Here's another one

[FetteredChinos]

thanks for that robber..

i did actually code some stops and TP's into my version of the system (as published, but with a few extra tweaks), and over 4 years, the results are a little over 10,000 pips with a max drawdown of 700 pips.. every year was profitable, unlike SIBKISS mark 1. averaging getting on for 60pips per week.

i would post the sheet on here, but 4 years of hourly data with formulae amounts to a 15MB worksheet.

im gonna see if it can be improved, and also transferred onto the other crosses.

am quite intrigued by the USD/JPY actually. nothing i tested before has worked on that beast.


regards,

Mr Brightside

 

[robber]

FC, I would like to back test this properly in Metatrader with various stops and TPs. Would you have any objection to me posting the basis of the system over on moneytec to find someone who can code it for me?


thanks


Neil

 

[FetteredChinos]

course not robber.. be my guest..

the criteria i found best are...

over 4 years of data, from 2001 to jan 05

a stop level of between 150 and 200 pips..

a target level of approx 300 pips.. this acts as a reverse too. when the TP is hit, reverse the original trade. (this reverse looks to add about 3000 pips to the profit.. taking us past 13,000 pips over 4 years. thankyou very much, bishop.

plus, instead of closing on friday close, try monday.. for some reason it gives a smoother equity curve, probably because it gives the trade more room to "breathe" and resume its mini-trend.

id love to be proved wrong..

first test trade occurring at 2pm today. lets see what happens...

FC

 

[robber]

FC, heres a copy of what I have requested for metatrader.

Note the price at 08:00 GMT. Look again at 14:00 GMT. (If possible these
times should be user defined variables or adjustable within the coding of
the expert)

If price is higher at 14:00 open a long position, If its lower open a short.

User defined 'target' is required (currently about 300 pips)

User defined 'stop loss' is required (currently 150 to 200 pips)

Close on Friday at 19:00 GMT ( It may be possible to run this trade until
the following week for more profit. To test this the variable 'closeday'
should be any value up to 13.)

There is one further complication to this system; Once the target is reached
a reverse trade is also placed. This is closed either by hitting 'target' ,
'stop loss' or time limit. If you manage to build this into the expert it
would also be very useful to be able to switch it off or to have two
experts; one with this function, and one without.


All other variables to be as per existing SIBKIS expert, eg lots,
slippage, risk, mm, spread and so on.



Will let you know if I get anywhere with it.


Neil

 

[FetteredChinos]

cheers mate. that pretty much sums it up..

from my excel testing, when the stoploss is hit, reversing isnt actually that profitable, so the risk probably isnt worth taking on board just for a few extra pips..

not the same when the target is hit though.

as for close days, as far as i can make out, mondays seem better, but i didnt test for other days in the "2nd" week.. with stops and targets i can only feel that the longer hold period may be more lucrative..

am currently looking at intra-day stuff again.. when will i ever learn......


FC

 

[trendie]

hi FC,

a question re: the breakout start you are journailing.
GBP/USD has been very narrow-range past two days. ( I dont feel so bad about y'days trades, in that context ).

If it is narrow-ranging, and theoretically, about to burst to upside or downside, would you consider resetting your S/R in light of this ? ( you are currently risking 200 pips to get 300 pips. )

if the GBP/USD shot up, would you use a narrower S/R for the upside ?

 

[FetteredChinos]

no fella. a strategy/system is just that. im journaling myself (oo-err) in order to force myself to stick to the rules..

if the 200pip stop is hit, then i dont reverse. it is only the target level that reverses..

no point in backtesting if you dont stick to the rules.

i agree though, the market is gonna go kaboom tomorrow with the NFPs..


you know we're NFP, yeah, you know me...

 

[Roberto]

a strategy/system is just that.

Can't argue with that.

no point in backtesting if you dont stick to the rules.

Can't argue with that either.

the market is gonna go kaboom tomorrow with the NFPs.

Can't even argue with that. Come on, FC, when are you going to say something that I can really ARGUE with?

 

[FetteredChinos]

ok,

is Bella Emberg fit?

 

[trendie]

FC, your test-data is from 2001 to Jan 05 ?

Are you using the same time-frame to determine the best strategy ?
What I mean is, should you be using the same data to identify the strategy AND test over it ?

Shouldnt the startegy be identified over, say 2001 to end-2003.
Then, the strategy resulting from that be tested over the year 2004 ?
( I cant remember the exact terms for this )
edit: I think its called -forward testing.


Otherwise, if you use the same data for identifying the strategy AND the test-data, you are effectively curve-fitting.

Hope the above makes sense.

 

[FetteredChinos]

hmmm true, and i see where you are coming from..

BUT, the main criteria i use when testing is the smoothness of the equity curve..

if it is pretty much straight line, then i can be fairly sure that it isnt curve fitting..as you cant curve fit over such a long period.. you can over 3 months or so. but if performance is fairly steady over 4 years of intraday data, then i am fairly confident we are onto a winner. i will post a jpg of the equity curve, if i can find one lying around the place.

FC

 

[barjon]

..

no point in backtesting if you dont stick to the rules.

..

fc

when you backtest one of your ideas do you backtest for a period (say years 1 and 2) to establish a benchmark (and maybe refine the rules) then try some pseudo forward testing by testing the result through a subsequent period (say years 3 and 4)?

jon

 

[barjon]

whoops, just seen trendie's said something similar - went away for coffee half way through writing mine

 

[FetteredChinos]

et voila..


as you can see, no dodgy spikes in equity in either direction.. just a gradual accumulation...

 

[FetteredChinos]

fc

when you backtest one of your ideas do you backtest for a period (say years 1 and 2) to establish a benchmark (and maybe refine the rules) then try some pseudo forward testing by testing the result through a subsequent period (say years 3 and 4)?

jon

yes barjon, that is what i did in this case..

i started with the original criteria of 8am-12am and tested for 03 and 04..


then extended same criteria to 02. still profitable..

didnt work so well for 01..

so i modified the criteria trying to target highest gross pips..

the only variable was the 2nd time period and the profit target. keeping things simple is hardly curve fitting...

 

[FetteredChinos]

and i should add, going for the highest gross pips i could find also yielded the smoothest curve..


hmm..

 

[barjon]

. keeping things simple is hardly curve fitting...

Never said it was did I? I know you're a simple soul with a prediliction for the curvaceous

 

[G-Man]

Can't beat a good set of curves I say.

 

[FetteredChinos]

sheesh... current short trade was 100pips underwater without a snorkel at one point this morning..


recovering nicely, and now into profit..whether it holds through the fireworks of tomorrow is another matter.

 

[trendie]

sheesh... current short trade was 100pips underwater without a snorkel at one point this morning..


recovering nicely, and now into profit..whether it holds through the fireworks of tomorrow is another matter.

Tell me about it!!
I shorted it at 1.8854.
Messed up trying to set a OCO and ended up closing the trade for a poxy 4 pips profit.
Missed out on over 50 pips.

Double-sheesh.