VIX Foretells Short Term Correction

I sent out a quick note on a VIX sell signal yesterday; Here’s the full analysis:

The CBOE Volatility Index (VIX) hit a new 52 week low on Wednesday. At one point intraday, the VIX touched 16.98, the lowest level since August 2000. This suggests to us a very high level of complacency amongst investors. It is a cliché that “Wall Street likes to climb a Wall of Worry,” and that worry seems to be missing form investors’ psyches. Historically, that lack of fear has been associated with intermediate market tops, which suggests to us that a consolidation could start fairly imminently – quite possibly, before the next 30 days elapses.

We looked at each of the VIX signals going back to the prior low of 8/25/00. We looked at each instance where the VIX rose at least 10 points to hit levels over 30. We also studied where a drop of 10 points led to the VIX breaking below 18. In all of these instances, the VIX signal was followed by a significant move. What surprised us was that, in the midst of the Bear Market of 2000-2003, the upside moves based upon this VIX signal were as strong as they were. What you heard is true: Bear Market rallies are particularly vicious.

The VIX gave 6 “complacency” sell signals, and 5 “panic” buy signals. All of the panic signals happened at least 6 months ago (so we have at least that much data on them); Only 3 of the complacency signals, however, happened over 6 months ago. Three recent signals (of a drop of 10 points breaking below 18) have occurred since July 2003. The data set is less complete, and therefore less reliable. Consider this a work in progress.

The VIX buy signals produced subsequent rallies on the Nasdaq ranging from 25.6% to 48.5% over 6 months (peak to trough). The average was 37.8%. The S&P500 and DJIA had gains of about 22%. SPX ranged from 20.1% to 24.5% (avg=22.1%), while the Dow gains were from 18.2% to 32.3% (avg=22.3%). See attached Excel doc.

The losses following the VIX sell signal where we have 6 months of subsequent data were significant. Those three VIX signals led to losses on the Nasdaq averaging 39.3%. The July and September signals have produced peak to trough drops of only 5% and 6%. On the S&P, the prior losses averaged of 25.7%, while the Dow dropped 22.7% More recent signals have produced drops of less than 10%.

It is ordinarily imprudent to rely on a single data point – even though the VIX has proven itself to be a rather reliable signal. The difficulty at this time is that we are unable to control for the variable of the massive stimulus at present. With never before seen levels of deficit spending, tax cuts, interest rate cuts and money supply stimulus all at once, it is all but impossible to compare yesterday’s VIX signal with any other analogous time. There simply has never been this much stimulus coursing thru the system at once.

See VIX chart below.

Download Excel file VIX Analysis

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. commented on Oct 20

    Carnival of the Capitalists

    This week’s edition of Carnival of the Capitalists is up at Jay Solo’s Verbosity. I’ve read through most of the entries, and the ones that have piqued my curiosity so far are: Kevin Brancato’s entry on income mobility. An overview…

  2. Jonathan Wilde commented on Oct 20

    Remember that recently the old VIX changed symbols to the VXO, while the new VIX was given the symbol VIX.

    The VXO hit as low as 18.60 on Oct 16th but did not break below 18. It still signals complacency in the market, but the level of complacency is not quite to the levels reached on those six occasions in the past.

  3. Barry Ritholtz commented on Oct 20

    I recognized that the CBOE changed their VIX computation; But you have your choice when reviewing data — you can look at the new VIX, which reflects the S&P500, or the old VIX — renamed VXO, which is the OEX 100.

    Either way, when you look at a chart of one or the other, you are looking a consistent data set.

    What I did was look at what VIX data (you could use either set, I used VIX) that were at extremes.
    I reviewed a variety of different conditons, eliminating those that did not foretell a big reversal.

    What I found was a drop of 10 points to below 18 was significant, where we had at least 6 months worth of subsequent data; The jury is still out on the July 03 and September 03 signals.

    I could have applied the same thing to the VXO (but why bother?).

    Incidentally, the reversals when the VIX spikes upwards is much easier to see and predict.

  4. Jonathan Wilde commented on Oct 20

    Ahh, I think I see. You used a consistent data set by using the backcalculated new VIX to get the historical data?

    Incidentally, the reversals when the VIX spikes upwards is much easier to see and predict.

    Yeah, that’s how I’ve made money during the last three years. The spike upwards are moments of widespread panic and are much sharper; they build momentum and in a timespan of a day or two, result in massive fear, uncertainty, and doubt. Which is the ideal time to buy.

    Complacency OTOH comes on slow and gently. It never builds to a crescendo like fear does.

    The difficulty with buying when the VIX spikes is having courage when everyone around you is selling; fear is easily recognized but action is difficult to take. The difficulty with selling when the VIX is low is in actually recognizing the complacency itself; action is easy to take, but euphoria is difficult to recognize.

Posted Under