scooby-doo BOJ JPY pairs trading robot | Page 299 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Trading correlation pairs by using the other pairs 843 replies

Export All pairs or selected pairs to .csv with script 3 replies

Pairs of Currency Pairs 4 replies

Trending Pairs / Ranging Pairs 0 replies

Robot trading 120 replies

Attachments: scooby-doo BOJ JPY pairs trading robot
Exit Attachments

scooby-doo BOJ JPY pairs trading robot

  • #5,961
  • Feb 14, 2010 7:49am Feb 14, 2010 7:49am
hi Steve,

you have included 'linear regression channel' into TCF2 with good returns. i wonder what you think of
'Third-power Polynomial Regression Channel' as described here: http://codebase.mql4.com/4332,
to detect earlier a changing trend ?
  • #5,962
  • Feb 14, 2010 8:02am Feb 14, 2010 8:02am
I stopped testing that bot a while back for the reason Swingman brought up. I was checking the numbers against MATAF consistently at the time and was noting the very obvious differences. I was so worried about the differences that I must confess I stopped testing the other bots on this thread that have the same code in them. I cannot bring up matters that have to do with code as I cannot articulate something I do not understand.

Others on this thread were and are still reporting great results with them so I thought the code must be right and Mataf may be wrong ,after all, the bots are doing great.

The more recent bot by Scoobie in particular is doing so well and there are other exceptionally good bots here by Steve and Maxou. These are the ones on the radar right now. IMHO, Steve you may want to clear this up with Swingman by PM possibly. Those who are testing and/or using that bot may be the better for it.

Cheers.
  • #5,963
  • Edited 8:48am Feb 14, 2010 8:28am | Edited 8:48am
Quoting SwingMan
I'm sorry friends that I've posted about something I am sure that is wrong.

@mbkennel, you read that @scooby-doo has taken over the formulas, and he do not own programmed or tested this. With your knowledge you can made some practical tests because this is in interest of all program users.
I think this is getting overblown. It's not a matter of "right" or "wrong" (both formulas compute something as they are supposed to) it's a matter of "which is most useful for the purpose at hand".

That purpose is, "checking whether the currencies in question are behaving as if the JPY is the primary source of movement".

My original intuition is that the formula in the current code does this fine---but I can be swayed otherwise by empirical evidence.

I'm thinking about it more and am starting to see some value of doing it the "other" way, but in truth, generalizing the issue.

In essence it's a matter of timescales upon which one does the detrending. The obvious point of comparison would be to lower the timescale (make faster) the moving average---that's what it's really about!

Classical formula: subtract of mean of time window over which one does correlation.
Modified formula: subtract a causal moving average.

More generally the "mean subtraction time" doesn't have to coincide (if you are just interested in a useful indicator) with the "correlation window time".

If you put the MA length of the existing function (assuming SMA) to be equal to the time window over which one gets the correlation, one gets something closer to the classical formula.

So now I believe it's not a matter of right or wrong (they're both the same in the end), it's only a matter of parameter tuning! All the other parameters in the bot have some good thought to them I believe.

This is worth a test. Let's just make an indicator with a variable parameter for timescale and running average/batch average choice and look to see what setting appears to capture true JPY movement the best. I'm withdrawing my previous opinion that the original formula is superior, thinking about it some more (in sober daylight :-) ) I can see benefits of the other one so I'm now 50:50, but I do still firmly believe that the question is experimental usefulness (better vs worse) as opposed to "right" and "wrong".
  • #5,964
  • Feb 14, 2010 9:03am Feb 14, 2010 9:03am
Would not the logical thing to do be to compare the results of the differences of the codes in question side by side so to speak??

I am no coder or mathematician and what I suggest might be some huge undertaking, I do not know, but I would think inquiring minds would want to see the comparison.

The work in this thread has been nothing short of astounding, I see no reason that should change now.
  • #5,965
  • Edited 9:14am Feb 14, 2010 9:09am | Edited 9:14am
Quoting scooby-doo
Who is superman2 sending me buy/sell signals.

what a fcukin muppet.
why Scoobs TCF V2 shows always long trade not allowed, short trade not allowed??????
even with all set ups given of Scooby....just dont understand....
testing great Scooby BOJ basket robot original V2.2....just GREAT till now!!!


