scooby-doo BOJ JPY pairs trading robot | Page 262 | 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

  • #5,221
  • Feb 2, 2010 5:14pm Feb 2, 2010 5:14pm
Quoting maxou888
Ater 2 months of demo, I went Live for Original 2.2 last night.
All params are defaut, except hours. Broker is MIG.
4 trades (2 last night, 2 this morning) : 3 winners, 1 loss (gbpjpy opened at 23:39 closed at 04.35 with 11 pips loss, in demo it closed with 10 pips profit !!)
Total net profit : +863 €
Dear Sr.

Would you mind to tell which pairs are you using in your live account? I know you say that are all defaults so it would be only gbpjpy and eurjpy?

I'm starting to test in Alpari UK and intended to update results here.

Regards.
  • #5,222
  • Feb 2, 2010 5:48pm Feb 2, 2010 5:48pm
Quoting scooby-doo
Also, taking an average value across the pairs to base a trading decision on is very dangerous the more pairs you include.

If you trade 4 then actually in certain circumstances it only takes 2 of them to be OB or OS to send all the trades.

So, to compensate this I am looking to see how many of the 4 are actually < 20 or > 80 with 3 being the absolute minimum before sending all of them.

Scoobs.
good idea. However, if you do the restriction like that (3 out of 4) then the average level of OB/OS will of course typically be less, it's like taking the limit to 15/85 or so.

What about average level, plus an upper bound on standard deviation?
  • #5,223
  • Feb 2, 2010 6:10pm Feb 2, 2010 6:10pm
Quoting mbkennel
good idea. However, if you do the restriction like that (3 out of 4) then the average level of OB/OS will of course typically be less, it's like taking the limit to 15/85 or so.

What about average level, plus an upper bound on standard deviation?
Because of the cross-pair relationship the key is making sure that USDJPY is actually really OB or OS.

Scoobs.
  • #5,224
  • Edited 7:09pm Feb 2, 2010 6:56pm | Edited 7:09pm
I just got around to reading this email, sent to me recently. Sounds like some of my coding attempts:

Jack Daniels Christmas Cookies
1 cup of water
1 tsp baking soda
1 cup of sugar
1 tsp salt
1 cup or brown sugar
4 large eggs
1 cup nuts
2 cups of dried fruit
1 bottle Jack Daniels Black
Sample the Jack to check quality. Take a large bowl,
Check the Jack again, to be sure it is of the highest quality,
Pour one level cup and drink.

Turn on the electric mixer. Beat one cup of butter
In a large fluffy bowl.

Add one peastoon of sugar. Beat again. At this point
it's best to make sure the Jack is still ok, try another
Cup just in case.

Turn off the mixerer thingy.

Break 2 leggs and add to the bowl and chuck in the cup
Of dried fruit.

Pick the frigging fruit off the floor.

Mix on the turner.

If the fried druit gets stuck in the beaters just pry
It loose with a drewscriver.

Sample the Jack Daniels to check for tonsisticity.

Next, sift two cups of salt, or something. Who geeves
A sheet. Check the Jack Daniels. Now shift the lemon juice and strain your nuts.

Add one table.

Add a spoon of sugar, or somefink. Whatever you can
Find.

Greash the oven.

Turn the cake tin 360 degrees and try not to fall
Over.

Don't forget to beat off the turner.

Finally, throw the bowl through the window, finish the
Jack Daniels and make sure to put the stove in the wishdasher.Cherry Mistmas !

  • #5,225
  • Feb 2, 2010 8:58pm Feb 2, 2010 8:58pm
Hey Steve,

I'll have to try that JD cake recipe. Those piss infused cakes rock ! Let alone the buzz whilst making the feckin' ting.

Back to biz

Optimized for 6 pairs for TCF v2 (based on my previous with backtests)
EURUSD,USDJPY,GBPJPY,EURJPY,GBPUSD,USDCAD

Heres the fxbook link http://www.myfxbook.com/members/getd...gd71-mod/15997

getdown71
  • #5,226
  • Feb 2, 2010 9:03pm Feb 2, 2010 9:03pm
Quoting SteveHopwood
I just got around to reading this email, sent to me recently. Sounds like some of my coding attempts:


Finally, throw the bowl through the window, finish the
Jack Daniels and make sure to put the stove in the wishdasher.[/b][/font][/color][/color][/size][/font][/b]Cherry Mistmas ![/center]
[/center]...
禠孠Victor V
  • #5,227
  • Feb 2, 2010 9:19pm Feb 2, 2010 9:19pm
Quoting getdown71
Hey Steve,

I'll have to try that JD cake recipe. Those piss infused cakes rock ! Let alone the buzz whilst making the feckin' ting.

