Monday, January 26, 2009

Strong SOX Action Could Be Good For Nasdaq

I’ve discussed in the past the fact that strong SOX action can often be a good harbinger for the the market. While both the S&P 500 and the Nasdaq failed to gain even 1% on Friday, the SOX rose more than 4%. It’s especially unusual for the SOX to post such strong gains without bringing the Nasdaq composite along with it. It has provided a nicely bullish expectation for the Nasdaq going forward.


3 comments:

Anonymous said...

Rob,

first of all I beg your pardon for not being a native speaker.

I really appreciate your work, but from my perspective I see a few problems with that kind of evaluation because is doesn't differentiate between times frames and doesn't bring the system's odds in comparison to the at-any-time odds for the Nasdaq closing higher x days later.

1. Although the figures may be totally correct, the conclusion (bullish perspective for the Nasdaq looking forward) could be incorrect because there is no differentiation between time frames. The setup produced between 19 up to 23 trades (which is a bit questionable itself because it means that 4 trades must have been triggered during the last 20 trading days which couldn't have been closed up to now).

Based on my own analysis (Nasdaq 100, not Nasdaq Composite), during the last 7 years only 3 trades were triggered, all looser (all three trades within the last 252 trading days, the last one on January 6, 2009). The majority of winners which seems to make the system profitable were almost all triggered during the bull market between 1995 and 2001, an environment which obviously isn't comparable to the current market environment (all three signals triggered during the last 7 years were looser). Would be interesting if you could check the system again with data from 2002 until present and/or data from 1995 to 2002.

2. The percentage of winning trades and the Profit Factor should be brought into perspective in comparison to the at-any-time odds for a Nasdaq closing higher x days later (without any setup) in order to verify if there is really a tradable edge based on a statistical relevant sample size and significantly above at-any-time odds probability and profitability.

Example:
- SOX rises 4% while the Nasdaq 100 can't even manage a 1% gain
Buy Nasdaq 100 on close. Sell 3 days later.
Since 1995: 16 occurrences, 7 winner and 9 looser (43.75%)
The at-any-time odds for a Nasdaq 100 close higher 3 days later since 1995 are 54.37%, much better than the system's odds (although the system's profit factor would be better than the at-any-time profit factor).

My conclusion: This kind of abnormality (SOX rises 4% while the Nasdaq 100 can't even manage a 1% gain) is of course worth a deeper evalution, but should be evaluated during different time frames (e.g. the last month, 1 year, 3 years, 10 years, all-time) in order to check if the system only works in a specific market environment (e.g. bull or bear markets) and should be brought into perspective against the at-any-time odds for a Nasdaq closing higher x days later in order to verify if there is
any tradable edge at all (or if randomly choosing an investment date would deliver better results).

Thanks a lot again for your outstanding work and service.

Best regards
Frank

Anonymous said...

i remmember seeing this data about 3 months ago. When it came out the mnarket totally blew chow after that. I beleive it was back in may. I think we rallied for 1-2 days after and then imploded I say this not to try to invalidate the data but more from the perspective of does this seem to perform stronger in a bull than a bear market ?

Anonymous said...

The setup produced between 19 up to 23 trades (which is a bit questionable itself because it means that 4 trades must have been triggered during the last 20 trading days which couldn't have been closed up to now).

I believe you are interpretting this incorrect. I'm pretty sure Rob is filtering out overlapping signals. Obviously there weren't four such signals in the past month.