Developing a mechanical system from scratch

[CanOz]

What about a short term system that primarily looks for super-high volume. I've had some luck with back testing entries that rely almost entirely on volume (in combination with only the most rudamentary price based descrimination, such as up or down days) in mid caps and would be interested in exploring it further.

Excellent idea Doc! So look for volume and then if trend up = long, if trend down = short....what a great simple approach...Anyone ever tested anything like this?

So we need to qualify 'super' with a MA on volume. hmmm....

 

[R0n1n]

I spoke to my mate about that EMA system, he wrote it off the top of his head during lunch hour. asked him to back test it and he says its return is about 17% .

I also think volume should play a role in the system. At the end of day most of the indicators are a derivative of Price, Volume and Time.

 

[happytrader]

I spoke to my mate about that EMA system, he wrote it off the top of his head during lunch hour. asked him to back test it and he says its return is about 17% .

I also think volume should play a role in the system. At the end of day most of the indicators are a derivative of Price, Volume and Time.

Hi ROn1n

This is so funny! You say this guy trades for a living? Hmm!

Cheers
Happytrader

 

[R0n1n]

he must be making fun of me....

what kind of return are we looking for in the system we are working on.

For volume based BUY we could look at a rule like : if a stock has 50% higher than normal volume AND its up 1% on the day.

 

[CanOz]

he must be making fun of me....

what kind of return are we looking for in the system we are working on.

For volume based BUY we could look at a rule like : if a stock has 50% higher than normal volume AND its up 1% on the day.

I need to get home and experiment on AmiBroker, but we can use a moving average to define normal, and then decide what is super high volume.

We can use another ma to ensure we are in a trend when it occurs...this can be an EMA 20 or whatever. You could even buy after a large decline, like a selling climax...take the third consecutive up bar after a high volume trend change....many many options here...lets just keep brainstorming them out....

We're going to need a better MS or AB coder then me though...or else you'll be waiting a few months

Thanks for your input Ron1n.

Cheers,

 

[wayneL]

On Waynes adjustable exits.
Cant use them in a system as the variables have to be fixed.
Fine for discretionary,but how would you pick which variable to use?

Just a quick clarification.

"Adjustable" was meant for development purposes, to give an eyeball idea what might be interesting to backtest. not that the stops should be adjusted after the system is set.

 

[CanOz]

Just a quick clarification.

"Adjustable" was meant for development purposes, to give an eyeball idea what might be interesting to backtest. not that the stops should be adjusted after the system is set.

Good morning Wayne....whats your thoughts on entries?

 

[wayneL]

Good morning Wayne....whats your thoughts on entries?

I reckon for a shorter term system, some sort of pullback, flag or triangle entry is good (but hard to code) as these are lower risk (in terms of stop placement).

Volatility breakouts could be interesting too, but again difficult to code to get the precise setups I'm thinking of.

Not much help I'm afraid

 

[R0n1n]

I spoke to my mate about that EMA system, he wrote it off the top of his head during lunch hour. asked him to back test it and he says its return is about 17% .
.

as per above I had asked him to back test it and he had said 17%, but got a mail from him with the report from backtest. It says 125% profit... Can someone check the nos please..

he may be playing with my mind....

 

[tech/a]

This looks like a test on a singular stock not a portfolio.

Over what period tested?

159 trades with a 2 period average hold smells like a singular test.

Printout looks like tradestation.

Singular tests in isolated periods (He could have selected a single stock in a 12 mth period where it flew),have no statistical significance.
Try 300 stocks over 8 yrs and then 20000 portfolios and you are getting much better.
I also suspect (Most recient peak) he is using ZIG ZAG indicator which cannot be used in systems testing as its Dynamic,in hindsite it appears fixed in practice it isnt fixed until the next Peak or trough is formed at whatever % ZIG ZAG you set. An absolute no no for testing.
Zig Zag is also used for peak and trough indicators.So same applies.

I need the code to run through Tradesim. But suspect its invalid.

 

[motorway]

Is something like this possible ?

I have little experience in what the limitation
of possible code is




1. A low is made.
2. A rally takes place.
3. Market moves down to a higher low but at least a certain % higher than the prior low. ( maybe no more than 60% retrace )
4. within 4 or 5 days of that recent low ( would not want this too long );
When,
Close is > than the previous close.
Close is > than the close of low in step 1.
The range is > than previous bar’s range.
Today's close is > than the mid range of the bar

Go long next bar

for short reverse the triggers

You would also Need to define the trend these triggers will operate on
maybe from the next time frame up .
eg triggers daily
on a trend defined by weekly

It would define a pull back entry with the trend long or short
hopefully avoid whipsaws and false breakouts



motorway

 

[theasxgorilla]

It would make for an interesting study though:
EOD traders versus IntraDay traders. WHo makes more per year annually in terms of R.O.E ?

The responses to this arguement suggest to me that a lot of people don't understand the science of trading (Not your response Nizar, not at all, I just chose it to post off as you posed the question well).

