Australian (ASX) Stock Market Forum

Short Term System Development

tech/a

No Ordinary Duck
Joined
14 October 2004
Posts
20,417
Reactions
6,356
Re: Developing a System to Trade Live

OK
Ive decided it wont kill me to post on one thread.
This topic interests me and there are enough here showing interest and persisting so if its OK with you I'll se this exercise out.

Ive made headway but still away from completion but I think youll find the direction I have taken as very interesting.

Decided to tackle what I feel are the hardest to trade---the pennies.

My universe is (Currently as its a work in progress) stock which trade under 20c---Ive veiwed many charts and these appear to me to be the ones which have the capacity to fly +100% in the shortest time.

For Entries I like breakouts in particular the FIRST or SECOND breakout of a move after a prolonged accumulation period which was preceeded by a fall---usually a dramatic fall. I'm still playing with days prior to breakout,I'm using 70 days and have played with 180 and 250.There are idiosycronies with breakouts which warrent further discussion and examples shown on charts---but not yet.

Chart 1 shows the "Ideal" entry.Testing is not currently being limited to this entry as to do so would cut back the number of trades dramatically---however if profitability is compromised by not filtering---guess what I'll filter!

For Exits
I like a combination.As they are so fast moving I like to use either an AND or an OR condition in the exit.My initial eyeball observations show me that a single exit condition wont cut it.I need at least 2 to confirm time to EXIT.(I'm actually using Andrews TOX trade as a live exit example to try out---CHART 2 is the chart I private mailed Andrew showing a few ideas.)

The 4 exits I like to combine are.
(1) Simply the NEAR TREND
(2) Near confirmed support
(3) Parabolic
(4) A custom close fitting M/A.

I submit here the Metastock code I'm using to get the feel and determine possible parameters BEFORE I go to the testing phase.I want to have a good idea that there will be a method which has atleast a chance for positive numbers.
Already I see 2 problems with coding so someone maybe able to help out here.
(1) Metastock cant code trendlines in its basic form we need someone to write a DLL to allow this I have some contacts and if necessary will buy it and offer it to those who want it at a dispersal cost.

(2) I havent written a code for a close support line.This is very important from what I see so far.

I also have some code which makes sure you take the FIRST signal will worry about that for testing as we go on.(Ive placed it in an EXPERT(Those with M/S know what I mean))

So far stops dont seem necessary.
The tightness of exits seem to limit losses but there will be more losses than wins.Seems to have a good expectancy but I havent run it yet.
Anyway thats where I am so far.

FORMULAS
EXPLORATION TO FIND PROSPECTS
---this one is for stocks that broke out in the last 63 days ---can be altered to X days.

Alert(HHVBars(H,70)=0,63) AND Fml("liquidity")>100000 AND C<.20

Ive set liquidity at $100K dont see this as a problem.

Paste below into indicator builder.

{"liquidity" } Mov(V * C,21,S)

Custom Moving Average 4 times smoothed{ ©Copyright 2005 John Rowland}

Mov(Mov(Mov(Mov(Mov(C,2,E),4,E),4,E),4,E),4,E)

{Parabolic Arc}

Z:=Input("Periods",2,250,4);
Mov(SAR( 0.02,0.20 ),Z,E)


To the charts
and feedback welcome.
I'll be back again when I have something more to input.
 

Attachments

  • Short term method.gif
    Short term method.gif
    34.5 KB · Views: 524
  • tox 1.gif
    tox 1.gif
    36.2 KB · Views: 508
  • Short term 1.gif
    Short term 1.gif
    22.3 KB · Views: 568
Re: Developing a System to Trade Live

tech/a said:
OK
Ive decided it wont kill me to post on one thread.
This topic interests me and there are enough here showing interest and persisting so if its OK with you I'll se this exercise out.

Ive made headway but still away from completion but I think youll find the direction I have taken as very interesting.

Decided to tackle what I feel are the hardest to trade---the pennies.

My universe is (Currently as its a work in progress) stock which trade under 20c---Ive veiwed many charts and these appear to me to be the ones which have the capacity to fly +100% in the shortest time.

