OK, since many of us here like to explore trading system, I propose a group project.
Many people know of the trading system Oddball (as featured in the Dec 2000 issue of Active Trader ... see
http://www.oddballsystems.com for background)
In a nutshell, the system measures the Advancing NY issues (rate of change) and buys/sell when the Rate of Change rises above/below certain levels.
On the Omega list, a gentleman named Bob Fulks posted an equity curve and hints about a self-adaptive version of Oddball he had.
The equity curve is here --->
http://www.purebytes.com/archives/omega/2002/gif00033.gif
The post describing details of the system along with the performance report is here --->
http://www.purebytes.com/archives/omega/2002/msg06378.html
Some observations ... the system trades 376 times over 51 months. The average # bars in winners is 28, implying the system trades once every 4-5 days.
I've toyed with different ways of dynamically adjusting the buy/sell levels .. first taking an average of the ROC measurements:
This is Tradestation 6 code, with Data1 as @ES.D and Data2 as $ADV
inputs:
sstart( 0930 ),
eend( 1600 ),
movavg( 30 );
variables:
bz( 0 ),
sz( 0 );
bz=average(roc(c data2,8), movavg);
sz=bz;
if time >= sstart and time<=eend then begin;
if roc (c data2, 8 ) > bz then buy this bar close;
if roc (c data2, 8 ) < sz then sell short this bar close;
end;
The results aren't great, but it does improve limiting the trade time to 2 - 4 pm est.
Some ideas to play with are taking the log of data 2, using the std dev of the roc average to adjust the buy/sell levels, using the highest/lowest roc level over a set lookback period, etc.
Brainstorm, think about different ways this can be done. We can create a strategy that performs well and we all can trade
Cheers,
Rich