Friday, August 28, 2009

False Breakdown Often Leads to Move Higher

The fact that the market tried to break down from its recent range Thursday and then reversed to close at a new high could be interpreted as bullish. Below is a test that describes Thursday’s action:

(click table to enlarge)


Pretty solid results across the board suggest an upside breakout appears more likely than a pullback at this point.

5 comments:

James Goode - On The Money said...

Looks interesting but big question: of the winning trades, how many began at a multi-month high, versus a multi-month low? Your parameters could be valid at both, and the difference is potentially important. If the winners came just as often at highs as lows then a trade beckons, otherwise... might not be as tempting as it looks!

Rob Hanna said...

Good question and something I normally would look at but neglected to last night.

I checked times it either made a 20-day high or a 20-day low. In both cases the results remained positive 1 and 5 days out. There were 13 instances where the SPY closed at a 20-day high. Performance was still positive but not as convincingly. Avg 5-day return was about 0.5% instead of 0.9%. Only six instances occured with a 20-day low hit before the reversal to close at a 4-day high. 4 winners but the average trade was over a 3% gain. Neither of these tests had enough instances to instill a high level of confidence but it appears the edge may be a bit lower than the original test when near 20-day highs (like now) and stronger when near 20-day lows.

Adam Berkowitz said...

hi rob

i just discovered your site this week and i really like your work...

i am a big fan of attempting to identify market behavior and then building robust systemic trading models on top of them...one of the keys for me to continue to do this successfully is to manage the risk...

i am also a big fan of tradestation for both its modeling and trade execution capabilities...

i believe your studies would have even more value if you would include the 'max intraday drawdown' field in your posted optimization reports...this would help us determine how much intraday risk the model bore while the trade was open...

i wish tradestation allowed user defined fields in the optimization report as well...then we could divide the 'net profit' by the 'max intraday drawdown' to get the intraday risk adjusted profit factor, as opposed to just the closed trade profit factor that we commonly use...i know there are work arounds to coming up with the calculation i am talking about, i just wish we could include this ratio in the optimization report...

in any case, i love your site and have become a big fan of your work...

all the best,
adam berkowitz

Frank said...

Rob - does today's action (opposite of yeterday's) now hold true in the opposite direction?

P said...

a slight variation to the pattern including Friday's action:
on SPY since 1994 when low of yesterday is the lowest low of the past 4 day and today is the highest open of the last 2 days
See results:
http://short-termtrading.blogspot.com/2009/08/spy-trading-pattern.html

It does not look bullish for next week because the close was lower than the open on Friday.