For Entries I like breakouts in particular the FIRST or SECOND breakout of a move after a prolonged accumulation period which was preceeded by a fall---usually a dramatic fall. I'm still playing with days prior to breakout,I'm using 70 days and have played with 180 and 250.There are idiosycronies with breakouts which warrent further discussion and examples shown on charts---but not yet.

Great to have you back tech!

Firstly, if you need to start a new thread to get the Metastock code help sorted out just go ahead and some lurkers may reply quickly.

As for the strategy, my first thought was, yikes! we tried this in the penny stocks breakouts thread and it was tough. But hopefully this system will be a tradeable one.
In terms of the breakouts, should we find more 'reliable' accumulation patterns (eg rectangular consolidations prior to breakouts, or ascending triangles with corresponding volume) to narrow our focus further? Are you talking of pullbacks after a breakout? (ie 'dramatic fall'). Also, perhaps a filter to note volume increases (eg increase in activity but not quite so far as to make it a breakout- this may signal accumulation starting in earnest (like in BMX over the last six months). Hope I'm not jumping the gun or missing the point.

BTW, I couldn't see the magenta line in the second chart, is it your custom one?
 
Re: Developing a System to Trade Live

tech/a said:
My universe is (Currently as its a work in progress) stock which trade under 20c---Ive veiwed many charts and these appear to me to be the ones which have the capacity to fly +100% in the shortest time.

I've found some tipsheets that specialise in junior resources co's purposefully look for stocks around 10c or less with some milestone in reach (eg about to finalise bfs or about to start exploration) knowing that psychologically some punters will buy upto 20c on pure speculation or hype. That type of behaviour would also help us as 20c (a round number) is a strong psychological level in most cases (the minimum for a float on the asx?) and once breached can see a quick spurt up. In terms of the universe would it be necessary to only include stocks with a year or two of trading history?? You'll be getting a fair few recent floats with that price level.
 
Re: Developing a System to Trade Live

Rich.

Dont get to excited I only want to post on this thread!

Anyway the magenta line is the custom indicator.The M/A.
New floats wont be picked up as the buy signal is (Currently) a breakout of a 70 day high---if it hasnt traded 70 days then it wont come up.

Perhaps a bit of an explaination as to how I develope the methodology to formulate a system will help those who struggle with getting past the first step.You'll note Im not testing anything YET.

I write down what I wish to achieve.
In this case it is ]{Catch fast moving cheap stocks that fly and jump in and out for maximum gain}
Sounds easy!!
Then I get a heap of charts that do that and spend hrs looking at them identifying common factors.
What can I see thats common with each or MOST charts.

I WORK BACKWARDS

Then SETTING the FORMULAS of DISCRETIONARY ELEMENTS to what I see in the charts

Before I want to run any tests I want to know what it is Im going to run.
This is labourious but this is WHY we will get a result in the end---IF THERE IS ONE.Working the other way around will lead to a great deal of frustration as you GUESS which parameters COULD work---sorry guys but thats what I could see happening----no offence.

Im still at this point of eyeballing and testing (Through placement of an expert I have written and continually fiddle with!)on dozens of charts.
As you can see I feel I have identified elements that have the potential to achieve what I want.

From here I need to know if its practical---does it give me enough trades, will it be liquid enough,what sort of bank will I need,do I need to take EVERY trade,Is there enough trades---etc etc.

Techtrader took 7 mths to design and implement---2 other great methods came from that--this I hope will be similar.
The path I have chosen may also end up to be a dead end----But it feels good so far! Thats encouraging to me.

Interested in feedback and TOX is a live work in progress if I find another Ill post it up to see how it goes.
I'm watching a few AND(Thats its code AND) being the best but is a second stage breakout!
 
Re: Developing a System to Trade Live

I have an interest in a short term system as well. Do we finish this and start on a short term system? I have been doing some playing with it and the results are coming up okay with extra filters.

MIT
 
Re: Developing a System to Trade Live

Gee, I'll have to refresh more often. I was replying to a message 5 messages ago.

