dshort.com: Tale of Two Regressions

William Witzeman from El Centro, California emailed me earlier this week with an alternate take on the linear regressions in our Connect the Dow Dots discussion. He writes:

When the bottom dropped out of the stock market last September, I though the move was extreme. And when we finally got on the same track as the Great Depression in March, things appeared to be quite overdone. Over the same elapsed time, the current crisis is more severe and perhaps destined to be longer lasting than the Tech Crash or Oil Crisis. It seems implausible that this bear market ended in March after a mere 17 months.

So I propose a different line. I would like to see what the linear regression would look like if drawn from the 2007 peak to just before the Lehman collapse and then extended to month 34 (excluding the actual data from the Lehman collapse to present).

OK, William, let's take a look. But first, a note for those unfamiliar with the concept: The regression we're discussing is a straight line running through the scattered data of daily closes so that the amount of scatter (distance from the line) is the smallest possible, as defined by the least squares method.

Here is a chart with a linear regression through the S&P 500 since the October 2007 peak and the hypothetical regression shown as a dotted line. The dotted line is based on the 234 daily closes in the S&P 500 from the peak to the day before the Lehman collapse in September 2008. But we've let Excel extend the regression, based only on those 234 data points, to match the duration of the 34-month Dow Crash.

What does this hypothetical regression tell us? Although it has no predictive power about the future of the current market, it does support the view that the Lehman disaster turned a nasty bear market — perhaps something like the 2000 Tech Crash — into something much scarier.

Is the financial crisis now behind us? Will we ultimately look back on March 9th as the end of the bear market? Or do we have more cliff-diving ahead?