Seeking Alpha

Dr. Ken's  Instablog

Dr. Ken
Send Message
Behavioral aspects of market moves; using quantitative measures as indices of behavior.
View Dr. Ken's Instablogs on:
  • Highest Correlation of Equity Markets and the Dollar in Decades? Or, Just Sloppy Statistics?

    Take a look at the Wall Street Journal article entitled “Stocks, Dollar in Rare Sync” that appeared in the October 29th issue of the publication.  The article quoted a “quantitative research analyst” who reportedly described the recent correlation between equities and the dollar as “astonishing”, saying that he hasn’t seen such a relationship since 1971.

    How is this analyst calculating these “correlations”?  From a review of available data, it seems that he is using the time series of S&P 500 and the US Dollar Index (USDX) prices.  He does not appear to be using the returns of these assets.  Whoever created the charts that appear in the article appears to be ignoring the statistical problems that arise from using time series data that have a high degree of autocorrelation.

    For instance, the correlation of 4-month moving average price data (i.e. simply running a correlation of the SPY price and the USDX) does,  in fact, yield the -0.80 correlation that is described.  When one runs a Durbin-Watson test on this data, however, one finds that the correlation is not as meaningful as assumed – the overlapping day-to-day prices have a high degree of positive autocorrelation, rendering dubious the significance of the high (negative) correlation.

    Nevertheless, when one runs a correlation of the daily returns (avoiding the problems of positive autocorrelation), there is still reasonable negative correlation.  Maybe not -0.80, but -.20.  This level of correlation between SPY and USDX, however, is much more common than the article suggests.  Not once in four decades, but rather once every few years.   And, over the last few weeks, the correlation has dissolved to near zero – in other words, the two assets are now completely uncorrelated.

    Disclosure: Long or Short SPY, Currency ETFs, from time to time
    Tags: SPY, Correlation
    Nov 01 7:59 AM | Link | Comment!
  • Is There Any Value to the Put/Call Ratio?
    Investors can buy or sell Index Option Puts and Calls on the Chicago Board Options Exchange, and have been able to do so since 1983. The number of transacted Call index contracts is different from the number of transacted Put index contracts on any given day (almost always fewer). The ratio of Put to Call transactions has been termed the “Put/Call Ratio” and investor lore teaches that the Put/Call Ratio is an indicator of future market direction. The Ratio has been considered to be a contrarian indicator; in other words, when the Put/Call Ratio is high the market will rise. By contrast, when the Put/Call Ratio is low, the market will fall. Stated yet another way, investors have it completely backwards; when they believe the market will fall (and consequently buy Puts) the market generally rises and vice versa.
    Does the data bear out the Put/Call Ratio as a contrarian indicator of future prices?  Well, it depends on what exactly one is measuring, and over what timeframe.
    Clearly, over the short term, the Put/Call Ratio is an abysmal predictor of market direction. Between 2004 and 2010, the correlation coefficient for the Put/Call Ratio and the next day’s SPY change (as a surrogate of the market’s return) was a dismal 1.8%. Even the last day’s SPY return is a better predictor of the next day’s SPY return, with a correlation coefficient of 12%. What about the 20-day simple moving average (NYSE:SMA) of the Put/Call Ratio? Does this predict the next day’s return? The next 20 day return? The answer to both is no; the correlation coefficient between the 20-day SMA and the next day’s return is only 0.5%, and the next 20 day return is only 2.2%.
    Let’s play one more trick. Let’s use an exponentially weighted moving average (NYSEMKT:EMA) of the Put/Call Ratio, with a very low weighting factor of 0.002 (meaning that one day doesn’t change the EMA very much at all). And, let’s see how that value predicts the next year’s SPY change (12m forward return). Now the correlation coefficient is 62%, and the chart looks impressive!


    The EMA of the Put/Call Ratio begins to rise in mid-2005 (note, the left side Y-axis is inverse; lower on top and higher on bottom), anticipating a drop in the 12m forward return that began in 2006. Similarly, the EMA of the Ratio begins to fall again in late 2007, almost exactly in lockstep with an increase in the market’s 12m forward return.
    It all sounds good, and if you are like me, you fast-forwarded to check out the current Put/Call Ratio EMA. It was at 1.44 on October 22, 2010 – reasonably low – predicting a high market return over the next 12 months. But before you go way long on the market, look at this next chart. It’s a plot of SPY versus the 12m forward SPY return over the same time frame as above.


    The correlation here is even better – fully 76%! What this means is that the price of the SPY was well-correlated with the 12m forward return; when SPY was low, the market did well over the subsequent 12 months and when SPY was high the market did poorly. Said another way, there was great reversion to the mean for SPY over a 12m timeframe. The bull market lasting through 2007 was followed by a fall; the crash of 2008 was followed by a rise; both over 12m timeframes.
    So, SPY price appears to be as good or better a predictor of future market returns than the Put/Call Ratio. In fact, I would suggest that the Put/Call Ratio is not the primary independent variable here at all – it merely reflects changes in the market. When the market falls, investors become nervous and seek protection (insurance) by buying index Puts (and the VIX rises). When the market rises, investors become complacent and the volume of index Puts decreases (along with a fall in the VIX). This is not earthshattering and certainly not a new observation. But hopefully this data drives yet another nail in the coffin of the Put/Call Ratio as a useful indicator for making trading decisions.

    Disclosure: Short SPY
    Tags: SPY
    Oct 23 6:53 PM | Link | Comment!
  • Very High Put-Call Ratio on September 7th
    The CBOE Index Put-Call ratio hit the historical high level of 2.47 yesterday the 7th of September. The ratio has not been this high since August 3rd, 2007. The SPX was near its all-time high then, and a string of high Index Put-Call ratios was observed during the summer of ‘07. Interestingly, the Index Put-Call ratio has exceeded 2.0 only 15 times since the beginning of 2008 (about 2% of trading days).
    A high Put-Call ratio can be an index of investor sentiment; a behavioral indication of pessimism at the worst and cautious protection of capital after a run-up at best. It is poorly correlated with next-day SPX returns, however, with a correlation coefficient of near zero.
    So, the meaning of this unusually high ratio is unclear, but probably worthy of mention.
    Sep 08 8:06 AM | Link | Comment!
Full index of posts »
Latest Followers


  • Major tracking error today; Yen is down against the dollar, but all the ETFs (e.g. FXY) are up... Arbitrage opportunity?
    Mar 17, 2011
More »
Posts by Themes
Instablogs are Seeking Alpha's free blogging platform customized for finance, with instant set up and exposure to millions of readers interested in the financial markets. Publish your own instablog in minutes.