Tech, this looks like something I was doing for awhile a couple of years ago. Looking for shares that have formed a base with the volume slowly growing and the price slowly moving up. I would buy before the breakout with a tight stop and wait. I'd have 5-6 positions and there would be something happening at least once a week.

It went pretty well but it doesn't trade as well as it looks when you try to sell once the volume drops as the spread can be relatively wide and they gap all over the place. The daily high and low can be achieved the the bid/offer gap
So what looks nice on the EOD data is difficult to achieve.

However, saying all that I remember on a forum a few years ago that somebody turned 10k into 100k trading like this (Liquidity wouldn't let him go further).


MIT
 
Re: Developing a System to Trade Live

mit said:
I have an interest in a short term system as well. Do we finish this and start on a short term system? I have been doing some playing with it and the results are coming up okay with extra filters.

MIT

I thought the intent from the start was for a short term system in this thread as that's what was discussed in the initial posts. Perhaps we can continue this thread as is and any long term systems can be started via a new thread. We are still in the early stages but because tech has posted some detail at least we have some direction (we seemed to be losing it there for a bit) and a theory to start with.
 
Re: Developing a System to Trade Live

I encourage those working on the exercise to find and research their own,long term or short term method.

If however your struggling for direction I welcome all the help I can get.
Right at this point I am leaning towards MIT's statement that the move I'm trying to catch and that I see in many charts in reality is very difficult to catch.This could be limited by coding ability but also practicality.Unless we investigate it we wont know.

I want people to be the devil's advocate and if developing a system so do you! If someone finds a hole in what your doing and you cant fill it--- they may well save you many $$$s.

So if you see something that doesnt add up for YOU post!

My idea is but one.

HINT
Perhaps if people post charts of MOVES THEY WANT TO CATCH then that could also act as a starting point.(Thats the way I identify what I want to trade).

Ill post up some other ideas (Charts) and ideas later when I have more time.
 
Re: Developing a System to Trade Live

RichKid said:
I thought the intent from the start was for a short term system in this thread as that's what was discussed in the initial posts. Perhaps we can continue this thread as is and any long term systems can be started via a new thread. We are still in the early stages but because tech has posted some detail at least we have some direction (we seemed to be losing it there for a bit) and a theory to start with.

Yeah Rich,

It might be an idea to divide all the posts up between the two and rename one short-term and one long-term. I personally don't care which direction we go: as long as there is direction! That was frustrating me a bit. Perhaps any future decisions regarding direction should be polled or a panel formed and they vote.

:2twocents
 
Re: Developing a System to Trade Live

loakglen said:
Yeah Rich,

It might be an idea to divide all the posts up between the two and rename one short-term and one long-term. I personally don't care which direction we go: as long as there is direction! That was frustrating me a bit. Perhaps any future decisions regarding direction should be polled or a panel formed and they vote.

:2twocents

Sounds good to me, any direction is much better than none.
 
Re: Developing a System to Trade Live

Come on guys direction is pretty simple.

PICK ONE.
Ive picked mine---for the time being.

One of the greatest phsycological drawbacks for any trader is making a decision,a decision to buy and worst of all that decision to TAKE A LOSS!

Procrastination isnt the road to Profit!
 
Re: Developing a System to Trade Live

tech/a said:
Come on guys direction is pretty simple.

PICK ONE.
Ive picked mine---for the time being.

One of the greatest phsycological drawbacks for any trader is making a decision,a decision to buy and worst of all that decision to TAKE A LOSS!

Procrastination isnt the road to Profit!

Hey, welcome back. Nice to have you back on board. :)
 
Re: Developing a System to Trade Live

I'll do both and run both so when theyre as not to put all my eggs in one basket. Be good to do while the gloss hasn't worn off systems testing. We seem to have enough background in different facets of this game so It shouldn't be too hard. Be intreesting to see all the variables between the two.
 
OK

Ive proceeded to testing.
These are preliminary tests and by no means exhaustive

So far I'm testing 3 universes over the last 7 mths to 1/1/05
(1) Stocks to 20c
(2) Stocks to 50c
(3) Stocks to $1

In singular testing ALL are profitable.
In multiple MonteCarlo portfolio testing only (1) is 100% profitable.
The other 2 have around 4% chance of failure and those that fail are only around 5% max.

