Matthais Parcha BUS483 Research Report

I'm happy to say that we're encouraging university classrooms to use Equities Lab in their classwork.  Here's an example of some of the work that has resulted.  Notice how well the long/short strategy does…  It's been reprinted, unedited, other than some (partial) attempts to format what used to be a Word Document for WordPress.

 

GOIZUETA BUSINESS SCHOOL

Final Project

Applied Investment Management (BUS 483)      

Spring 2014, Professor Busse

Mathias Prucha

4/24/2014

Rationale for Approach

General Goals

The overall main goal was to create steady, positive returns while avoiding large draw downs and a high standard deviation. The ideal market neutral strategy would generate a high positive alpha combined with a beta close to zero, indicating there is very little exposure to the market. Having high absolute returns was therefore not the criteria for assessing potential strategies. A high sharp ratio, as a risk adjusted return measure, and a low correlation with the benchmark index was the preferred criteria. 

As common used strategies like momentum or value based approaches tend to have good returns but also high volatility in terms of draw downs and standard deviation I decided to go for something different. My intuition was to exploit some sort of mispricing and hope that the market will correct it in the near future. More precisely a disproportion between the past performance of a stock (e.g. a more technical indicator like relative strength which tracks the performance of a stock compared to the overall market) and the actual operating performance of a company. The idea is that the market is disconnected from the real company financials for various reasons (over/under reaction to news and other irrational behavior). This market inefficiency should only exist in the short term and correct as soon as possible if the market efficiency hypothesis is right at least to some extent.

Long (Figure 1: Long-only Screener)

After applying the given restrictions I chose to be stricter on the market capitalization and take

$500M instead of $250M as a cut off to avoid a large small cap exposure and therefore a higher volatility. As operational performance indicator the cash flow growth per share (>15%) over the last year seemed to be a good choice. A larger cash flow growth is a positive sign for both the operational and the financial health of a company. In contrast to this significant growth the stock still has a very low price/earnings ratio (<9) for the current year. So there are two value based indicators which contradict each other. On top of that I added a technical indicator, the relative strength compared to the S&P 500 over the last month. The restriction posed that the stock has not outperformed the market by more than 30% over the last month. This should exclude high flying momentum stocks or other potential market overreactions. In order to filter out potential negative outliers (companies in severe distress) I excluded stocks which had a closing price of less than 10% of its closing price 20 days ago. Moreover stocks needed to have a beta of 0.9 or higher (for more market exposure). This filter improved my performance while volatility stayed pretty constant. For my goal of a lower turnover I restricted trading. The portfolio gets only rebalanced in January, June and November each year. This worked out best although I have no real rational explanation for choosing these specific months (January – January effect, June – investors cash out before holiday season, November – tax selling starts). Overall the strategy could be described as a hybrid value strategy that eliminates outliners on both sides.

Short (Figure 1: Short-only Screener)

The same given restrictions for trading volume, market cap and share price were applied. Similarly to my intuition for the long strategy I tried to come up with another sort of mispricing, where the market actually has very high expectations for a stock but the past performance is clearly negative. As indicators for high expectations I chose Price to Book (>6) and Price to Earnings (>30) ratios. Both suggest that investors consider the stock has “high growth” but this makes these stocks also very volatile as market sentiment can change quickly.  The idea is that these stocks are expected to outperform (high P/E and P/B) the market if they actually underperformed investors will pull out their money. So the restriction posed that a stock had to underperform the S&P 500 by more than 8% over the last six months. Past underperformance should therefore continue as frustrated investors lose their confidence in the high expectations they had initially. “High growth” stocks tend to get hit particularly hard when investors have to revise their expectations down. To reduce turnover I implemented the rebalancing restrictions similar to the long strategy (rebalance in January, June and November).

 

Discussion of Results (see Table 1)

Long 

The annualized return is 26.16% which significantly outperforms the S&P 500 in same period of time (excess return of ~5%). A relatively low standard deviation of around 30% provides a sharp ratio of 0.8 after subtracting the risk free rate. Due to a high beta of 1.33 (regression analytics are on the excess returns of the S&P 500) the strategy does very well in bull markets but performs worse than the market in bear markets. Drawdowns are particularly noticeable in the 2008 market crash (-61%) and the 2011 euro debt crisis (charts in Figure 2). There is a higher correlation with the Russell 2000 (0.83 vs. 0.7) which indicates a larger exposure to small cap stocks compared to the S&P 500. The R-Squared of 70 indicates that 70% of the returns are explained by the market. Overall the strategy generates a high annual alpha of almost 19%. The total trades are reduced to 6277 (about 2.5 trades per trading day) due to the earlier mentioned rebalancing restrictions.  So transaction costs are not overly high which is very important for the successful implementation of the strategy.

Short

Equities Lab results are like the portfolio was actually longed not shorted, so I multiplied the daily returns by -1 to get my real short returns. The strategy has a negative annualized return of  

-2.87% and a standard deviation of 32%. After subtracting the risk free rate the sharp ratio is        -0.133 which seems to be not very attractive. The main goal, however, is to offset the large draw downs of the long strategy. As the beta is highly negative with -1.21 and the correlation with the S&P 500 and the Russell 2000 is also very negative (-0.45 respectively -0.40) the strategy seems to fulfill its purpose. A look at the chart (Figure 2) reveals that the strategy does best during the financial crisis of 2008 and the euro debt crisis of 2011. That was the hedge I wanted. The largest drawdown happens in the huge recovery of 2009 and early 2010 (-69.89%). 

