Friday, March 27, 2009

Two Strong Up Days Under The 200 Revisted

For the 2nd day in a row the market finished up strongly on Thursday. Over a year ago on the blog I posted a simple study that looked at S&P performance following two consecutive days where it rose at least 0.75% and closed below the 200ma. At the time the 0.75% hurdle was a good sized 1-day move. In the last year a move of that magnitude hasn’t been significant. Still, I thought it would be interesting to go back and run the results again using the same parameters.

These results are similar to the ones I showed last year. In fact the negative influence is now even more pronounced. To see how poorly the market has performed under these conditions in the last year I re-ran the test to show just the time period since the original blog piece:

The fact that the 0.75% hurdle has become easier to achieve hasn’t weakened the bearish influence of the setup. I attribute a large part of reason for this to be the exceptionally choppy environment the market has been in over the time period.


Anonymous said...

Rob, I have two intelligent questions and one very dumb one.

Would a filtering variable such as CURRENTLY ON A BREADTH-THRUST CONTINUATION SIGNAL or NOT // OR // CURRENTLY IN A MULTI-90% UP DAY CLUSTER ENVIRONMENT or NOT -- (or somesuch filter capturing the recent powerful short-intermediate upside) skew the results toward the positive side of the ledger?

Would a filtering variable such as CURRENTLY “STANDING ON A STILT” (meaning overbought on an already vicious near-parabolic straight-up move) or NOT skew the results toward the negative side of the ledger?

Now for the really dumb question. How come there are 106 trade results one day later, but only 67 trade results ten days later? Did the Earth and Time come to an end on some occasions? If there is no price or RS target, merely time elapsed, shouldn’t there be 106 results 10 days later as well?


Rob Hanna said...


The answer to both of your 1st 2 questions is "probably".

Question 3: Because there was a significant amount of overlapping signals. Only the 1st one was taken in the event of overlap. Typically if you take more than 1 signal at a time it will make the results look evn more extreme (in this case bearish).


Anonymous said...

OK, makes sense and explains it.

Now, to make your life even more miserable, and your chances of leaving your computer screen and enjoying some outdoor sunshine even more remote.. what about running a filtering variable reflecting Seasonality-- both the Good Dow/Bad Dow favorable/unfavorable 6-month periods (originally researched by Norm Fosback and Yale Hirsh), and the monthly end-of-month into beginning-of-next month Seasonality periods..?

Considering that the formal Fosback Seasonality System has won 20-year risk-adjusted (Sharpe Ratio) performance awards from the Hulbert Financial Digest, this is a non-trivial question.

There IS undoubtedly such a seasonal effect-- and we are right now entering a double-favorable Seasonal period (time of year, turn of month) over the next few days.. so it matters realtime, and not just academically...


Rob Hanna said...

I'm swamped today and won't be able to go into details here. Not sure I'd have much to add about the best/worst 6 months strategy.

With regards to end of quarter markup you may want to check out a post I did 1 year ago:

Bain said...

Hey Dan, why don't you run those filters and get back to us with the results.

Anonymous said...


Why don't I run the tests? Because, alas, I'm not a quant. I still do all my Market calculations on an old HP 12C hand-held calculator.

I wish the above were a joke but it's not.

I didnt necessarily mean now; I meant my comments mainly as a conceptual alert to fellow viewers of this blog, and to you, as we navigate the next few days.

However, I do think that in your "liesure" (LOL) time it might be worth pursuing.

I've always been fascinated by the question of WHAT CONSTITUTES A 'SIMILAR' ENVIRONMENT?

If there are any other quants who frequent this blog and possess anything like Rob's means and prowess who want to take up
Bain's suggestion, I would be first in line to absorb what you come up with...


Anonymous said...


Just got to check out your year- ago posting, referred to above.

Actually, the "Seasonality System" of Fosback and Hirsh, as refined by Sy Harding and Mike Burk (the OTHER Mike Burk, not the Michael Burke of Investors Intelligence), has nothing to do with End of Quarter markups. I don't know if such marking up is actual, or urban legend. The Seasonality System is a gradual, actuarial reality having to do primarily with end of month cash flows due to automatic reinvestment of dividends in investor holdings.

If there ARE quarterly window dressing effects, they would be absorbed automatically into the actuarial reality of the Seasonality System-- but only as a minor factor.

However, since Fosback was able to create objective Rules of entry and exit based on the tendencies, Mark Hulbert has been able to track it like any other 'strategy' in his Newsletter and in his Marketwatch column. And 20-year risk-adjusted performance awards don't lie. So it's fair to say it's part of the landscape out there...