The futures are in full melt-up mode, but this reliable indicator suggests there is weakness ahead.
The indicator is very simple. If SPY is trading beneath its 200 day moving average, short it at the close when there are more than 3000 stocks trading above their 5 day moving averages. The short is closed when the number of stocks trading above their 5 day moving averages falls beneath 700.
The win rate for this indicator is 81.25% and the average trade is 1.78%.
The arrows show the shorts (red arrows) and the covers (green arrows). The red line in the lower pane shows the indicator.
wow! nice indicator!
Good stuff.
Reminds me of Larry Connor’s RSI(2) and RSI(4) SPY system where he used a similar set-up (though with the RSI(2-4). I like short term reversion to the mean…especially on the SPY. The QQQ has been tricky for me though. Seems to get crazy and trend more than revert at times.
What does DeMark say now? He warned that if the up move was too strong, it would likely signal a sharp downward leg ahead.
DeMark is too busy counting 13s
Like the thinking. What’s the universe for the 3000 stocks? NYSE, all stocks?
that’s a great question re: universe for the 3k stocks.
Could be issues if using just NYSE
The universe is all major exchange listed, i.e. no OTCBB, stocks meeting a volume/liquidity criteria. I cannot remember what the exact criteria was that I used, but I’ll check in a bit and update.
The universe is all stocks with a 50 day average volume greater than 100K shares and a liquidity greater than 1,000,000 where liquidity equals the 50 day average price X the 50 day average volume.
what is the trade sample size of the study? thanks!
Sample size was 32 trades (going from memory). Sample size is difficult with this count type of indicator since as you go farther back in time, there are fewer stocks, and so that messes the count up. If we lowered the requirement to say 2500, we would have more trades, and they would go farther back in history. Does that make sense?
Woodshedder,
it makes sense. I often wonder how the sample size makes a difference i.e. 32 vs 50 vs 100 vs 200 etc…I see a lot of studies on your blog and other blogs with sample size in 10s, not in 100s. Can you share some thoughts on that? Perhaps there is a way to quantify how sample size affects the outcome, not sure if this makes any sense…
Thanks,
tihfa
PS Your persistent and frequent blogging has been very encouraging and motivating. Hope to have a blog of my own one day to learn more and help out some day. Thank you for your time and energy!
Sample size is a huge issue, as it should be anytime we are using statistical analysis.
Assuming a standard (Gaussian) distribution, 30 samples should be sufficient. As you know, markets do not follow a standard distribution, so in reality, I have no idea whether 30 is a large enough sample to generalize results.
To make matters worse, abnormal markets, which typically happen during extreme bear (and sometimes bull) phases, are rare. Thus, because they are rare, there are simply not very many samples to be had…So, during times when accurate modeling is most needed as the chances for profit and loss are the highest, we have the smallest amount of samples to test.
I don’t know how to quantify this. But I can tell you that average market returns with small sample sizes will be more volatile than with large sample sizes, fwiw.
Glad you enjoy the blog!
One way I have been trying to handle all of the trades regardless of sample size (whether 10s or 100s ) is by dropping to lower time frame and studying the price behavior especially for trades lasting 1-5 days i.e. RSI, DV, MA cross over type setups…Most of meaningful moves take about 1-2 days for short term trade setups from what I have observed. I have been toying with several setups and they all share same price action underneath…BTW I am in the midst of developing an automated mechanical system on Trade Station platform for index trading. Perhaps once that is all done will be starting a blog or a trading journal…or I should while still under development…
Take care,
tihfa
You might have noted really fascinating points ! ps nice internet website .