Docstoc

GLOBAL ASSET ALLOCATION AND STOCK SELECTION

Document Sample
GLOBAL ASSET ALLOCATION AND STOCK SELECTION Powered By Docstoc
					    GLOBAL ASSET
ALLOCATION AND STOCK
     SELECTION
        ASSIGNMENT # 1
SMALL CAP LONG-SHORT STRATEGY
           FIRST-YEAR BRAVES
 Daniel Grundman, Kader Hidra, Damian Olesnycky,
            Jason Trujillo, Alex Volzhin
        Methodology
 Goal: to identify long-short strategy for trading
  US small cap stocks using Fact Set.
 Universe Definition: US stocks with market cap
  from $300M to $2B.
 Strategy: Buy 1st quintile, Short 5th quintile.
 Benchmark: S&P 500
 In-sample period: Jan, 1995 – Dec, 2004
 Out-of-sample period: Jan-Dec, 2005
               Factors
 We tested many factors but settled on three:
     One-month return
     Six-month return
     Current price to 52-week high
 Additionally, we tried various combinations
  of these factors (two-factor and tree-factor
  models)
     Strategy Based on
      1-Month Return
1-Month Return   1-Month Alpha
    Strategy Based on
     6-Month Return
6-Month Return   6-Month Alpha
 Current Price to 52-Week
           High
Price to 52-Week High Return   Price to 52-Week High Alpha
Other Explored Factors
       In addition to the previous 3 factors, we tried several
        other metrics:

         Book to Market Price

         Price to Earnings

         Dividend Yield

         Return on Equity

         Revision Ratio

       However, we found all of them to be of little value.
      Book to Market
          Price
Book to Price Return   Book to Price Alpha
Price to Earnings
P/E Return   P/E Alpha
        Revision Ratio
Revision Ratio Return   Revision Ratio Alpha

                 Returns good returns:
    Our one-factor models delivered
     •   1-Month Returns Model +6.98%
     •   6-Month Returns Model +4.26%
     •   Price to 52-Week High +3.55%
 However, two-factor models were even better:
     •   1-Month Return & Price to 52-Week High
         +6.95%
     •   6-Month Return & Price to 52-Week High
         +4.55%
                       Bivariate Model: 1-Month
                      Return & Price to 52-Week
                                  High
                 Price to 52 Week High and 1 Month Return            Price to 52 Week High and 1 Month Return
                                                                                       Alpha
                 6
                                                            4
                 5
                 4                                          3
Monthly Return




                 3                                          2
                 2                                          1
                 1                                          0
                  0                                              1            2         3          4            5
                                                            -1
                 -1   1      2      3      4      5
                 -2                                         -2
                 -3                                         -3
  Beta for Bivarate P to
52High & 1 Month Return
          Model
       Bet a

       Fra ctil e

                B eta (P rice to 5 2 Wee k H ig h)



2 .4

2 .2

2 .0

1 .8

1 .6

1 .4

1 .2

1 .0


                    1                2               3   4   5   NA
          Bivariate Model: 6-Month
         Return & Price to 52-Week
                     High
     P-52 High and 6 Month Return Model                P-52 High and 6 Month Model Alpha

4                                             4
3                                             3
                                              2
2
                                              1
1
                                              0
0                                                  1         2      3      4      5
                                              -1
     1        2        3        4         5
-1                                            -2
-2                                            -3
    Multivariate Model
Multivariate Model Return   Multivariate Model Alpha
                    Scoring
 We used scoring for bi-variate model (1-month
  return and price to 52-week high)
 For 1-month return:
    •   1st quintile +5, 5th quintile -5
 Price to 52-week high:
    •   1st quintile +3, 5th quintile -3
 More weight on 1-month return because single-
  factor model based on 1-month return is superior to
  that based on price to 52-week high.
                          In-Sample Two-Factor
                                 Model:
                        1-Month Return & Price to
     Alpha
                        52-Week High with Scoring                Alpha

     In-Sample Model w/ Scoring Return
     Fractile                                                In-Sample Model w/ Scoring Alpha
                                                                 Fractile

            Alpha (Total Quintile Score)                                Alpha (Total Quintile Score)

4                                                           4

3                                                           3

2                                                           2

1                                                           1

0                                                           0

-1                                                          -1

-2                                                          -2

-3                                                          -3

-4                                                          -4
                1             2            3   4   5   NA                   1             2            3   4   5   NA
Beta for Bivarate 52-P and
 1- Month Return Scoring
          Model
        Bet a

