Friday, August 22, 2014

When is the Stock Market Overbought and Oversold?

Abnormal Returns posted a few particularly good links yesterday; check out the study of how returns early in the market day predict those later in the day.  I'll have more to say about that topic later this weekend, as it dovetails nicely with recent observations regarding who is participating in markets.  Also worth a read is Adam Grimes' post regarding the dark side of technical analysis.  A lot of market folk lore falls short of wisdom.  As we saw a while ago, cognitive biases have a way of entering many forms of market analysis--even those that appear to be rigorously quantitative.

The bottom line for the above posts is that there are worthwhile patterns out there in markets, but it is hazardous to outsource the identification of those.  Nothing substitutes for doing the work yourself and seeing, in your own experience, what works and doesn't work.  That not only yields knowledge, but also produces the genuine confidence required to trade noisy and risky financial markets.  It's tough to hold an position through normal adverse movement when the idea is not truly your own. 

Let's perform a data exercise that challenges what we know.  When is a market overbought or oversold?  Many market indicators included in data services will highlight those levels, perhaps above 70 and below 30 in an oscillator that moves between zero and 100.  But do we really know that those are meaningful levels?

I went back to 2006 and took a look at the percentage of SPX stocks trading above their 5-day moving averages.  (Data available via Index Indicators).  I broke the market down into quartiles based upon the day's closing level of VIX.  Here's what we get:

The lowest volatility market quartile averaged 59% of stocks above their five-day moving averages, with a standard deviation of 19.  The next lowest volatility market quartile averaged 54% of stocks above their moving averages, with a standard deviation of 24.  The third volatility quartile averaged 51% of stocks above their moving averages, with a standard deviation of 27.  The highest volatility market quartile averaged only 45% of stocks above their moving averages, with a standard deviation of 32.  

We know that volatility has a directional component in the stock market, so the averages are not so surprising.  Note, however, those standard deviations.  If we define overbought and oversold as fixed indicator levels--say, 30% is oversold--then we're accepting a reading of about half a standard deviation in high volatility regimes and a reading of almost 1.5 standard deviations in low volatility periods.  

At the recent lows, we got to a point where about 10% of SPX stocks were trading above their five-day moving averages when VIX was trading around 17.  That was a much rarer occurrence than if the same reading had occurred with a VIX north of 30.  Same indicator reading, two different meanings.

What is a warm day?  45 degrees on the Fahrenheit scale is a warm day in Connecticut winter and a cool day in the summer.  Context matters:  what is overbought and oversold highly depends upon the market season.  An important part of interpreting any piece of market data is knowing the season you're in.

Further Reading:  Honing Your Trading Process