I need help please....not quite new in trading , but new in using Your excelent EA....
thanks very much to all
argon
  • #5,966
  • Feb 14, 2010 9:17am Feb 14, 2010 9:17am
thanks a lot guys...
argon
  • #5,967
  • Feb 14, 2010 9:56am Feb 14, 2010 9:56am
Quoting argon
thanks a lot guys...
argon
already talken in previous posts ( seem to be only a comments problem)
  • #5,968
  • Feb 14, 2010 11:35am Feb 14, 2010 11:35am
Quoting argon
can somebody help me with instructions for all Scooby's EA thanks a lot guys...
argon
Do your own reading, like everyone else has done. Think of it as a 'right of passage' if you like. Try starting at the beginning, that's usually a good place to start (i.e. the First post in this thread).
  • #5,969
  • Feb 14, 2010 11:52am Feb 14, 2010 11:52am
Quoting fxtrader08
Dear Swingman,

I am sorry that some members appear to be treating you with a little hostility.

It is concerning that you believe this to be wrong as I and many others on FF respect your opinion greatly.

Would it be possible for a change to be made in the indicators and to run both side by side? Perhaps this could be the reason why DD is so high at times?

Warm Regards,

fxtrader08
I respect Swingman's opinion too. There are some great minds on here.
Really I respect the vast majority of people's opinions here, because everyone has something useful to contribute in their own way. For example most people on here are much, much cleverer than I am, (I have to read some stuff 2 or 3 times for it to sink in, and then still some things go over my head) but that doesn't mean that my opinion carries no value.
It's not about who's the cleverest, and we shouldn't take it personally if someone disagrees with us. In the end, it's about pulling together to achieve the common aim and we shouldn't lose sight of that.
Swingman (and Watube) has brought something to everyones attention. Doing the calculations a different way might make for even better results, or worse, or results might be exactly the same. But surely it should at least be looked at?
  • #5,970
  • Feb 14, 2010 12:30pm Feb 14, 2010 12:30pm
Quoting dinofx
hi Steve,

you have included 'linear regression channel' into TCF2 with good returns. i wonder what you think of
'Third-power Polynomial Regression Channel' as described here: http://codebase.mql4.com/4332,
to detect earlier a changing trend ?
Hehe. Even the name is offputting.

The LR calculations I use give a slope value, so I am sticking with this for now as a vaguely understand what I am doing with it.

Thanks anyway. I will take a look.

  • #5,971
  • Feb 14, 2010 12:34pm Feb 14, 2010 12:34pm
Quoting watube
IMHO, Steve you may want to clear this up with Swingman by PM possibly. Those who are testing and/or using that bot may be the better for it.

Cheers.
I do not understand a word George or Matt have uttered about this. I have the same understanding as does an ant have the concept of tying shoelaces. I am not getting into this.

I do not even understand the correlation code that Scoobs provided.

As I said earlier, it does not appear to be bust, so I aint fixing it.

  • #5,972
  • Feb 14, 2010 12:42pm Feb 14, 2010 12:42pm
Quoting SteveHopwood
Hehe. Even the name is offputting.

The LR calculations I use give a slope value, so I am sticking with this for now as a vaguely understand what I am doing with it.

Thanks anyway. I will take a look.

thanks, ok for slope, quite different. I drope one on chart to follow yours. May change coz of timeframes or ATR (which i use with 200 on m15 to cut off bad/small signals).
  • #5,973
  • Feb 14, 2010 12:46pm Feb 14, 2010 12:46pm
Quoting SteveHopwood
Hehe. Even the name is offputting.

The LR calculations I use give a slope value, so I am sticking with this for now as a vaguely understand what I am doing with it.

Thanks anyway. I will take a look.

Steve,

'Third-power Polynomial Regression Channel' is more usually know as a COG indicator - Centre of Gravity. If you look at the 'm' setting for the COG you will see if set to 2 you get one turning point in a straight line linear regression channel. Set m=1 and you get straight line. Normally m=4 is used which gives a better fit to PA. Try setting m=6 to 8 if you want the tightest fit.

More info here
http://www.forexfactory.com/showpost...8&postcount=18

Not sure if this info is any use ....

Cheers
  • #5,974
  • Feb 14, 2010 12:51pm Feb 14, 2010 12:51pm
Quoting macman
Steve,

'[b][font=Arial]Third-power...
Cheers macman. It is usually near-impossible to get any useful info out of custom indicators, so I am not getting into this one either.