        Fra ctil e

                 B eta (Tota l Q ui n til e S co re )
 2 .3

 2 .2

 2 .1

 2 .0

 1 .9

 1 .8

 1 .7

 1 .6

 1 .5

 1 .4

 1 .3

                     1                   2              3   4   5   NA
Out-of-Sample Testing
  We used the period from January, 2005 to
   December, 2005 for the out-of-sample
   testing of our best model (two-factor: 1-
   month return & current price to 52-week
   high).
  Annualized Returns -
   •   Benchmark Return: 0.4%
   •   Our model without scoring: 11.79%
   •   Our model with scoring: 12.07%
 Out-of-Sample Two-Factor Model: 1-Month
 Return & Price to 52-Week High w/o Scoring
                                    Alph a

Out-of-Sample Model Return          Fractile   Out-of-Sample Model Alpha
                                           Alpha (Price to 52 Week High)

                             0.5


                             0.0


                             -0.5


                             -1.0


                             -1.5


                             -2.0


                             -2.5


                             -3.0


                                               1           2               3   4   5   NA
  Out-of-Sample Two-Factor Model Beta: 1-
    Month Return & Price to 52-Week High
               without Scoring
       Be t a

       Fra ctil e

                B eta (P rice to 5 2 Wee k H ig h)
3 .8

3 .6

3 .4

3 .2

3 .0

2 .8

2 .6

2 .4

2 .2

2 .0

1 .8
                    1                2               3   4   5   NA
        Out-of-Sample Two-Factor
         Model: 1-Month Return &
        Price to 52-Week High with
                  Scoring
Out-of-Sample Model w/ Scoring Return   Out-of-Sample Model w/ Scoring Alpha
  Out-of-Sample Two-Factor Scoring
Model Beta: 1-Month Return & P to 52-W
        Be t a
               High with
        Fra ctil e

                 B eta (Tota l Q ui n til e S co re )

 3 .6

 3 .4

 3 .2

 3 .0

 2 .8

 2 .6

 2 .4

 2 .2

 2 .0

 1 .8

                     1                   2              3   4   5   NA
          In-Sample Results (1/2)
Heat Map In-Sample WITHOUT Scoring:
   P52 & 1-Month Without Scoring In-Sample Quintiles   • Quintile 1 has NOT the highest average return.
 Year        1       2          3         4        5
                                                       • Only 3/10 years have the highest returns.
 1995      56%     85%        70%       40%     -16%
 1996      60%     59%         5%       12%      23%   • Here we are concerned by 2003 when we
 1997      24%     25%        45%       10%      27%
 1998      46%     25%        48%        9%      -5%     actually got the lowest returns in Quintile 1.
 1999     150%    229%        76%       91%     103%   • The spread would have crushed us!
 2000      52%    123%         6%      -37%     -75%
 2001       3%     45%         4%       48%     -47%   • Quintile 5 has the lowest average return.
 2002     -14%    -33%       -53%      -50%     -43%
 2003      60%     93%        82%       72%     148%   • 5/10 years have the lowest returns.
 2004      23%     16%        -3%        3%      -3%
Arithm.                                                • Here we are concerned by 2003 when we
           46%     67%        28%       20%      11%
Mean                                                     actually got the highest returns in Quintile 5.
         In-Sample Results (2/2)
Heat Map In-Sample WITH Scoring:
     P52 & 1-Month With Scoring In-Sample Quintiles     The scoring screen alleviates our concerns:
  Year      1         2         3        4          5
  1995   108%       38%       48%      16%        11%
                                                        • Fractile 1 has the highest average return.
  1996    31%       27%       26%      36%        16%
  1997    77%       39%       30%      32%       -25%
                                                        • 8/10 years have the highest returns.
  1998    57%       23%       37%       8%        -1%
  1999   131%      126%      112%     128%       104%
                                                        • The scoring eliminates the 2003 crush!
  2000   138%       17%       41%     -69%       -75%
  2001    69%        9%       37%     -40%       -47%
                                                        • Fractile 5 has the lowest average return.
  2002     9%       27%      -41%     -47%       -75%
  2003   176%       94%       86%      76%        26%
                                                        • 10/10 years have the lowest returns.
  2004    72%       21%       5%      -12%       -33%
Arithm
          87%       42%       38%      13%       -10%
. Mean
Out-of-Sample Results (1/2)
Heat Map Out of Sample WITHOUT Scoring:
 P52 & 1-Month Without Scoring Out-Of-Sample Quintiles
 Fractile     1        2         3       4          5      • Quintile 1 has the highest average return.
 Summary     1.17     -0.13     -0.18    -0.16    -1.95
                                                           • Only 3/12 months have the highest returns.
12/31/2004   -2.36    -6.16    -10.88    -9.26    -16.48
 2/01/2005    7.44    -2.72      1.86     0.51     -2.83   • Here we are concerned by these 2 months
 3/01/2005   -4.90    -3.94     -3.06    -3.68     -7.37
 4/01/2005   -4.66    -8.84     -8.23    -7.71     -9.81     where we actually got the lowest returns in
 4/29/2005    6.54     7.95      6.44     7.96     16.91     quintile 1.
 6/01/2005    2.59     1.82      2.08     3.78      1.27
 7/01/2005    6.53     8.48      8.06     8.89      7.10   • Quintile 5 has the lowest average return.
 8/01/2005    0.15    -0.59      0.69    -0.19     -5.82
 9/01/2005    1.31     1.60     -2.10    -2.37      0.27   • 8/12 months have the lowest returns.
 9/30/2005   -4.20    -6.38     -3.99    -5.29     -9.54   • Here we are concerned by these 2 months
11/01/2005    6.13     8.23      6.42     7.12      9.80
12/01/2005    0.70     0.87      2.50     0.38     -2.12     where we actually got the highest returns in
Geometric
   Mean
             15%       -2%      -2%      -2%      -21%       quintile 5.
                                                           • The Long/Short spread is satisfactory: 36%
             Out-of-Sample Results
                     (2/2)
Heat Map Out of Sample WITH Scoring:
  P52 & 1-Month With Scoring Out-Of-Sample Quintiles
 Fractile     1        2         3       4          5     The scoring screen alleviates our concerns:
 Summary     5.59      0.95     0.39     -2.27    -6.48   • Quintile 1 has the highest average return
12/31/2004   -3.42    -4.40    -7.35    -16.83   -15.23
 2/01/2005    4.98    5.91     1.93      -1.73    -7.00     and outperform the unscored screen by far!
 3/01/2005    1.21    -4.00    -3.57    -10.67    -8.57   • Quintile 1 has the highest average return.
 4/01/2005   -0.76    -2.39    -8.23    -13.13   -15.52
 4/29/2005   11.01    7.14     7.29     17.00     6.75      10/12 months have the highest returns.
 6/01/2005    5.47    2.29     3.41      -2.37    -0.48   • Quintile 5 has the lowest average return and
 7/01/2005   14.22    6.17     8.27      6.34     4.05
 8/01/2005    6.10    -1.87    -0.20     0.13     -7.36     underperformed the unscored screen by far!
 9/01/2005
 9/30/2005
              6.40
              0.51
                      2.19
                      -3.04
                               0.11
                               -4.34
                                         0.41
                                         -7.27
                                                 -10.91
                                                 -16.36
                                                          • Quintile 5 has the lowest average return.
11/01/2005   18.57    5.21     7.32      6.43     0.11      9/12 months have the lowest returns.
12/01/2005    4.88    -0.80    1.70      -0.75    -3.80
Geometric
                                                          • The Long/Short spread is satisfactory:
             92%      12%       5%      -24%      -55%
   Mean                                                     147%.
                                     Long/Short Distributions
                                  Positively Skewed After Scoring
                     Long/Short Returns Distribution                                                                   Long/Short Returns Distribution
                         P52-1Month In-Sample                                                                            P52-1Month Out-Of-Sample



