Yesterday, SPY closed up 1.6% over Monday’s close. After a large percentage gain, what has the next day and subsequent days looked like?
The Rules:
- Buy SPY (or $SPX) at Close if It Will Close Up > 1.5%
- Sell at the Close X Days Later
- No Commissions or Slippage
- All SPY history used
- $SPX history goes back to 1960
The Results:
Results for $SPX are above. Note that the next day gains average around .18%. These results reflect over 600 samples with trades going back to 1960, but since $SPX is not tradeable, the results should not be trusted completely.
SPY results are above. They are quite different from the results for $SPX. There were about 300 samples with trades going back to 1993.
Analysis of Results:
I have broken out the trades that closed above vs. below the 50 day moving average. Note the difference in volatility of the trades beneath the average against those that closed above the average. Bearish markets are more volatile.
- I don’t trust the $SPX results completely as it is not tradeable. However, it does appear that over the past 50+ years, the index has tended to follow a large percentage gain with more, albeit smaller, gains.
- SPY shows a different picture. It has tended to consolidate or trade in a range for almost two weeks after a large percentage gain. It has tended to give back some of its gains the next day.
- More than anything else, perhaps the difference between $SPX and SPY can be explained by observing that this setup used to work well over a long history, but over more recent history has not worked as well.
Observe the equity curves below. These were generated by buying at the close on the day the setup occurred and selling at the next close.
$SPX equity curve is above.
SPY equity curve is above.
What stands out the most is how bad this setup performed in 2008. It has never really recovered from that drop. Since SPY has fewer data points to average together, the past 3 years has really weighed down the average trade. $SPX, going back more than 50 years, has more than twice the data points. Hence, the period from 2008-present has not weighed down the average to the extent that is has on SPY.
One thing is for certain: this is not a setup that one should be trading in an environment that is highly mean-reverting. Other than that, this study has not left me leaning strongly in either direction.
Nice work Wood. Have you read Wayne Whaley’s stuff about the Inverse Zweig Breadth Thrust that occurred 2 days ago? If so, would be curious to read your thoughts. The reverse signal proves to be interesting stuff in theory.
Also worth noting that a day like yesterday is very rare in that while the spoos were up 1.6% breadth was under 6:1. That hasn’t happened since early 2009.
I haven’t heard of the WW stuff. I’ll have to google it tonight.
I was aware of the breadth divergence but didn’t have time to test it last night. Thanks!
Thanks, your post help me set mild expectations for my day trading today. Went very well.
dis, thanks for the feedback. I love to hear those kinds of things. Glad it helped today.
Why the divergence in SPY and the underlying index? I guess my question is more in regards to the magnitude as opposed to the existence.