Skip to content

A Simple Fundamental and Technical Rank Swing Portfolio: Combining RROC with Analyst Estimate Revisions

October 14, 2009

Ok, so its time for CSS Analytics to move a little bit closer to its traditional background focus: combining fundamentals and technicals. For those of you that are not aware, CSS Analytics sells stock rankings to institutions and in future posts we will preview some of the performance our proprietary research. In this series we will look at how to expand upon the super simple RROC strategy using a variety of techniques. For newbies, the RROC strategy entails buying the worst 20 performers in the S&P500 index based on the 1-week % price change (or ROC), and shorting the top 20 perfomers. One of the issues with this strategy despite its solid performance is the risk versus reward–the equity curve isn’t very smooth,  and the fact that it tends to hold “torpedo” stocks that would be scary to pull the trigger on. Certainly given these two characteristics, it is not recommended that a trader should hold a concentrated portfolio using the RROC strategy. Naturally fundamentals help to address the previous problem of holding torpedo stocks—companies that have improving fundamentals should be less susceptible to extreme moves than companies with deteriorating fundamentals. The problem is that many fundamental criteria possess “yesterday’s news” and have a tendency to lag price moves. Analyst estimate revisions of a company’s next fiscal year earnings projections in the most recent month tends to be a good short term predictor of stock returns. If earnings estimates are raised, stocks tend to outperform the market, and the half life of this effect is not much longer than a month. The reverse happens when earnings estimates are lowered. Thus for swing trading strategies, estimate revisions help put the “wind at our back” so to speak, and help to ensure that our risk/reward is more favorable. This allows the possibility of more concentrated portfolios with higher returns without sacrificing too much in terms of risk (more on this later). Nonetheless, one persistent pattern with estimate revisions is that it lags the most at the onset of new bear markets– just look at the long only strategy drawdown starting in 2008. Suddenly the market “darlings” are over-owned and vulnerable to sharp sell-offs. A 200 day moving average rule or position sizing based on the market trend easily remedies this issue.

Below are the results for a simple rank combination of estimate revisions and RROC. That is, I created a percentile ranking for each factor for every stock in the S&P500 and I ranked them all from best to worst.  An equal weight was assigned to avoid any overfitting though this is far from optimal—in actuality a higher weight on RROC with a smaller weight on estimate revisions produces the best risk/reward portfolio. As is my signature, I will assign an acronym: the FT rank. While the results are not as high during the bear market as the original RROC, they are less volatile and perform much better during the previous years spanning back to late 2003. The combined portfolio produces a 25% annualized return with a high Sharpe Ratio. This compares to the S&P500 which has been roughly flat over the same time span.  The maximum drawdown of the FT  long/short portfolio is a respectable 19%—you might notice that the combined portfolio has far superior characteristics than the individual portfolio components. This is due to adding another fundamental factor which helps to produce a portfolio that is less likely to be mismatched—ie we are not shorting high RROC stocks that also have high estimate revisions. The probability of achieving any technical edge is maximized by considering relevant technicals (not all are relevant and for a given time frame) in conjunction with fundamentals. In the next post we will improve this strategy further.FTrank2

7 Comments leave one →
  1. quant permalink
    October 15, 2009 10:05 am

    Thanks very much for this and other research. Two questions:

    (1) Is there possibly survivorship bias — what about stocks that were in the S&P 500 at the start of the test but not now?

    (2) What are your data sources, especially for the earnings estimate revisions?

    • david varadi permalink*
      October 15, 2009 10:29 am

      hi, there is no survivorship bias here it is accounted for in the software which is Zacks Research Wizard—which is the source of estimate revisions data as well.

      cheers
      david

      • Carl permalink
        May 10, 2010 11:41 am

        Hmmm – you may want to double check with Zacks. The last time I checked with Zacks they acknowledged a survivorship problem in that database. I discovered it when a deep value screen showed great results – the reason was the bankruptcies had been removed!

Trackbacks

  1. Improving RROC with Fundamentals: A Combo Ranking « CSS Analytics
  2. Thursday Potpourri | HighYields.com
  3. Improving RROC with Fundamentals: A Combo Ranking | Financial engineering source center
  4. Checking in on the 5 Day Rate-of-Change Simple System | System Trading with Woodshedder

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: