scooby-doo BOJ JPY pairs trading robot | Page 345 | 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

  • #6,881
  • Edited 4:17pm Feb 27, 2010 3:43pm | Edited 4:17pm
GU is both ranging and trending/breakout out:

(**To see the chart text, you'll have to double-click on the image to enlarge it**)

Red boxes = failed mean reversion trades (3) (first set's entries/exits on far left missed just by a few pips each time, LOL, ie. whipsaws)

Blue boxes = successful mean reversion trades (2)

Red ellipses = failed SAR breakout trades (on failure of mean reversion trade) (1), ie. here we target the 1.0 ADR line when the mean reversion at 0.5 ADR line trade is stopped out.

Green ellipses = successful SAR breakout trades (on failure of mean reversion trade) (2)

NB: none of these trades were taken (these are looking retrospectively at two randomly chosen major charts/pairs). Flexibility to change parameters is key, as you can see here. (Narrow misses on the left).

Jury is out whether it makes sense to SAR, but if we can fiddle with the parameters, maybe SAR at failed mean reversion trades will work.

http://i46.tinypic.com/2rqoqkn.gif
  • #6,882
  • Edited 4:21pm Feb 27, 2010 4:08pm | Edited 4:21pm
Hi, Invisible,

I have a dumb question as someone with very little experience running bots. NOT been my thing, so far.

How would I get the MPTM trade manager etc. running (with your slightly changed values) on the same chart as a I put your version of Scooby.mq4? (When I placed it on the chart with the Scooby EA, it "erased" the Scobby EA.)

So do I put it on a separate chart? And then if I do that, will it then interfere with the Zoogle EA I am running on the same copy of MT4, or will it retain it's own TP/SL settings etc. independently of the MPTM controlling your Scooby EA version?

Sorry for dumb questions, but these issues are definitely not my forte.

Also, what is your TP/SL in the MPTM for the Scooby EA, or can you give me the average size of your winning trades vs. average size of losers? (Maybe there is no fixed TP/SL in MPTM if you are running a trailing stop etc., I don't know).

Again, thanks for help. Good work!

P.S. 100 SMA/EMA is definitely followed by markets, re http://forex.fxdd.com/category/forex-trading

As is 20 SMA/EMA. 34 or 30 SMA/EMA fine (makes little difference off by 3-4 etc.).

*If it ain't broke, don't fix/change it! But if you do at some point, maybe do per Scoobs 25 (no different really from 20/1; and drop the 34 for the 50, which is watched more; and retain the 100; and maybe include the 200, which definitely is watched just as much as the 100 is). But I'm not sure if we really need the 200 (re don't multiply variables needlessly in a system), as the 100 is so good.

*But I see the logic of at least one Fib ratio between your MAs, 21:34=0.618 Fib, and then using the very intently watched 100 SMA. Lots of big market players watch this one, without having to go all the way up to the very long 200 SMA, though prices/big players will bounce positions off both, so retraces to them are really good entry points, on average, in a trend EA).

Quoting Invisible
Anyway, this version of Scooby.mq4 has made an excellent start. I also have a couple more ideas I want to test, so I will keep this original version running and run the other ideas alongside it. Apart from the low floating DD (just over 3%) and good profitability (11% in 3 days), what I also liked was what I saw in terms of where the bot was placing trades. Eyeballing the charts every so often showed it entering at some really good points, which surprised me really as the methodollogy behind it is relatively simple.

I will keep you all posted...
  • #6,883
  • Feb 27, 2010 4:28pm Feb 27, 2010 4:28pm
I can help with some answers regarding the mtm setup you want.
It is an ea, so it does have to go on another chart. You can indeed run it even if you have other ea's running (like zoogle). Just set it to only control per trade comment, or magic number and it will only look at the trades you want it to.
I run the tick sender app, but if you don't just open a separate gj chart (for more tick activity) and drop mtm on that.
  • #6,884
  • Feb 27, 2010 4:34pm Feb 27, 2010 4:34pm
Great, thanks!

So, if it's set to the same magic number as Invisible's version of the Scooby EA (the one I want to control), it will control that EA then?

What would one put in trade comment if one was doing it that way instead?

Quoting tazmaan
I can help with some answers regarding the mtm setup you want.
It is an ea, so it does have to go on another chart. You can indeed run it even if you have other ea's running (like zoogle). Just set it to only control per trade comment, or magic number and it will only look at the trades you want it to.
I run the tick sender app, but if you don't just open a separate gj chart (for more tick activity) and drop mtm on that.
  • #6,885
  • Feb 27, 2010 4:43pm Feb 27, 2010 4:43pm
@bit99ret

If you run my Scooby mod on the same MT4 as another EA, then you can set the MPTM to manage the trades based on the magic number of the EA. (I think it is "1801"). This means that the MPTM will only manage the trades of the EA that has this magic number and will leave the other EA's trades alone.