The results for the 20c stocks I found remarkable with over 100% anualised return on 10K

Ive called the system PENTEKK---(Terrible I know).

Detailed Report
(PENTEKK NS 70 20)

Simulation Summary
Simulation Date: 7/20/2005
Simulation Time: 7:39:09 PM
Simulation Duration: 0.08 seconds

Trade Summary
Earliest Entry Date in the Trade Database: 1/4/2005
Latest Entry Date in the Trade Database: 7/19/2005
Earliest Exit Date in the Trade Database: 1/12/2005
Latest Exit Date in the Trade Database: 7/20/2005

Start Trade Entry Date: 1/4/2005
Stop Trade Entry Date: 7/19/2005
First Entry Date: 1/4/2005
Last Entry Date: 7/14/2005
First Exit Date: 1/19/2005
Last Exit Date: 7/20/2005

Total Trading duration: 197 days

Profit Summary
Profit Status: PROFITABLE
Starting Capital: $10,000.00
Finishing Capital: $15,621.30
Maximum Equity/(Date): $5,621.30 (7/20/2005)
Minimum Equity/(Date): ($745.34) (3/24/2005)
Gross Trade Profit: $7,498.52 (74.99%)
Gross Trade Loss: ($1,877.22) (-18.77%)
Total Net Profit: $5,621.30 (56.21%)
Average Profit per Trade: $208.20
Profit Factor: 3.9945
Profit Index: 74.97%
Total Transaction Cost: $1,620.00
Total Slippage: $0.00
Daily Compound Interest Rate: 0.2267%
Annualized Compound Interest Rate: 128.5165%

Trade Statistics
Trades Processed: 230
Trades Taken: 27
Partial Trades Taken: 0
Trades Rejected: 14
Winning Trades: 15 (55.56%)
Losing Trades: 12 (44.44%)
Breakeven Trades: 0 (0.00%)

Normal Exit Trades: 20 (74.07%)
Delayed Normal Exit Trades: 0 (0.00%)
Open Trades: 7 (25.93%)
Protective Stop Exit Trades: 0 (0.00%)
Time Stop Exit Trades: 0 (0.00%)
Profit Stop Exit Trades: 0 (0.00%)

Largest Winning Trade/(Date): $2,813.61 (4/1/2005)
Largest Losing Trade/(Date): ($260.41) (4/7/2005)
Average Winning Trade: $499.90
Average Losing Trade: ($156.44)
Average Win/Average Loss: 3.1956

Trade Duration Statistics
(All Trades)
Maximum Trade Duration: 87 (days)
Minimum Trade Duration: 2 (days)
Average Trade Duration: 23 (days)
(Winning Trades)
Maximum Trade Duration: 87 (days)
Minimum Trade Duration: 6 (days)
Average Trade Duration: 32 (days)
(Losing Trades)
Maximum Trade Duration: 24 (days)
Minimum Trade Duration: 2 (days)
Average Trade Duration: 12 (days)

Consecutive Trade Statistics
Maximum consecutive winning trades: 5
Maximum consecutive losing trades: 5
Average consecutive winning trades: 2.50
Average consecutive losing trades: 2.40

Trade Expectation Statistics
Normalized Expectation per dollar risked: $2.7000
Maximum Reward/Risk ratio: 43.45
Minimum Reward/Risk ratio: -2.13
Average Positive Reward/Risk ratio: 5.92
Average Negative Reward/Risk ratio: -1.29

Relative Drawdown
Maximum Dollar Drawdown/(Date): $824.19 (3/24/2005)
Maximum Percentage Drawdown/(Date): 8.3010% (3/24/2005)

Absolute (Peak-to-Valley) Dollar Drawdown
Maximum Dollar Drawdown: $895.34 (8.9530%)
Capital Peak/(Date): $10,000.00 (18000101)
Capital Valley/(Date): $9,104.66 (3/24/2005)

Absolute (Peak-to-Valley) Percent Drawdown
Maximum Percentage Drawdown: 8.9530% ($895.34)
Capital Peak/(Date): $10,000.00 (18000101)
Capital Valley/(Date): $9,104.66 (3/24/2005)

I tried a 40 day breakout and that to was profitable but not as spectacular.
These results are based on a 70 day breakout.This has implications for trades that fail and then re trigger (The new high has to be the highest for 70 days.)
Big retracements mean that this can take some time.Shorter timeframes will trigger earlier as after 40 days for example a large retracement will no longer be in play.

