Short Term System Development

[tech/a]

Yep

Found it.

Error is in the UNIVERSE of stocks selected.
I cannot get a profit out of it now.
Not unusual as Ive attempted finding a short term method many times before and yet to find a profitable one (Doesnt mean there isnt).
The results I found un believable as well---this is more believable.

The error was in that I was loading stocks into my universe that I knew triggered a buy over the last year.This is ofcourse hindsite trading as you wont know that prior to your trading.
I hadnt chaged the settings from when I wanted to find stocks that we would have liked to trade,then determining the settings to best trade them.

Simple error ---huge difference.

Hrs of work ruined with one error---better to know now!!!

 

[tech/a]

I have made some comments here after many hrs of testing over a period of 3 yrs with regard to short term methods for those who are interested.

These will be the lines of developement I will now persue.

https://www.aussiestockforums.com/forums/showthread.php?p=16372#post16372

 

[tech/a]

For those interested im still going on this.
Currently eyeballing ALL charts that will be in the universe then making up 2 universes which will in my veiw toss out those with best possibilities.
As base universe I'm again using the BT Margin list (which will disappoint some who will want a method trading the pennies).My reason here as I'm interested in volatility but not regular extremes.Plus a great deal of Fundamental analysis has been done for me.

Universe (1) will be those in strong UPTRENDS with the veiw of hopping on and off.
Universe (2) will be those that are in a current corrective move with the veiw of hopping on when they emerge from the correction,often long term corrections and a basing indicate a longerterm up move.CTX UTB ALL all started from this sort of correction.

Its time consuming and as I dont have a lot of it (time) things from my end are a bit slow.

Anyone else still at this---if so interested in your approach.

 

[Milk Man]

Hi tech,

I reckon it would be better if you used a margin list at the time of testing for more accurate results. If youv'e kept these from tech-trader development it would be good info for all of us. Otherwise, how would we go about obtaining what was on these lists in the past?

As far as my short term system goes , mainly due to having to correct data faults (and I subscribed to my provider for 12 months :swear: )

I thought of testing all stocks, with a liquidity requirment, then only trading them if they have been profitable in the past. To test this I've been testing until 2002, then culling out those which don't return well enough, then testing 2002 until current. The method i'm trying doesn't have enough trades though.I'm going to test on a simple MACD crossover so theres more trades to see what comes of it.

I've also tested a bit of an uptrend formula which works quite well. In my long term formula a stock must make a 12 month high thats 25% higher than the highest high of 12 months ago.

 

[tech/a]

Loakglen

As its a short term method I dont think survivor ship is an issue.
I'm thinking a few days to a week or so.

 

[Milk Man]

Loakglen

As its a short term method I dont think survivor ship is an issue.
I'm thinking a few days to a week or so.

No worries; definitely a problem with longer terms IMO.

 

[tech/a]

We didnt seem to get far with this.
However had a thought which I thought would be interesting.

I thought we could use the tipping comp for our prospects.
My thoughts are that I would look at the selections from a technical view and hold in a portfolio those which I thought to be best placed over the next month.
Id place a stop of 20% of purchase price on each (As this has tested to be the best stop position with smaller caps.) So a 10c stock would have a stop at 8c.
Id have a fixed position size of 5K for each.
Risk is then around 2% if we have 10 prospects---more if less and less if more.
I thought then to use an exit of a variable 3 day M/A stop which signals a sale on a downturn of the M/A not a cross.See Diag.

I will enter them in Stator which I have and be able to update each night.
Over time I can compare months or keep some going (Portfolio's) etc.