Saturday, February 24, 2007

Detecting Participation in Breakout Moves: A Best Practice in Trading

From Brett: Our next Best Practice post comes to us from Aiva_Trader, who examined the last five months of data in the NASDAQ 100 Index (QQQQ) to establish the odds of continued upward movement once QQQQ breaks above its current day's high. Specifically, the study broke down occasions in which fewer than 30% of NDX stocks were participating in the move (i.e., not surpassing their current day's highs) vs. those occasions in which more than 70% of NDX were also participating in the breakout. Here are the findings:

From Aiva_Trader:

When QQQQ breaks its current day high (all occasions):

The close 1 minute later is up by an average of .026% and closed up 68% of the time.
The close 5 minutes later is up by an average of .030% and closed up 59% of the time.
The close 15 minutes later is up by an average of .043% and closed up 56% of the time.

When QQQQ breaks its current day high and fewer than 30% of index stocks participate:

The close 1 minute later is up by an average of .002% and closed up 68% of the time.
The close 5 minutes later is down by an average of -.028% and closed up 44% of the time.
The close 15 minutes later is down by an average of -.05% and closed up 39% of the time.

When QQQQ breaks its current day high and more than 70% of index stocks participate:

The close 1 minute later is up by an average of .05% and closed up 71% of the time.
The close 5 minutes later is up by an average of .05% and closed up 62.5% of the time.
The close 15 minutes later is up by an average of .07% and closed up 58% of the time.


From Brett: These results fit well with my own research. Breakout moves are more likely to reverse when a large proportion of sectors and individual stocks are not participating in the breakout. This helps to differentiate a market that is topping out (becoming more selective as it approaches highs) from a market that is breaking out (ready to establish fresh new highs). I suspect Aiva_Trader's observations could form a core concept for a trading system, particularly if the signals were filtered by such criteria as whether or not the market was overbought at the time of new highs, time of day of the breakout move, etc. It would also be fruitful to examine holding periods from 30-90 minutes out. My own recent research found that fading new 60 minute lows in SPY and IWM has been profitable at those holding periods. Many thanks to Aiva_Trader for the valuable observation in the NASDAQ market. Knowing the odds that a market has gone your way in the recent past is, for me, a best practice--and a helpful aid in avoiding bad trades.

3 comments:

niklaskoehler said...

Hi Brett,

great idea with the new highs data, which should be followed on important time frames (i.e. number or better percentage of stocks of an index with daily, 5 day, and 20 day new highs) during the day.


Sure, one can estimate that way if a move has substance.

However, where can such data be found?

Thanks!
Nik

Brett Steenbarger, Ph.D. said...

Hi Nik,

New high/low data can be tracked with an intraday screening tool. I use Trade Ideas for that purpose. It's also possible to link a spreadsheet to your real-time data feed and automatically populate fields that will alert you to new highs/lows at any time frame. Tracking the behavior of the sector ETFs as we approach new highs is a quick and dirty way of assessing the breadth of participation. Thanks for the question--

Brett

Cucca said...

For Niklaskoehler, you can also go to the NasDog 100 flash quote site at:

http://quotes.nasdaq.com/quote.dll?page=nasdaq100