What That 90% Day May Mean

02/25/09 9:10 AM EST

Good Wednesday morning...We begin the day with some minor selling pressure in the pre-market futures.  They sold off immediately following President Obama's speech yesterday evening, and have recovered a bit in choppy fashion since then.

Whenever we get a big up or down day in the market like the past couple of days, there is always a lot of buzz about whether or not the day was a "90% day".

The typical meaning of that phrase regards 90% of the points gained or lost being skewed one way or the other, as well as 90% of the volume, which was popularized by the well-respected Lowry's technical analysis service.  That concept is what is behind the Down Pressure gauges we post for the S&P 500 and Nasdaq 100, which are quite effective as short-term overbought/oversold indicators.

We've looked at these types of 90% days many times over the years, in different forms.  We typically only look at either the Up Issues Ratio or Up Volume Ratio, however.  The latter is the one I want to touch on this morning.

Yesterday, the Up Volume Ratio on the NYSE closed at 91%, meaning that out of all the (composite) volume on that exchange, 9 out of 10 shares were traded in a stock that closed higher than the day before.  That's a pretty rare sign of heavy buying interest.

It's even rarer when it comes the day after the market closed at a new 52-week low.  In fact, it has happened only 6 times since 1950 using the S&P 500.  The chart below shows each of those occurrences, and indicates whether that low from the previous day held over the next six months or not.

As we can see from the chart, it was pretty effective as a longer-term buy signal.  In 1957 (both times) and 1970, there was some back-and-forth trading in the shorter-term, but the low that was hit prior to the 90% day was not violated on a closing basis.

In 1962, the S&P did dive lower, violating the low four days after the 90% day.  But it only spent four days under that low, then shot higher and rose for the next couple of months.  Actually, prices rose for the next couple of years - that 1962 instance marked a major market bottom, as did the instances in October 1957 and May 1970.

So that seems OK, but then we get the last chart, which includes the most recent two occurrences from October of last year.  Both came on 10%+ one-day gains in the S&P, which was simply not sustainable.  While the previous day's low close held for a few sessions, ultimately it was broken both times.

Somewhat similar to 1962, after the violation of the October 10th low, the S&P only spent 3 days under that level before rebounding again.  After the violation of the October 27th low, it spent only two days under that level before another jump.  That ultimately led to another wave of selling pressure, but those lows were decent support.

I would have like to have seen total volume explode higher on Monday, then a violation of 740 early Tuesday, before we got an upside reversal that completely erased Monday's losses.  Usually those kinds of scenarios have a greater chance of holding, at least for a few more sessions, and it makes me more suspect of Tuesday's rebound.  Based on precedent of these 90% days, there seems to be a decent chance of Monday's low marking an intermediate-term exhaustion point, but if it's violated and we can't recovery within 3 or 4 sessions, those precedents will fall by the wayside.

Bottom line - Intermediate-term

We went over several studies in December (here and here and here) indicating that what we witnessed during November marked a major bottom.  But after what we went through to begin the New Year (e.g. the spike in Dumb Money Confidence and Intermediate-term Indicator Score, the failed breakout at 920 on the S&P, the subsequent losses of support at 880 and 850, and the failure to bounce off short-term oversold conditions), that probability diminished substantially.

Because of that January failure, I had been leery of buying into weakness until we either saw more of a pessimistic extreme in the Dumb Money, or an improved technical picture.  The Dumb Money is getting there with a current reading of 25% - low enough to be considered historically extreme - but that still doesn't approach the previous pessimistic extremes under 17% that we saw at each of the prior temporary lows since 2007.  That doesn't mean we can't rally, it just means that it's more difficult to define the probability of doing so.

As for the technical picture, the S&P 500 broke under the 800 area that had been support, and it's now hanging by a thread at the November lows.  Given the AAII data mentioned last week and a few other potential positives (now including the 90% session from Tuesday), I can't outright dismiss the potential of a meaningful low forming at any point.  The tough part is that without an extreme in the Dumb Money and the murky technical picture, I can't find a decent risk/reward scenario for supporting a longer-term trade at this point and would expect any rally emanating from these levels to be turned back near probable resistance at 800 - 830 for now.

Bottom line - Short-term

Yesterday was the 14th time since 1928 that the S&P 500 fell at least -3% to a new 52-week low one day, then rose at least 3% the next.  Most often there was some short-term follow-through, as the index showed a positive return over the next 2 sessions 10 out of 13 times, averaging +2.6%. Since 1950, it has occurred 6 times and we saw upside follow-through 5 of those times.

That's where the good news ends. Buying after two days and holding for a week resulted in only 31% winning trades, and an overall average return of -3.7%. That's pretty pathetic, and it helps confirm what we've discussed over the past couple of days - we should see a nice short-term snapback rally, but it's very difficult to extrapolate anything out of it more than a short-term snapback rally.

There have been four times the S&P 500 SPDR (SPY) dropped at least 3% with lower volume than the day before, then rose at least 3% the next day on higher volume.  It followed through to the upside over the next 2 - 5 sessions each time, but after that there was no consistency among them.  The dates were 09/01/98, 08/06/02, 10/16/08 and 10/28/08.

It's the first time we've ever seen a +3% "inside day", being defined as today's low being higher than yesterday's close, and today's high less than yesterday's open.  There have been two times where we've seen it with a 2% up day (04/05/00 and 05/13/02), after which we saw some short-term upside follow-through over the next 2 and 4 sessions, respectively, and both ended up falling to new lows.

Yesterday morning we looked at several compelling factors pointing to an imminent multi-day rebound, and the stats from yesterday help support more upside follow-through, at least for another 2 - 4 sessions or so.  If that takes us up into the 800 - 830 zone on the S&P during the first couple of sessions of March, it should set up a decent selling opportunity as we test the low put in on Monday.  If we simply fail from here and give back yesterday's gains, I do think we'll see another multi-day (at least) rebound on a violation of 740 and bounce back above.

All the best,