Algoservice Finance Blog: 02/18/12

Choosing a trigger matrix ?

what i understand under a trigger matrix

Would be that peace of code which comes just before the order it self ,
Lets picture that

(we checking 6000 symbols, we dropping all except stong - buy issues leaving us with 10% of the 6000 that is very round estimate of 600 contracts )
(now these 600 go into the back test which is day based 1 year history ca 20 - 50 orders, that reveals all the crap we dropping estimate 2/3 which go negative)
(200 left contracts so from 6000 only 200 are actual yielding positive)
(considering risk ? what we do now is to determine the top 33% of these 200 that mean(gain%) is the floating pct% is the border where we would except a contract in our buy order )
(consider not every day is a entry to every single contract mounting to natural elimination of the left 66 contracts again i have to estimate but i believe per day we have 5-10 possible entries)

Now here it comes

This presents a day pool last 10

+--------+------------+--------+--------------+-----------------------------------------------------+
| symbol | orderDate  | shares | onClosePrice | oneYearBacktestPerformancePct % on 10k     exposure |
+--------+------------+--------+--------------+-----------------------------------------------------+
| CASC   | 2012-02-16 |    225 |        58.72 |                                               32.12 |
| OGXI   | 2012-02-16 |    831 |         14.5 |                                              20.495 |
| PRSC   | 2012-02-16 |    785 |        15.14 |                                              18.849 |
| CBI    | 2012-02-16 |    259 |        44.42 |                                             15.0478 |
| TWI    | 2012-02-16 |    455 |        24.97 |                                             13.6135 |
| PWE    | 2012-02-16 |    517 |        21.97 |                                             13.5849 |
| M      | 2012-02-16 |    317 |         35.3 |                                              11.901 |
| BDC    | 2012-02-16 |    280 |        39.75 |                                                11.3 |
| RDWR   | 2012-02-16 |    337 |        32.64 |                                              9.9968 |
| KLAC   | 2012-02-16 |    214 |        50.68 |                                              8.4552 |
+--------+------------+--------+--------------+-----------------------------------------------------+

For a varity of reasons, personly the lack of funds to engage more then 3 trades a day,or and the trading licence which only allows certain amount of trade a day.we are forced at this point to choose from the pool only a part.
This is most delicate procedureas we have to ensure that the statistical integrity is not compromised at this point.
In order to ensure integrity or ideally to generate a positive emerging property,we have to employ a matrix at this point.So far so good now we arrived at the headline of this article. After all these years playing with these scenarios i am still not hundred percent sure what the ideal solution on this point is.What i'am sure of is that the opinion's go apart as well.Some or most and part of me are thinking,why bother at all simply go with the highest back-test result end of story,this would have certain advantages which are not to dismiss (integrity, simplicity).On the other hand with the right matrix we would be able in theory to harness (adaptation, flexibility, additional yielding ) so lets cook it i have prepared some visualisations which represent a possible course of action.possible scopes (risk,odds,momentum,cycle)