35                                                                                                        5
30
25                                                                                                        4

20                                                                                                        3
15
                                                                                                          2
10
5                                                                                                         1
0    -35%   -30%   -25%   -20%   -15%   -10%   -5%   0%   5%   10%   15%   20%   25%   30%   35%   More
                                                                                                          0   -35%        -30%            -25%            -20%           -15%            -10%
                                 P52 - 1 Month       P52 - 1 Month Scoring                                                                    Returns RangeP52-1Month with scoring OOS
                                                                                                                     P52-1Month without scoring OOS
             Concerns
 Transaction Costs
 Short Selling Constraints
 Execution
 Volatility/Exit Signals
 Fact Set
               Concerns
Transaction Costs
  Monthly rebalancing
   •   Many months have >50% change in
       fractile components.
  Large number of securities
   •   ~60 Stocks per fractile per month
                 Concerns
Short Selling Constraints
   Dealing only with small cap securities.
      May be limited opportunity to short
      sell  some securities.
              Concerns
Execution
 How to execute as an actual trading
 strategy.
   •   When to run model?
   •   When do you make trades?
                  Concerns
Volatility and Exit Signals
  Portfolios are not Beta neutral and overall
  betas are usually above 1.
  No parameters set for liquidating portfolios.
    •   In sample we had several very bad months.
    •   Given the high volatility of small caps,
        there is the potential for very large
        losses.
              Concerns
Fact Set
Limited knowledge of the tool.
Results seem almost too good.
In practice we would run tests to verify that
what we believe is happening is actually
happening.
           Limitations
Primary limitation is the fund size for
which this is compatible.
   •   Relatively few securities
   •   Low market capitalizations
Solution: Change screen
   •   Wider market cap range
   •   Low trading volume requirement
             Summary
We find the results of our analysis to be
very compelling.
The big challenge is efficient and proper
execution.
Proper study of transaction costs is
required.
We would also recommend a further
review of the data before moving
forward.