  • #5,975
  • Feb 14, 2010 1:27pm Feb 14, 2010 1:27pm
Quoting SteveHopwood
Cheers macman. It is usually near-impossible to get any useful info out of custom indicators, so I am not getting into this one either.

Steve,

I had a play around a while ago with a COG indy (not sure if its the same one) and it repainited like crazy.

Cheers,
Nick.
  • #5,976
  • Feb 14, 2010 1:58pm Feb 14, 2010 1:58pm
One thing you need to start thinking about is what todo if you placed a trade at 11.00am and 2 hours later your trade has not hit TP, SL or other exit criteria and there is high impact news approaching.

This is another reason why coding a fully automated EA is very difficult.

Scoobs.
  • #5,977
  • Feb 14, 2010 2:09pm Feb 14, 2010 2:09pm
Quoting scooby-doo
One thing you need to start thinking about is what todo if you placed a trade at 11.00am and 2 hours later your trade has not hit TP, SL or other exit criteria and there is high impact news approaching.

This is another reason why coding a fully automated EA is very difficult.

Scoobs.

I cutoff trades 2 hours before and 2 hours after due to possible trend change. If I get in at the 2 hour mark before and hit breakeven within the 2 hours till news time I take it... otherwise I gotta take the chance it will hit B/E sometime after that.

Worked pretty well but again if you are trading with the trend in 3 TF's it probably won't be so bad there may be a high percentage chance of hitting your B/E before it may hit an S/L.
  • #5,978
  • Feb 14, 2010 2:47pm Feb 14, 2010 2:47pm
can you tell me plz what do you mean by "right of passage "? you mean I have to have my own reading ?....ok will try ..thanx
  • #5,979
  • Feb 14, 2010 4:01pm Feb 14, 2010 4:01pm
Quoting scooby-doo
One thing you need to start thinking about is what todo if you placed a trade at 11.00am and 2 hours later your trade has not hit TP, SL or other exit criteria and there is high impact news approaching.

This is another reason why coding a fully automated EA is very difficult.

Scoobs.
Coding the exit under these circumstance is not a problem. Deciding what we want to do is what requires decision taking.

One thing I do think is essential is that users of automated trading systems use small lot sizes and take lots of little trades. This way, we can ride out the occasional news spike. I do not see us retail traders with our small accounts accepting large lot size trades, depending for their success on taking a small number of pips via an autotrader.

Most of us are hoping to use an auto-trading robot because we are not highly-trained, highly-experienced professional traders. We are happy to accept lots of small profits leading eventually to greater ones.

So, what should we do about high-impact news releases? Ride them out is the answer. Most of them are over-hyped anyway. Even NFP doesn't have much impact these days. If they endanger our accounts, then we are over-trading. If not, then spikes end. Dem prices dey go up, den dem prices dey go down, den dem prices dey go up agin..........................

The mistake people make is to misunderstand the nature of an auto-trading robot. They trade the robot without understanding what it is capable of doing; they tend to see the pros without understanding the cons, yet this has little or nothing to do with news releases.

I repeat: nobody should use a trading robot until they have traded the system the robot trades manually and fully understand what they are doing. This falls on deaf ears.

Actually, I can take this one step further. They should not even trade the robot until they can code it themselves. I do not usually bother voicing this sentiment because I know what a waste of effort it would be to do so. Most people want robots to do their thinking for them. Understandable, if idiotic.

So the answer to the question, "what todo if you placed a trade at 11.00am and 2 hours later your trade has not hit TP, SL or other exit criteria and there is high impact news approaching." is not a problem to code into a robot. Once traders know what they want to do and have learned to do so, they can code it into the bot. Until then, they rely on the likes of you and me to tell them what to do. Ride it out, is my advice.

  • #5,980
  • Feb 14, 2010 4:26pm Feb 14, 2010 4:26pm
Quoting scooby-doo
One thing you need to start thinking about is what to do if you placed a trade at 11.00am and 2 hours later your trade has not hit TP, SL or other exit criteria and there is high impact news approaching.

This is another reason why coding a fully automated EA is very difficult.

Scoobs.
OK, your scenario assumes rightly that a high impact news release has the potential to strongly move price one way or the other (stating the obvious here).

So I guess hedge with pending orders either side of your current position, then re-evaluate after the news event has passed or close the position a few minutes before the news and accept the loss or profit.

Just my $0.01 worth.
3 traders viewing now