On Market Volatility & Downturns

Market downturns are historically difficult to gauge, ex ante, in terms of their depth and length.

Further, it would appear that no two are alike.

To wit, the proprietary long/short allocation signals we employ have been useful, but not perfect, in this downturn.

Developed in the timeframe of the technology bubble collapse early this decade, the two indicators which comprise the signal alerted us that a downturn worthy of exiting long positions and taking short ones was imminent.

However, the actual January S&P500 plunge of more than 6% came faster than the signal for it.

What went wrong?

I suspect two things are different, with respect to the indicators. One is the much larger amount of investment assets now controlled by fast-moving hedge funds, relative to early in the decade. The other is the sheer speed of market reaction, helped in part by modern, high-speed, multi-channel dissemination of business and other news.

I don't typically focus on daily market changes, at least for investing purposes. I follow the market on a daily basis, but more to remain in touch with investor sentiment than to attempt to take advantage of any short-term market moves.

Further, not only are no two market downturns exactly alike, due to the evolution of environments surrounding equity investment over time. There are also few significant lengthy equity market downturns to study. Thus, unlike performing quantitative research on what works in typically upward-trending S&P500 equity markets, doing research on lengthy down markets is much tougher because there's simply less data, from fewer instances. Perforce, the certainty with which one can draw inferences and build indicators is much less than that for long-allocation periods.

Nevertheless, my partner and I fretted that we should have seen, by the beginning of December, that a large-scale decay in equity values was looming.

As a result of our move to options portfolios last year, we had a string of call portfolios which, beginning in late October, were showing consistently worsening losses, despite gains in our companion small equity portfolio.

After revisiting our current indicators and making some ad hoc modifications, we were still dissatisfied with the sense that we were engaged in 'fine tuning' a rather less-sensitive pair of indicators.

As we discussed this, I conjectured that there was probably some much shorter-term volatility-oriented measure that might inform us. Perhaps a monthly-based measure of daily change.

Ironically, I had posted this piece on a measure we labelled MFQ, for "Market Fluctuation Quotient," in late November. In that post, I concluded,

"If you are a frequent trader, than the MFQ is relevant. It clearly captures the recent, heightened day-to-day volatility, when that term means changes in closing market value direction.

We, however, are longer-term investors. Even in our equity options portfolios. As such, the MFQ doesn't really affect our investment decisions very much.
Still, it's interesting to note how divergent the two different perspectives on market 'volatility' can be- a series of standard deviations of month-to-month S&P returns, and a series measuring S&P daily close sign changes.

Clearly, volatility is in the eye of the beholder, and his/her frame of reference."

Technically, that was correct. However, the MFQ, as my partner still complained, lacked a magnitude component.

Building on this, after our recent discussions of a few weeks ago, I went back to something more classic. Taking the daily S&P returns for each month since 1990, I calculated a straightforward standard deviation of daily total returns of the index.

When this is plotted with to one of our two existing allocation indicators, we believe the two become leading indicators of a significant, longish equity market downturn.

The degree to which the monthly intra-day standard deviation of the S&P rose significantly, and remained high, shortly before a pronounced and lengthy market downturn of 2001 and the recent spate of losing index months, is unerring.

When put in context by the other, existing indicator, labelled "Factor A" on the nearby chart, it is a perfect pairing of leading indicators. Factor A turns steadily down just when the intra-day standard deviation skyrockets, in advance of a market downturn. But either one, on its own, is not predictive.

Armed with this new indicator, we are viewing the current market turmoil through new lenses, and, thus, with a new perspective.

The monthly intra-day standard deviation rose ominously in November. Had we been using this measure at the time, when I wrote the prior, linked post, we would have been out of long positions as of December. And probably already into put portfolios.

January's standard deviation rose again, to near-November levels. The evolving value for February is near January's level.

We think it's going to be at least a few more months before it is safe to remain long in equities or their derivatives.