You need to have my Scooby mod on one chart (M15 and I use GBPJPY). Then you open another chart (I also use GBPJPY for this) and attach the MPTM to this chart and load up the set file from the ZIP I attached to the previous post.

To see my results for the EA so far, please download the ZIP as it has the EA, set files and also results in there. Like said, I am not sure why it has done so well so far, but it definitely looks promising. I will be running this next week also along with two variants alongside it all on FXDD demos to see how they compare.

Cheers

Invisible
  • #6,886
  • Feb 27, 2010 5:24pm Feb 27, 2010 5:24pm
Quoting bit9ret
Hi, Steve,

On both versions of the bot, getting these "string" error messages:

- "Symbol name for market info function must be a string"
Not even going to go looking at this one. Trust me, I learned a long time ago about this. Every call to this function is a string; trust me, I know. This error arises when you change time frames on a chart with an ea loaded on to it - never a good idea.

Quote
- "Symbol name for order send function must be a string"
- "Order send failure with error 4062: string parameter expected"
Same answer as above. Not even I make mistakes like this any more.

Quote
And on the new version the following also crops up:

- "Order send failed with error 131: invalid trade volume"
That is your mistake. Either you are trying to: send a lot size unacceptable to your criminal or: your lot size when divided by 2 results in a lot size unacceptable to your criminal.

  • #6,887
  • Feb 27, 2010 5:37pm Feb 27, 2010 5:37pm
Quoting bit9ret
Great, thanks!

So, if it's set to the same magic number as Invisible's version of the Scooby EA (the one I want to control), it will control that EA then?

What would one put in trade comment if one was doing it that way instead?
You would set ManageByTradeComment to true, then enter the trade comment of the trade(s) you wanted to manage in the TradeComment input.

For example, I want mptm to manage my Fibonacci trades. They all have "Fib" as their comment and I want to manage by trade comment. My mptm inputs are: ManageByTradeComment = true: TradeComment = "Fib"

Hope this helps.

  • #6,888
  • Feb 27, 2010 6:05pm Feb 27, 2010 6:05pm
bit9ret, I am 'thinking aloud' here. Considering:

"So can we consider a mod 2 version of this bot that will SAR long/short if the ranging day stop is broken at +/-0.7 ADR (0.5 entry point + 0.2 ADR stop)? So we'd exit the attempted mean reversion position at the stop and switch/reverse long net 1 unit instead, ie. do some fancy footwork there?

.............................

It's a narrow 0.2 ADR window we could "tap dance" in, ie., +/- 0.5 ADR is broken; stopped out at +/- 0.2 ADR beyond that when day "decides to" trend/break out; and simultaneously SAR long/short all the way up/down to the +/- 1.0-1.2 ADR trend day level. "

This would be much easier to code if those blasted USA criminals had not adopted their no/restricted hedging anti-trader policies. The fact is, they have and I have to code around it. At least, I do until every one here does the sensible thing and abandons them for non-US criminal accounts.

Hint hint, everyone.

In the meantime, this is what I understand. Using a sell trade as an example, and using the robot defaults for the purpose of this example.

Either:

 

  1. when the market reaches candle open price + 50% ADR, the robot sends a pending sell that fills after the appropriate retrace. Stop is candle open price + 0.7% ADR.
  2. at open price + 0.7% ADR, the trade hits its stop. At this point, the robot kicks in a market buy trade:

    1. tp at candle open price + 100% ADR
    2. sl is? Any suggestions?


Or:

If the market is in a runaway trend and the stage 1 pending trade never fills. then this is the sequence of events:

 

  1. at candle + 0.7% ADR, the robot deletes the pending sell trade and sends an immediate market buy trade:

    1. tp at candle open price + 100% ADR
    2. sl is? Any suggestions?

Have I got the right idea?

Cheers

  • #6,889
  • Feb 27, 2010 8:05pm Feb 27, 2010 8:05pm
[quote=SteveHopwood;3504975]bit9ret, I am 'thinking aloud' here. Considering:

"So can we consider a mod 2 version of this bot that will SAR long/short if the ranging day stop is broken at +/-0.7 ADR (0.5 entry point + 0.2 ADR stop)? So we'd exit the attempted mean reversion position at the stop and switch/reverse long net 1 unit instead, ie. do some fancy footwork there?

