[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: More Robust OddBall



PureBytes Links

Trading Reference Links

>I have three new "more robust" versions of the OddBall code with 
>various features. They are available on Mark Brown's Oddball site. 
>There is also a document describing the code. The introduction to the 
>document is quoted below.

Great work, Bob.  Excuse the whine: I only wish you woulda' put that
effort into an EOD oddball system!

'm fooling with that myself, but with very limited success (using
EOD data only, not tick data).  Using just a 1-day RoC and a money
management stop, it works OK when the cash index is traded like a
futures contract.  However it doesn't do as well when I substitute
ES data.  Also, it works for only 3-4 years of history.  I have not
been successful finding parameters that work further back.

This would indicate to me that the market's memory relating to
the advancing issues just isn't there on a daily basis (i.e. what
happened in Advancing Issues yesterday doesn't strongly affect what
price will do today and tomorrow).  In light of that, it is strange
that when I do things with longer-term measure of RoC, I can get
improved performance.

I don't plan to use this EOD-Oddball as a trading system itself, but
rather as a filter for another decent trading system (Swinger2-EOD).
Using the cash index only, combining the two systems gave promising
results.  Using ES data, the EOD-oddball doesn't make a good filter,
however.

Regarding my observation that substituting other than the cash index
for either Oddball or Swinger degrades the performance:  I also
tried substituting the SPY exchange traded fund, with disastrous
results!  The daily-bar correlation between SPY and the cash index
sucks, although the overall movements agree (if the cash index goes
up, SPY will go up, but the high and low versus the open and close
won't have any correlation).  It was such a consistent loser that I
tried fading the signals with SPY, but that didn't work either.

There seems to be an arbitrage opportunity here, but I don't know
what it might be.

-- 
  ,|___    Alex Matulich -- alex@xxxxxxxxxxxxxx
 // +__>   Director of Research and Development
 //  \ 
 //___)    Unicorn Research Corporation -- http://unicorn.us.com