Is This Decline Really All That Unusual?

http://www.sentimentrader.com/subscriber/daily_comment.php

August 20, 2007

1:30 PM EST

Is This Decline Really All That Unusual?

Over the weekend, I posted a comment that went over some of the remarkable events of the past couple of weeks, and particularly from Thursday.

We have witnessed a truly extraordinary set of events, and the few historical precedents that are available strongly suggest that we have achieved enough of a washout, with enough panic, that we should have carved out at least an intermediate-term bottom last week.

Pretty much everywhere we look, though, the nay-sayers are suggesting that the correction hasn't been deep enough (even though it's the worst one in many years...), or it hasn't gone on long enough.

So I thought it would be instructive to take another look at history. This time we'll ignore sentiment, we'll ignore breadth - well, heck, let's just ignore anything that doesn't have to do with price and time.

What we're going to look at today is every intermediate-term decline in the history of the S&P 500 to see if we can get some additional clues as to what's going on here.

The charts below were created from looking at every 52-week high in the S&P 500 from 1950 - present. I then marked the bottom of every intermediate-term low to record the stats for the correction. By "intermediate-term low", what I required was that the price had to mark at least a 3-month low, meaning that prices didn't violate that price for three months before and after that date.

First, let's look at the magnitude of declines during these corrections. Each blue bar in the chart below represents one individual correction. I identified 62 of them since 1950.

The corrections ranged from just under -5% to more than -50%, with an overall median of -10.2%. The median is the best representation of "average" in this case since the larger declines could really skew the figures, so half of the corrections were less than -10% and half were more than -10%.

The current correction is darn close to the average, at -9.4%, so this first look is right in line with prior lows.

Now is where things get a little hairy, though - the number of days that traders have suffered through declining prices.

The median number of calendar days in all the prior corrections was 75 days, with a minimum of 13 and a maximum of 1082 (yikes!). The current drop has lasted only 28 calendar days, assuming last Wednesday marked the lowest closing low of this decline.

There were 7 other corrections that were quicker than this one, and on average the S&P dropped -7.7% during those drops. So the current one has been a bit deeper than those other "quickies".

Overall, there was a good correlation between how long the correction dragged on and how much prices suffered as a result. The following chart is a scatter plot, with each dot representing a correction. For example, if you look at the three dots directly above the -50% number, you'll see that those declines lasted about 600 days, 900 days and 1100 days.

Now let's shift gears a bit and look into the future. Making the assumption that last week was an intermediate-term low, perhaps we might glean a bit of info about what could be in store by looking at some of the others.

First we'll look at the correlation between the length of the correction and the subsequent three-month return in the S&P 500.

The chart shows us that generally, the longer the decline lasted, the better the rebound was when the market finally did bottom. The correlation was pretty weak, though, as you can see the wide variance in the dots for the corrections that lasted less than 200 days.

The last one we'll look at is the correlation between the depth of the drop and the go-forward return. One would perhaps assume that the deeper the drop, the bigger the bounce, so let's check it out.

Well, that's kind of true - the correlation was positive between the two, but again it's fairly weak.

Once again, we see a wide variance near the left-hand side of the chart for corrections that averaged -10%. Subsequent three-month returns ranged from the pitiful (+2.9%) to the sublime (+24.3%).

Conclusion: Correction Has Been Short, But Deep Enough

I'm going to ignore the reasons why we might have seen a low, since I would just be repeating what I wrote this weekend. Instead, I want to focus on past corrections and whether they tell us anything about where we are now.

Less than a calendar month has passed since the S&P 500 was sitting pretty at a new yearly high, but in that time we've suffered a great amount. The length of this correction is extremely short - one of the shortest of the past 50+ years, if in fact we have put in a multi-month low. Out of the 14 declines that were around -10%, only three of them were shorter in duration than our current one.

That is a little disturbing, since it would be quite out of the ordinary, but at least it wouldn't by any means be unprecedented - we only have to go back to this past March to see that.

It's more comforting, in a way, that we have dropped as much as we have already. By almost reaching that -10% mark on a closing basis, we were very close to the average depth of all corrections.

Looking at the correlations between the drops and future returns, if we use the regression formula provided by the correlations, we get that based on the length of the decline, we should see a three-month return of +8.9%. Based on the depth of the decline, we should see a return of +9.5%. Pretty close.

It's silly to use such precise figures, but at least it gives us something to work with. And you may remember that many of the studies I went over during the past several days suggested that we might expect a three-month return of around +8%, so that's very close to the figures above, and they were achieved by a very different method.

After this analysis, I'm a little concerned about how unusually short this decline has been, but it has been severe enough - on a price basis and a breadth/sentiment basis - that I continue to go with the idea that the vast bulk of the selling is behind us when looking out at the intermediate-term.

All the best,

Jason Goepfert

President and CEO

Sundial Capital Research, Inc.

Forwarding or otherwise distributing this copyrighted material is a breach of your subscriber agreement. Violators are subject to termination of their subscription with any received subscription fees forfeited. Any references to historical performance are based on data we deem to be reliable, but are based upon feeds from third parties. We do not recommend subscribers take positions based on data presented here alone, but rather incorporate it into a comprehensive investment outlook.