Market Neutral

To combine the strategies an equal weight on both portfolios was assumed. Also regulation T margin was applied so that the total returns were long + short returns. The overall annualized return increased to 31% while the standard deviation decreased to 26.05% compared to the long only strategy. Therefore the sharp ratio is significantly higher with 1.13 (again taking the risk free rate into account). The short strategy turns out to be a valuable hedge as the maximum drawdown is only 31.6% now (compared to 61.9%). During the financial crisis of 2008 the max drawdown is only 25% compared to over 50% in the S&P 500. The betas cancel each other also out which ends up in a beta of 0.04 for the market neutral strategy. Almost all returns can be attributed to alpha (30.15%). The R-squared is extremely low with 0.09 which might indicate that the S&P 500 is not the best benchmark for a market neutral strategy. Overall the combined strategy generates consistent, positive returns while avoiding large draw downs at the same time.

Sensitivity Analysis (Table 2) 

As displayed in Table 2, I made some minor changes to the two most important parameters for both strategies and back tested the strategies under slightly different restrictions again. For both strategies annualized returns varied within an expected range of -/+ 4% which indicates a certain robustness of both strategies. The analysis also shows that the finally chosen restrictions are almost optimal in terms of the annualized return. Even though both strategies are simple looking just at the applied filters both seem to be able to exploit market mispricing consistently over time.

 

Possible improvements for Equities Lab 

For a better diversification and consequently a lower standard deviation the ability to include stocks traded on international exchanges (London, Frankfurt or Tokyo etc.) would be helpful. This would simply result in a better geographical diversification. Moreover it would be useful to be able to actually short stocks. To come up with successful short strategies would be much easier this way. In addition to that I noticed that some technical indicators like RSI (relative strength index) or Bollinger Bands are not available at all. This would also be a helpful improvement. All in all Equities Lab is an extremely useful tool. Sometimes the rather long loading times are a bit annoying though. 

 

 

Figure 1: Strategy Filters

Long- Only Screener

Short-Only Screener

 

 

Figure 2: Backtest Plots

 

 

 

Table 1: Back test Statistics Summary

 

               Long only

 Short only               

Long- Short

    S&P 500   – rf

 

        Average Daily Return                              

0.111%

0.010%

0.121%

0.028%

 

Annualized Return (Appr. 1)

32.337%

2.436%

35.559%

7.203%

 

Annualized Return (Appr. 2)

26.158%

-2.874%

31.037%

5.041%

 

 

 % of non-zero  returns                             

              99,1125%                          

              97,5861%                         

           99,9645%  

– 

 

 

 

 

 

 

 

 

 

         Standard Dev. (daily)                              

                1,9464%                          

               2,0569%                         

            1,6412%    

            1,2247%     

 

 

Standard Dev. (annualized)

30.899%

32.653%

26.054%

19.441%

 

 

         Sharp Ratio (Appr. 2)                              

                 0,7989  

                -0,1331  

              1,1347      

             0,2593       

 

Alpha (annualized)

18,932%

10,657%

30,156%

0%

 

Beta

1,3348

-1,2112

0,0411

1

 

R-squared (adjusted)

70,5368

52,0032

0,0943

1

 

             Calendar Period     

from 5/20/08 to 11/21/08  

from 11/20/08 to 4/15/10  

from 1/9/03 to 3/11/03                                          

from 7/19/07 to 3/9/09

 

Maximum Drawdown

-61,905%

-69,888%

-31,604%

-57,742%

 

Financial Crisis Max. Drawdown (from 10/308 to 3/5/09)

-25.292%

 

Correlation with S&P 500  

0,70264

               -0,45177                         

             0,53710     

– 

 

Correlation with Russell 2000

                 0,83797

-0,40190

0,69407

                                 

 

All displayed data is located under “Summary” sheet on the “Backtest Stats” spreadsheet

Short statistics are computed as the portfolio would really be shorted (Equities Lab output multiplied by -1)

The Maximum Draw downs are computed using a Excel drawdown VBA (http://investexcel.net/wpcontent/uploads/2012/04/MaximumDrawdown.zip)

Risk free rate from Ken French’s website

(http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html)

Table 2: Sensitivity Analysis

 

 

                                            Long only(annualized return)

 

 

CF to Shares Gro

           >14             >15

wth 1Y

>16

<8

<9

24.57%         25.06%

26.16%

25.66%

25.11%

25.85%

<10

21.22%         21.42%        21.38%

 

 

PE Est.

 

 

 

 

Short only (annualized return)

 

Rel. Strength T6

           < -7             < -8

M

< -9

>5

>6

-7.81%

      -2.89%       -3.07%

-6.10%

-2.32%

-7.88%

>7

      -5.58%       -7.14%       -6.46%

P/B

Returns displayed as you would actually long the Short portfolio

About henry information

Henry Crutcher is an avid family guy, board gamer (think Settlers of Catan, Puerto Rico, etc), computer nut, and all around geek. Hailing from Louisville, KY, he has noticed that the weather in Louisville is remarkably similar to the weather in Atlanta, GA despite the 407 miles that separate them. He has two daughters, one cat, and lots of trees. He loves the Miles Vorkosigan series from Lois McMaster Bujold, for its mix of SF, comedy and insight into how people work. He also comsumes more than his fair share of cheesy business/economics books, such as The Ascent of Money by Niall Ferguson, or Farewell to Alms, by Gregory Clark.

More Stories

1 Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.