New Lows as a Warning Sign

Bottom Line: On a 10-day moving average basis, new lows on the NYSE have overtaken new highs. While it sounds ominous, historically it more often meant that we were near the end of a downtrend than the beginning.

On Monday, the average number of new lows on the NYSE over the prior 10 days reached 125, greater than the 106 average number of new highs over the same period. That was the first time since February 2003 that new lows exceeded new highs (on a 10-day moving average basis), and was a sign to some that we are entering a dangerous period for the market. I was asked today if using this type of new low/new high crossover was an effective way to enter or exit the market, and it turns out the answer is “not really”.

On the surface, using such a system should make sense. After all, it seems as though you should generally want to be long when new highs continually exceed new lows, and short (or in cash at least) as long as new lows exceed new highs. Certainly, doing so will allow you to capture some big trends, but this type of system suffers from the same issue that most trend-following systems do – it gets you in and out too late.

Looking over the past 40 years, we can construct a very basic system based on the movements of the new highs and new lows. Let’s go long the S&P 500 when the 10-day moving average of new highs first exceeds the 10-day average of new lows, and exit our long when new lows first overtake new highs. Also, as a second system, let’s short the S&P when new lows first cross new highs, and exit our short when new highs exceed new lows.

	Long when Highs > Lows, Cover when Lows > Highs	Short when Lows > Highs, Cover when Highs > Lows
Avg Return	1.5%	0.8%
% Winners	36% 43 out of 121	26% 31 out of 120
Max Gain	40%	24%
Max Loss	17%	8%
Avg Gain	9.0%	4.8%
Avg Loss	2.5%	2.7%

From the table, we can see that neither one had a good average return. Also, both systems had extremely low winning percentages. The one positive aspect about each of them is that the average (and maximum) gains were well above the average (and maximum) losses. This is in keeping with most trend-following systems – they have a lot of losses, but the few gains they do have usually make up for all those losses. You have to capture those few big trends to make the system even remotely viable.

If you had tried to use this as a timing mechanism, going long when new highs exceeded new lows, then switching to short when new lows exceeded new highs, you wouldn’t have done very well. $10,000 beginning in 1965 would have grown to only around $20,200 currently, for an average annual return of just over 2.6%. The maximum drawdown was a huge 74%, as the system would have absolutely chopped you to bits over the past five years. As a matter of fact, since 1999, out of 38 signals (long and short) only 5 of them would have been profitable!

It is undeniable that this type of system would have kept you out of several of the major declines, and kept you in some of the best markets – such as over the past year - and there are hundreds of ways to tweak the system by adding in stop losses, profit targets, adjusting the length of the moving averages, etc. But over the past 40 years, as far as accuracy goes this system was a much better contrary indicator than anything – meaning that when new lows exceeded new highs on a 10-day moving average basis, it more often came near the end of a downtrend than the beginning.

Bottom Line: Our shortest-term model for the Nasdaq hit a new high at today’s close, and its recent history suggests that the risk/reward has now shifted back to favor the long side.

On Tuesday, I highlighted the bearish indication our STEM model was giving. The overtones from that model have now been alleviated to a great degree with the market declines over the past two days, and now we’re seeing an equally strong but opposite signal being given from another model. Today, the STEM.MR model for the Nasdaq closed at 80% - a new record. Since the beginning of the year, the model has reached 70% four times, shown by the green circles below.

Note that the peak in the model didn’t coincide with an exact low in the NDX in any of the four cases. On 1/29, the NDX dropped another 9 points before rallying 31 points. On 2/23, it declined 12 points before jumping 22 points. On 3/15, it lost 8 points before rising 38 points. And on 3/22, it dropped just a few points over the next few days before the large 58 point rally.

Even though a peak in the model didn’t mark any of the lows, it did provide a very good “heads-up” that the risk/reward was beginning to favor the long side. This suggests that while today’s low in the NDX may not hold over the next few days, we should see more points to the upside than the downside. If we instead just continue to creep lower, then we have a good indication that the larger trend has further to go on the downside, and I suspect the March lows would be tested and likely broken.

As you can see from the Seasonality section of the site, the first few trading days in May tend to be quite positive, showing an average return about 4 times that of a random day and a greater-than-average chance to being up. Coupled with some of the extremely oversold short-term readings we’re getting from many of our measures, the odds seem high that we will get a tradable bounce beginning in the next day or two. However, as I noted on Tuesday, with the S&P 500 currently trading below the 1120ish level that I have been keying on, I think some degree of caution is warranted for longer-term traders. I think it’s premature to sell everything and go to cash, but I would be backing off making aggressive new purchases until we can either make some decent progress or see more panic in our longer-term measures.

This disclosure is not intended as trading advice in any form. It is meant as a note to subscribers that the author may have a position directly affected by the market outlook reflected in the commentary. Although the author takes great pains to remain objective in any commentaries, it is only fair that readers should know that the author may have taken positions in accordance with his market outlook. Positions can and do change at any time, without notice to the reader.