Back to biz

Optimized for 6 pairs for TCF v2 (based on my previous with backtests)
EURUSD,USDJPY,GBPJPY,EURJPY,GBPUSD,USDCAD

Heres the fxbook link http://www.myfxbook.com/members/getd...gd71-mod/15997

getdown71
I have mixed results running on demo with 10 pairs. So far, promising ones are GU,EU, GJ, EJ and AJ. All with default settings except i turn off hedging.

What's your settings? Maybe i can test at the same time with my broker?
禠孠Victor V
  • #5,228
  • Feb 2, 2010 9:45pm Feb 2, 2010 9:45pm
Quoting scooby-doo
Because of the cross-pair relationship the key is making sure that USDJPY is actually really OB or OS.

Scoobs.
OK, I'll believe you, you want a consensus. Does this mean "USDJPY is OB or OS, plus N others" is the best criteria.

Another question---there's a big difference in using RSI(2) if you consider the 'real-time' RSI vs the "previous time period" RSI. I.e. the first one changes with the moment as the current price equals the close.

Should we consider 'previous candle' or the moment the current candle hits the limit?

This is important for all robots---even Steve's original BOJ method uses a "shift" parameter of zero, meaning taking the current real-time price as "close" for computation.

So this means that say if a single tick puts RSI above 90, and the next tick puts it below 90, the robot will trigger the trade.

I know a real-time manual trader would use the feel to decide if it is worth it or not, but are there any suggestions for rigorously dumb algorithms?
  • #5,229
  • Feb 3, 2010 3:37am Feb 3, 2010 3:37am
Sorry Guy,

I just wanted to re-post. I remember there was a strength indie coded with the inputs for all the yen pairs. where you would just check the pairs you wanted to include in the indie. PC crashed and lost it. anyone still have this?

Thanks
GREED IS THE ULTIMATE INDICATOR
  • #5,230
  • Feb 3, 2010 3:48am Feb 3, 2010 3:48am
Quoting mbkennel
OK, I'll believe you, you want a consensus....
Ideally all of them should be OB or OS but I am not sure how many trade setups you would get in a week.

If a single tick puts it into OB or OS is fine.

Scoobs.
  • #5,231
  • Feb 3, 2010 4:08am Feb 3, 2010 4:08am
Quoting SteveHopwood
I noticed that the robot was closing hedged positions at a loss. This turned out to be a coding error - I was not including the hedge trades in the basket upl calculations. The bot was closing the entire position when the first two trades went into profit, ignoring the hedges.

There is a fix for this in the zip in post 1.

Hi Steve,

Downloaded & extracted the zip from post 1 but the file dates on the opposite bots seem to be unchanged.

Is it possible you missed adding the updated files to the zip?

OK - if I am wrong feel free to pour scorn - quite used to it from 'her indoors' -
  • #5,232
  • Feb 3, 2010 4:09am Feb 3, 2010 4:09am
Quoting mbkennel

So this means that say if a single tick puts RSI above 90, and the next tick puts it below 90, the robot will trigger the trade.

I know a real-time manual trader would use the feel to decide if it is worth it or not, but are there any suggestions for rigorously dumb algorithms?

This problem is common in the control and instrumentation field whether using electrical,electronic,pneumatic or mechanically originated signals etc. and the simple solution has been around for over 100 years embedded in various forms of control equipment & "systems" etc.

It is the use of a "Deadband" or "Hysterisis" filter to filter out spurious or false triggers. (even your mechanical or electronic geyser/immersion heater on/off controls have hysterisis to minimise hunting)

One can search the internet for the full background, graphs etc but for the moment is can be simply summarised as follows

Signal going up must reach Limit 1 (say >=93).... but to trigger OB trade same signal must then fall to limit 2 (say <=87) that is a deadband of 6.

Signal going down must reach Limit 4 (say <=17).... but to trigger OS trade same signal must then rise to limit 3 (say >=23) that is a deadband of 6.

I have used this simple filter with the faster scooby signal rsi settings with great success.

As with all things there is some downside, so the trick is to keep the deadband as narrow as possible to filter out the false signals and not introduce large time delays in entering a trade.

The optimum deadband for a particular Pair/TF/Period can be discovered through test optimisation or observation and measurement.

drxyber
  • #5,233
  • Feb 3, 2010 5:42am Feb 3, 2010 5:42am
Quoting fxcellent
Sorry Guy,

I just wanted to re-post. I remember there was a strength indie coded with the inputs for all the yen pairs. where you would just check the pairs you wanted to include in the indie. PC crashed and lost it. anyone still have this?