Lots and lots to go yet but initially encouraging---I'm now more confident that this can be cracked!!

I have slightly altered the variables but as Im fiddling wont publish them yet unless som amibroker person wants the master to tweek.

The pick is the last 7 mths P/L for the 20c PENTEKK thought a visual would be more explainatory.

There are some interesting aspects with the results which are highly predictable and I'm glad to say that other than the profit (Which is remarkable) the rest ---given the profit is as expected.

Of interest
(1) Winners around 50%--slightly more--good given the universe hasnt been filtered.
(2) Reward to risk high at 5:1
(3) Average number of winners and losers in a row 2.5---no drawdown probs.
(4) Maximum losers in a row 5--wont go broke here either as set risk to 1% of capital.---If it doent go then we are out.--wider stops kill us! (I tested 2%).
Set each trade as 33.3% of capital and capital $10,000---seems to work OK but will do many many more tests before I'm happy.

Back again when I have more info could be around weekend before I get back to this.
 

Attachments

  • PENTEKK 1.gif
    PENTEKK 1.gif
    29.7 KB · Views: 450
Very encouraging results!

As someone who's had experience in building a system before, what direction do you plan to take from here to refine the system. The knowledge of how you go about building this is just as important as what you've built.

Have you considered limiting in the way you've since considered with techtrader - limiting entries to periods when say, the XSO is trading above its 30EMA and having a built in exit also based around XSO?
 
Doc

Good question and a little farther down the track.

Thinking out load though with an average winning trade length of 32 days and losing of 12 my immediate thought is "whipsawing".
Some sort of filter would be advantagious---well I think it would.

Colby in "The Encyclopedia of technical indicators" found significant gains when filtering with the DJIA.
The whole gambit of composite index compilation and use as timing filters is in its infancy--many are just being introduced to it.
Software has to catch up.

Amibroker can compile Composites and Justdata can do it pre distribution.

In answer to your question specifically Doc---it will be something that I will look at as a final possible improvement to the system.
Frequency of trades will also be an issue as there were only 27 trades in 6 mths ---about 1 a week.

Refining does get to a point where practicality must be addressed.
As it is now perhaps the screen shot of the trades can explain a bit more.
Profit was made out of 4 trades.This is as I would expect--- few runners and many that fail or perform poorly.To have 4 outliers in 6 mths is a great result.The question I will answer is DID I GET MAXIMUM RETURN on them and did I dud myself on others---in otherwords is my exit TOO tight!
Thats where the answers will lie in the EXIT.

I will continue to walk through the thought process as I work with the system if that is some help to those trying to define the thought process---well the one I use--- in System developement if that is a help to you and others.
I get the feeling this is an area people are struggling a little with?
 

Attachments

  • PENTEKK 2.gif
    PENTEKK 2.gif
    54.9 KB · Views: 432
Doubling our Trading capital from $10000 to $20000
Has a DRAMATIC EFFECT and POSITIVE on the system


Detailed Report
(PENTEKK NS 70 20)

Simulation Summary
Simulation Date: 7/20/2005
Simulation Time: 10:09:43 PM
Simulation Duration: 0.07 seconds

Trade Summary
Earliest Entry Date in the Trade Database: 1/4/2005
Latest Entry Date in the Trade Database: 7/19/2005
Earliest Exit Date in the Trade Database: 1/12/2005
Latest Exit Date in the Trade Database: 7/20/2005

Start Trade Entry Date: 1/4/2005
Stop Trade Entry Date: 7/19/2005
First Entry Date: 1/4/2005
Last Entry Date: 7/14/2005
First Exit Date: 1/19/2005
Last Exit Date: 7/20/2005

