Monday, February 15, 2010

SPY Daily & Weekly Models

 On Sunday I published a 5 year study of the SPY Weekly Model.  If you look closely, that Weekly Model generated a little over 20 SPY points over the course of the past 12 months, a return (without leverage) of about 26%.

Below is a table of SPY Daily Model signals for the same 12 months, only instead of taking all signals, this model only trades when both the Daily and Weekly SPY Models are in agreement. 



These results address the advantage of trading only on the side of the dominant trend, in this case, the Weekly Model.  This is admittedly a small sample, but my historical testing supports these kinds of results in all indexes, stocks and ETF's for which I've run this simulation.  Not all of this backtesting is ready for publication, but this one is and I wanted to get it out tonight, since both the SPY Daily & Weekly Models are in sync going into this week's trading.


SPY Weekly Trend Model


This represents another way to use the Trend Models, combining multiple time periods to filter out lower probability trades.  You can do your own math on using the double (or triple) beta funds to trade these signals.  There are many more ideas along these lines which I will be sharing as we go forward into 2010.


A

15 comments:

Mike said...

I like this high probability approach...thanks for posting Allan.

Only drawback I can see is that it keeps you out of an initial pivot point after a serious drawdown...

In other words....the +-March 13th '09 Daily would have been missed...until the Weekly caught up with it....


Even so...across a broad spectrum of tradeables...I really think this is the way to go.

Thanks again.

-Mike

Mike said...

with regard to Mar 13th reference above...

I don't mind being a little late to the party....(waiting for being in sync with Weekly)

especially when fiesta of shorting it down prior to that was in play.

a conservative, middle of the road, approach. Plenty of money there.

-Mike

Anonymous said...

Allan at the beginning of the year you posted the daily performance for the year for TZA. Can you post the daily performance for the SPY, without reference to the Weekly signal, so that we can see the pure daily signal results and compare apples to apples with the SPY weekly and SPY daily filtered with the weekly.

Thanks,
Paul

Allan said...

Paul,

The stats for Daily-only Model for the past nine months:

23 trades (versus 13 for filtered model)
57.10 total points (versus 57.21)
avg trade +2.70% (versus +5.29%)
win percent = 61% (versus 77%)

In other and grossly simplified words, the filtered model traded about 1/2 as much with the same net points.

Anonymous said...

thank you

Anonymous said...

I have some SDS Mar. 38 calls which are under water given todays market performance ( as of 12:20).

It looks at this point that the daily(SDS) signal will switch from long to short while the weekly will hold short- a classic conundrum.

Allan,what is you view given the mixed signal situation?

Regards,

Harvey

Allan said...

Harvey,

Playing options on the Daily Model is problematic, high reward, high risk. You almost HAVE to use the Hourly Model to filter those trades, which would have had you out at Friday's close or at the Open today. At this point, its a matter of damage control and now knowing a whole lot more about your situation, I can't really give you anything meaningful. Write to me privately if you want.

A

Anonymous said...

Thanks Allan. My position is small so I'll hold tight given what the model says albeit my window is 4 1/2
weeks,

Harvey

Allan said...

I did a simulation of the above table using END-OF-DAY closing prices for entries and exits. The results were as expected, a little less robust, but still very respectable. In summary, the 12 month return was 41% (in lieu of 68.7) which is about 82% if using a double beta vehicle, like SSO/SDS.

A

Anonymous said...

OT: NNVC

Hope Allan doesn't object to my exhuberance but i am about to make honey in my pants!.

Bernhard said...

NNVC/GNVC(ISCO)up.

EDZ/FXP down.

Maybe you should post about the relevance of the Berkeley JV. Might be interesting for your readers.

You really think we are heading for bad times? ie Buy EDZ/FXP...

Great blog, still following everyday.

Kind regards,
Bernhard

Anonymous said...

To the above..

Last time I expressed exuberance to the surge in NNVC, RC called me down on it in a very derogatory fashion...
I only say to you...

Always happy for good news!

Let us all hope this is the beginning To all our hopes and dreams...

Anonymous said...

Bernhard,

Do you think we are heading for bad times?

Bernhard said...

Always hard to judge.

I've been reading alot and hearing all sides... I get lost in the information. While the market seems to be hallucinating, some advisors are saying sell everything. Hold cash. We are heading down. But what news will hit us before this happens.

Greece, kicked out of EU?...
War fails in Afghanistan?...
Another terrorist attack?...
Another bubble (secondary recovery bubble)...

I honestly dont know anymore... do you?

Mike_T said...

I'm curious to see how the hourly model does when filtered by the daily and/or weekly as well.

I like the higher percentage of winning trades this provides; easier on the whipsaw/emotions.

NNVC up 18.68% in a single day on a couple news reports shows how explosive it can be, and likewise reverse with lack of news. Waiting for a good entry as it settles/bases is prudent.