the spike higher knew something!

Only time will tell if BIDU can actually hold up or whether it will falter like SHLD.