Thanks
here ya go
Attached File(s)
File Type: ex4 RS_GROUP_JPY Strength Index with inputs.ex4   7 KB | 368 downloads
  • #5,234
  • Feb 3, 2010 10:01am Feb 3, 2010 10:01am
I have been following this thread with great interest and have also been demoing several of the robots. Thank you to Steve and Scoobs for all of your hard work and incredible information....it's been an eye opener!!

I have been running the DB on a demo account without any issue but when I try to load it onto a live account (very small) I keep getting the message that the robot has been suspended because I need to change the Magic Number input. I have tried several different numbers but to no avail. Am I doing something wrong or does the broker just not want me to load this robot onto a live account? I could understand why they wouldn't because since Jan. 22nd, when I changed the settings to 75/25 OB/OS, the DB has had 40 winners with 4 losers! My TP is 40 and SL is at 80 (just to give it some room) and so far I've only had one basket trade close (TP set at 5%), otherwise all the rest have closed at TP of 40.

Any advice you could offer would be greatly appreciated as I'd love to see if live results are the same as the demo. As I said earlier, it's a very small account so if it gets wiped out I won't be crying in my soup.

Thanks again for all the wonderful information in this thread!!
  • #5,235
  • Feb 3, 2010 10:10am Feb 3, 2010 10:10am
Quoting SteveHopwood
I noticed that the robot was closing hedged positions at a loss. This turned out to be a coding error - I was not including the hedge trades in the basket upl calculations. The bot was closing the entire position when the first two trades went into profit, ignoring the hedges.

There is a fix for this in the zip in post 1.

Hello Steve.
This problem also affects the original basket? Because I see this robot is still at 2.2 verson...
  • #5,236
  • Feb 3, 2010 11:35am Feb 3, 2010 11:35am
Quoting fdxcrr
I have been following this thread with great interest and have also been demoing several of the robots. Thank you to Steve and Scoobs for all of your hard work and incredible information....it's been an eye opener!!

I have been running the DB on a demo account without any issue but when I try to load it onto a live account (very small) I keep getting the message that the robot has been suspended because I need to change the Magic Number input. I have tried several different numbers but to no avail. Am I doing something wrong or does the broker...
This is something I forget to add to the user guides. You have to drag a fresh instance of the robot onto the chart, select the inputs tab and change the magic number immediately.

  • #5,237
  • Feb 3, 2010 11:36am Feb 3, 2010 11:36am
Quoting bobdirlei
Hello Steve.
This problem also affects the original basket? Because I see this robot is still at 2.2 verson...
No. Different hedging requirements.

  • #5,238
  • Feb 3, 2010 11:40am Feb 3, 2010 11:40am
Quoting macman
Hi Steve,

Downloaded & extracted the zip from post 1 but the file dates on the opposite bots seem to be unchanged.

Is it possible you missed adding the updated files to the zip?

OK - if I am wrong feel free to pour scorn - quite used to it from 'her indoors' -
Don't be so silly, you fool. I don't do things like that.

I wish. We all know I do this kind of thing all the time.

Not this time, though. I downloaded the zip myself to check.

  • #5,239
  • Feb 3, 2010 1:10pm Feb 3, 2010 1:10pm
Scoobs & Steve, great work

Here's a statement after just 2 days of running Scoob's EA in post 5153 with default settings. I have been closing some out just 'cos they're in profit, or applying a 15-25 pip TS if I was near the PC, but it definitely warrants further investigation.

I haven't seen anybody else comment on this EA, unless I missed it.

^ Chris ^
Attached Image(s) (click to enlarge)
Click to Enlarge Name: scooby_ea_post_5153.gif Size: 5 KB
Click to Enlarge Name: scoob_post_5153.jpg Size: 87 KB
  • #5,240
  • Edited 2:10pm Feb 3, 2010 1:55pm | Edited 2:10pm
Quoting Fx Optimist
Scoobs & Steve, great work

Here's a statement after just 2 days of running Scoob's EA in post 5153 with default settings. I have been closing some out just 'cos they're in profit, or applying a 15-25 pip TS if I was near the PC, but it definitely warrants further investigation.

I haven't seen anybody else comment on this EA, unless I missed it.

^ Chris ^
Cheers Chris.
I guess that's Scoobs version of the Ducks Nadgers or whatever the given name will be. I did download but not started testing it yet. Will get it set up on my VPS with the others.

Would just like to thank Scoobs again for his continued input. Without his ideas this thread wouldn't even have started.
3 traders viewing now