These type of posts, what happens after 3 consecutive higher closes, are perennial favorites. Yes, many bloggers out there cover this, but I believe it is helpful to revisit the studies from time to time.
The Setup:
The market has recorded 3 consecutive higher closes AND has simultaneously recorded 3 days where the closes were all higher than the opens.
I think most traders are familiar enough with the power of mean-reversion that they understand that consecutive higher closes typically lead to a pullback. That is the easy part of this. What I’m more interested in is what happens AFTER the pullback? Since we are watching to see whether or not the market has bottomed, we want to see the uptrend resume after a short pullback.
The Rules:
- Setup = When SPY has closed higher 3 consecutive days in a row AND each close has been higher than the open, buy SPY at the close.
- Sell at the close X days later.
- No commissions or slippage included
- All SPY history used
The Results:
Summary of Results:
Sample size was healthy on most of these tests. The 2 tests with the smallest samples were Setup & C<MA50 (28 trades held the full 20 days) and Setup + All Variables (6 trades held the full 20 days).
As I tend to do, I added additional factors to the test to more accurately model the current market. All tests show that a slight pullback is likely (as we expected). More importantly, after some reversion to the mean, the results suggest the market may continue higher.
- The worst short-term performance resulted from adding the additional factor of a close beneath the 50 day moving average.
- Yesterday’s volume was higher than the 50 day moving average. Adding that factor to the setup changed the results only slightly. Strangely, it resulted in the worst next-day performance.
- Requiring the close to be above the 200 day moving average resulted in the best performance. It resulted in the most shallow pullback, the least volatility, and the highest average return after 20 days.
What happens when we add together the setup with all additional factors? Doing so creates the most accurate model of the current market. However, being so specific, there were very few instances of this, meaning the sample size was very small. I put the results on a different chart because the other chart got very busy.
These results were the best of all, primarily because one of the 6 trades resulted in a gain of over 9%. The large gain occurred in 1997.
Bottom Line:
After some weakness, which we would expect after consecutive higher closes, the results suggest that the short-term uptrend may continue.
How influenced are “buy after x days up” studies like this by the recent two year rally? Buy after 1-day up, 90% profitable…? How does it compare to the same sample size based on random draws? I know, always making work!
Jeff, this is the first time this setup + all additional variables has occurred since 2007.
The basic setup worked very well until 2007. Since then, results have been poor.
Sample sizes are spread fairly evenly throughout the SPY history. I do not really see more occurrences over the last 2 years than during previous history. If anything, the setup has been occurring less often.
OEW just got a simultaneous uptrend confirmation and WROC buy indicator which usually precedes the uptrend confirmation … last time that happened was July 2010 which was a good time to be/stay long except for the August consolidation from where we rallied straight up about 300 points.
Wood,
I know how to write setup code in Ami but don’t know how to write the “Number of days held” code. Coiuld you throw a guy a bone? Also I assume you run each setup individually, correct?
Regards,
John