Second Wind, from Phil McDonnell

September 8, 2009 |

Dr PhilFor much of today's action the Nasdaq lagged behind the S&P index in percentage terms. The Nasdaq has a higher beta than the S&P so one would normally expect it to move farther and faster both up and down. Like a race horse making it final surge to the finish the Nasdaq managed to surge ahead of the S&P in the final minutes of trading to restore the normal order of things.

Naturally this raised two questions. First, what happens if the Nasdaq under performs the S&P on an up day? Second, what happens the next day after the Nasdaq is stronger than the S&P? To resolve these questions the following study looked at a little over 4100 days of trading for the Nasdaq index and used the SPY ETF as the trading vehicle for the next day. The following table shows the next day results after SPY is up and it is stronger than or weaker than the Nasdaq in percentage terms.

.         +SPY>Naz       +SPY<Naz

Avg     -0.00012        0.00008

SD       0.012612993    0.010327566

n         968            1271

% Up     0.508          0.509

t-stat  -0.290          0.275

The average returns are close to zero and the percentage of times up is about 51% for each case. However the t-stat is non-significant for both cases despite sample sizes of ~1000.

Dr. McDonnell is the author of Optimal Portfolio Modeling, Wiley, 2008

David Higgs remarks:

As of late there have been some days where the move of the markets was attributed to just one or two stocks due to short covering. But then this is a different fly in the ointment…





Speak your mind

5 Comments so far

  1. Andrew McCauley on September 8, 2009 8:55 pm

    A condition of 1% plus (stronger / weaker) results may move towards significance.

  2. Matt Johnson on September 9, 2009 12:16 am

    I wonder if you thought you’d find some trading edge. Even if you had, I feel, at least, trading it would be significantly harder to do than the study you performed.

  3. anton johnson on September 9, 2009 11:14 am

    I find that increasing the magnitude of the differential to >=1% results in returns consistent with randomness.

  4. vniederhoffer on September 9, 2009 5:36 pm

    that’s the purpose here. deflate ballyhoo. and provide suggestive fruitful things for the productive. vic

  5. Andrew McCauley on September 9, 2009 8:52 pm

    Anton, thanks for doing the work. Much appreciated.


Resources & Links