Blog - Investing Notes

December 29, 2006 - Follow-up on Streak Reversals: Some Nasdaq Disorderly Conduct

In reaction to our blog entry of 12/26/06 presenting an out-of-sample test of S&P 500 index streak reversals during 2004-2006, two readers pose follow-up questions. David Zaitzeff, a futures broker at Peregrine Financial Group, Inc., asks that we perform a similar analysis on the Nasdaq Composite index. Dennis Page of Beverly Hills MI asks whether we can replicate the entire TradingMarkets.com streak reversal study of seven million backtested trades on individual stocks from 1/1/95 to 6/30/06. Our responses are:

Using Nasdaq Composite index daily closes for 1/2/04-12/22/06, here is additional index "streak" research. The following chart shows the average returns, with subsample sizes (in parentheses) and one standard deviation variability ranges, for one trading day after various lengths of up and down streaks for the Nasdaq Composite index. We calculate returns using the closing price on the day the streak materializes and the closing price the next trading day. Most of the subsample sizes are so small that they are meaningless for statistical inference (and useless as trading rules). The two largest subsamples, for three-day up and three-day down streaks, fail to confirm consistent reversal effects. The average daily return for the entire sample is +0.03%. The average daily return for days after three-day up (down) streaks is -0.15% (-0.18%). Up streaks show reversal, and down streaks do not. However, with the variability (standard deviation) of the post-streak returns much larger than differences in average daily returns, these inferences are weak.

The next chart shows the average returns, with subsample sizes (in parentheses) and one standard deviation variability ranges, for five trading days after various lengths of up and down streaks for the Nasdaq Composite index. We calculate returns using the closing price on the day the streak materializes and the closing price five trading days later. Again, most of the subsample sizes are so small that they are meaningless for statistical inference (and useless as trading rules). The two largest subsamples, for three-day up and three-day down streaks, again fail to confirm consistent reversal effects. The average five-day return for the entire sample is +0.13%. The average daily return for days after three-day up (down) streaks is -0.14% (-0.17%). Again up streaks show reversal, and down streaks do not. However, with the variability (standard deviation) of the post-streak returns much larger than differences in average daily returns, these inferences are weak.

The next chart compares the one-day average reversal returns for the S&P 500 index and the Nasdaq Composite index after various lengths of up and down streaks (without regard to subsample sizes or variabilities). The behavior of the Nasdaq Composite index is most noticeably different after 3-day and 4-day down streaks, showing continued declines on average rather than reversals. Possible conclusions are: (1) Nasdaq Composite index down streaks are more persistent than S&P 500 index down streaks; or, (2) subsample sizes are too small for reliable conclusions (and we are at risk of being fooled by randomness).

The frequency of streaks is roughly the same for the S&P 500 index and the Nasdaq Composite index. There is a hint that streaks tend to be somewhat longer for the latter, but the subsamples are so small that we cannot draw this conclusion with confidence.

In summary, in contrast with the S&P 500 index, the Nasdaq Composite index does not support a finding of systematic streak reversal behavior during 2004-2006.

Replication of the entire TradingMarkets.com streak reversal study on individual stocks exceeds the combination of time, data retrieval capabilities and database programming we can apply. We do see two important extensions to what TradingMarkets.com accomplishes (assuming that the conclusion regarding streak reversal behavior is valid):

1. Checking the correlation of the timing of streaks across individual stocks.

If streaks are highly correlated (occur at the same times due to co-movement with the overall stock market), then there would not be enough independent streak reversal opportunities for an attractive standalone trading strategy. And, there probably would not be enough concurrent streaks in opposite directions to support a zero-cost long-short portfolio strategy.

If the correlation is limited, one might be able to trade streak reversals almost continuously, thereby possibly beating the market. And, if there are concurrent up and down streaks, a zero-cost long-short portfolio strategy might be feasible.

2. Checking for robustness across economic cycles (or bull and bear markets). It may be that trading against up (down) streaks works more reliably in bear (bull) markets.

Tie-ins to theory would also be interesting. Systematic streak reversal behavior would seem to contradict the Efficient Markets Hypothesis.

For research on other trading indicators, see Blog Synthesis: Some Trading Indicators.