Total Trading duration: 197 days

Profit Summary
Profit Status: PROFITABLE
Starting Capital: $20,000.00Finishing Capital: $34,984.67
Maximum Equity/(Date): $14,984.67 (7/20/2005)
Minimum Equity/(Date): ($974.68) (3/24/2005)
Gross Trade Profit: $18,378.32 (91.89%)
Gross Trade Loss: ($3,393.65) (-16.97%)
Total Net Profit: $14,984.67 (74.92%)
Average Profit per Trade: $554.99
Profit Factor: 5.4155
Profit Index: 81.53%
Total Transaction Cost: $1,620.00
Total Slippage: $0.00
Daily Compound Interest Rate: 0.2842%
Annualized Compound Interest Rate: 181.8035%

Trade Statistics
Trades Processed: 230
Trades Taken: 27
Partial Trades Taken: 0
Trades Rejected: 14
Winning Trades: 15 (55.56%)
Losing Trades: 12 (44.44%)
Breakeven Trades: 0 (0.00%)

Normal Exit Trades: 20 (74.07%)
Delayed Normal Exit Trades: 0 (0.00%)
Open Trades: 7 (25.93%)
Protective Stop Exit Trades: 0 (0.00%)
Time Stop Exit Trades: 0 (0.00%)
Profit Stop Exit Trades: 0 (0.00%)

Largest Winning Trade/(Date): $5,713.61 (4/1/2005)
Largest Losing Trade/(Date): ($474.12) (4/7/2005)
Average Winning Trade: $1,225.22
Average Losing Trade: ($282.80)
Average Win/Average Loss: 4.3324

Trade Duration Statistics
(All Trades)
Maximum Trade Duration: 87 (days)
Minimum Trade Duration: 2 (days)
Average Trade Duration: 23 (days)
(Winning Trades)
Maximum Trade Duration: 87 (days)
Minimum Trade Duration: 6 (days)
Average Trade Duration: 32 (days)
(Losing Trades)
Maximum Trade Duration: 24 (days)
Minimum Trade Duration: 2 (days)
Average Trade Duration: 12 (days)

Consecutive Trade Statistics
Maximum consecutive winning trades: 5
Maximum consecutive losing trades: 5
Average consecutive winning trades: 2.50
Average consecutive losing trades: 2.40

Trade Expectation Statistics
Normalized Expectation per dollar risked: $4.3000
Maximum Reward/Risk ratio: 72.08
Minimum Reward/Risk ratio: -2.14
Average Positive Reward/Risk ratio: 8.61
Average Negative Reward/Risk ratio: -1.18

Relative Drawdown
Maximum Dollar Drawdown/(Date): $1,341.91 (3/24/2005)
Maximum Percentage Drawdown/(Date): 6.6370% (3/24/2005)

Absolute (Peak-to-Valley) Dollar Drawdown
Maximum Dollar Drawdown: $1,341.91 (6.6370%)
Capital Peak/(Date): $20,217.23 (2/17/2005)
Capital Valley/(Date): $18,875.32 (3/24/2005)

Absolute (Peak-to-Valley) Percent Drawdown
Maximum Percentage Drawdown: 6.6370% ($1,341.91)
Capital Peak/(Date): $20,217.23 (2/17/2005)
Capital Valley/(Date): $18,875.32 (3/24/2005)
 
I'll have the first draft ready for release complete with commentry and instructions for those with systems testing capability to pick to pieces.

This is a very important aspect of testing---if it holds up to the masses pulling it to bits then chances are It will hold up while trading.

My tesing hasnt been exhaustive---yet but results are promising enough to throw out to the general populace.

I hope to have it written up and presented by the Weekend.
This W/E is very busy so if its not done by then it will have to wait till the week after.
 
Pretty impressive 55% winning trades and a win/loss of over 4:1 and only a 6.6% drawdown.

MIT
 
MIT.

Yes too good.
I'm trying to find out WHY its so high.
Think Ive found out too--time for bed.
 
Top