There is an opportunity factor to any trading system, mechanical or systematic (discretionary). Call it trade frequency if you like. The microcosmic nature of markets tends to mean that you can do as some do and zoom in to the 15m, 30m or 1h timeframe and you will see the same trend behaviour and patterns as you do on a daily or weekly timeframe, only with greater frequency. This can increases the opportunity factor or the trade frequency. Couple this with leverage (lots of leverage...think CFD 1% margin on indexes type leverage) and with one good trade a month on a single instrument you can achieve better than what others use an end of week system across a universe of stocks.

Of course if you don't know what you're doing the adage, 'speed kills' will apply.

 

[happytrader]

as per above I had asked him to back test it and he had said 17%, but got a mail from him with the report from backtest. It says 125% profit... Can someone check the nos please..

he may be playing with my mind....

Thats very interesting ROn1n

May I ask whether or not your friend was actually 'invested' in this system.

Cheers
Happytrader

 

[R0n1n]

This looks like a test on a singular stock not a portfolio.

Over what period tested?

159 trades with a 2 period average hold smells like a singular test.

Printout looks like tradestation.

Singular tests in isolated periods (He could have selected a single stock in a 12 mth period where it flew),have no statistical significance.
Try 300 stocks over 8 yrs and then 20000 portfolios and you are getting much better.
I also suspect (Most recient peak) he is using ZIG ZAG indicator which cannot be used in systems testing as its Dynamic,in hindsite it appears fixed in practice it isnt fixed until the next Peak or trough is formed at whatever % ZIG ZAG you set. An absolute no no for testing.
Zig Zag is also used for peak and trough indicators.So same applies.

I need the code to run through Tradesim. But suspect its invalid.

the code is amibroker, and the report was from it as well.

System used = my Exp1 5
Initial Equity = 100000
Max Open Positions = 6
Trailing StopLoss = 2%
Round Lot Size = 5
Min Position Value = 5000
All other default settings

period = 1 year
stocks= my portfolio

Happytrader HE is not using this for his usual trade.

 

[tech/a]

the code is amibroker, and the report was from it as well.

System used = my Exp1 5
Initial Equity = 100000
Max Open Positions = 6
Trailing StopLoss = 2%
Round Lot Size = 5
Min Position Value = 5000
All other default settings

period = 1 year
stocks= my portfolio

Happytrader HE is not using this for his usual trade.

Thanks very helpful.

Enjoy your systems developement people.

 

[CanOz]

as per above I had asked him to back test it and he had said 17%, but got a mail from him with the report from backtest. It says 125% profit... Can someone check the nos please..

he may be playing with my mind....

You should get Amibroker and you could test systems yourself...its definetly Amibroker.

Motorway thanks for the input, so your looking for a trend reversal? Like a failed attempt to put in a double bottom?

Again, i'm not the one to comment on code, Lesm, Mr.Woodo, GP or Kaveman ...to name a few, for Amibroker.

Anyone want to volunteer to test some entries?

I'll go back and run Ron1n's code and see if i get the same report.


Got to find Wayne's posted code too.

Cheers,

 

[motorway]

Motorway thanks for the input, so your looking for a trend reversal? Like a failed attempt to put in a double bottom?

No buying a pullback in a trend

buying the dips that confirm the trend
By price making higher lows and moving up

motorway

 

[CanOz]

as per above I had asked him to back test it and he had said 17%, but got a mail from him with the report from backtest. It says 125% profit... Can someone check the nos please..

he may be playing with my mind....

If you can get the actual code i'll run it too, the report looks like AB....i think it would be good to run it to show you exactly how difficult this is going to be.

Cheers,

 

[CanOz]

the code is amibroker, and the report was from it as well.

System used = my Exp1 5
Initial Equity = 100000
Max Open Positions = 6
Trailing StopLoss = 2%
Round Lot Size = 5
Min Position Value = 5000
All other default settings

period = 1 year
stocks= my portfolio

Happytrader HE is not using this for his usual trade.

The code should look more like this:

Buy = Ref( Close , -1 ) < Ref( Open , -1 )
AND Ref( Close , -1 ) == Ref( Low , -1 )
AND Ref( Close , -1 ) > EMA( Close , 20 )
AND Close > Ref( High , -1 );

Sell = Close < Ref( Low , -1 );

What you had there was the back test settings of the code, not entry/exit signals.

How about we try to do a narrative on Motorways idea?

you need to specifiy any conditions etc..then describe what you want...

Lets go with something simple...i'll code up OOPs or something and tack on that Chandelier exit...maybe i'll get that done tonite...who knows...

Cheers,

 

[happytrader]

the code is amibroker, and the report was from it as well.

System used = my Exp1 5
Initial Equity = 100000
Max Open Positions = 6
Trailing StopLoss = 2%
Round Lot Size = 5
Min Position Value = 5000
All other default settings

period = 1 year
stocks= my portfolio

Happytrader HE is not using this for his usual trade.

ROn1n I've got to wonder why this guy would throw you a system off the top of his head that he doesn't actually trade?

Canaussieuck May I ask whether this thread this is a serious attempt to develop a profitable short term leveraged system?

I don't mind if it is not, as it is all entertainment value, however, I would adjust my input accordingly.

Cheers
Happytrader