.............................
REply :
 Hi Steve, we need 2 more filters to determine ranging or trending day.
Try to put this 2 Indicators: ATR(14) & Standard Deviation(14)
Now here's the rules ...
If you find ATR(14) > then the Standard Deviation(14) it means xxx/yyy = Ranging/Sideways.
If you find ATR(14) < then the Standard Deviation(14) it means xxx/yyy = Trending.


The Logic behinds is; if the SD is smaller then the ATR it means Market in range of the 14 bars because ATR is measure the range (high to low) of the N bars (in this case 14 bars); BUT if SD is more large then the ATR it means Market break the boundaries of High to Low 14 Bars Range & the result Market is Trending.
-----------------------------------------------------------

It's a narrow 0.2 ADR window we could "tap dance" in, ie., +/- 0.5 ADR is broken; stopped out at +/- 0.2 ADR beyond that when day "decides to" trend/break out; and simultaneously SAR long/short all the way up/down to the +/- 1.0-1.2 ADR trend day level. "

reply :
Yes, I like this kind of ADR-stopped style SL.
I'm just considering if "stopped out at +/- 0.2 ADR(3 - 7 Days ADR) is more profitabled instead of 10% SL for Scooby original EA.



hope this can help
Cheers,
  • #6,890
  • Feb 27, 2010 9:44pm Feb 27, 2010 9:44pm
Steve,

I'm getting out of my depth here with the mechanics of bot entry orders, as I'm used to manual order entry/discretionary trades. But you have the logic of the system here clearly.

I can say what follows....

On TP and SL I am coming at in two different ways:

1) actual price behaviour - price looks like it tends to overshoot 1.0 ADR a tad on a running day, and tends to retest the 0.5 ADR level before running to the 1.0 ADR level.

2) the math of what we have to do achieve positive expectancy - as these SAR trades are unlike to succeed more than half the time (50% accuracy), the TP must be > SL. (See the two random charts I posted: 3 attempted SAR breakouts to 1.0 ADR price level failed, 2 succeeded)

Given these two points, I'm thinking:

- TP [+/-] 0.4 ADR (running us to a slight overshoot of the 1.0 ADR price level). We have to extract every inch of profit here we can because the accuracy of this SAR sys is suspect.

- SL at [+/-] 0.25 ADR (putting us just back just back beyond the +/- 0.5 mean reversion price level). Why? Because price seems to retest the mean reversion point at 0.5 ADR. No trailing stop, because there's a lot of noise, backing and filling of prices sometimes.

That gives us a TP/SL ratio of 0.4/0.25 ADRs = 1.6 RR ratio. We must have this if the accuracy of SAR trades is circa 50% or less.

(So TP = 0.7 with price running to 1.1 ADR = 0.4 ADR TP; SL = 0.7 entry point minus 0.25, taking us just back beyond the 0.5 mean reversion point to 0.45 ADR price level).

Note that I am getting cold feet somewhat in looking at the math/likelihood of this level of expectancy, whether this SAR system is viable. But that's that.



Quoting SteveHopwood
Either:

 

  1. when the market reaches candle open price + 50% ADR, the robot sends a pending sell that fills after the appropriate retrace. Stop is candle open price + 0.7% ADR.
  2. at open price + 0.7% ADR, the trade hits its stop. At this point, the robot kicks in a market buy trade:

    1. tp at candle open price + 100% ADR
    2. sl is? Any suggestions?


Or:

If the market is in a runaway trend and the stage 1 pending trade never fills. then this is the sequence of events:
[list=1][*]at candle + 0.7% ADR,...

  • #6,891
  • Feb 27, 2010 9:51pm Feb 27, 2010 9:51pm
On the "AD2008's" ATR vs. standard deviation idea, that is very interesting!

Never seen that before, but am not sure how we could mathematically compare the two in a bot because they are in different units (apples and oranges; and lines crossing on a chart of these are merely coincidental).

One misc, maybe crazy idea: what would a coded line chart indicator look like of various ADR periods plotted on ONE indi pane? On one indicator pane below the price chart, could we have several ADR lines ploted/overlapping: 3-day ADR line, 10-day ADR line, 100-300 day ADR line.

Why? I'm trying to envision this, but could the very short term 3 day ADR line lead, run to the top the chart, and cross back down over the 10 day or 100-300 day ADR line (either). Could that be our entry point for either a short or long mean reversion trade?? (ie. excess ADR movement bleeding off, peaking out, as signal for mean reversion). I looked at this with MT4's standard deviations indi on hourly charts, and it doesn't work, but maybe with ADRs....not sure.

(I'm thinking of your original bot, not the new SAR idea one).

I have no idea what these lines would look like. Maybe worth giving a shot coding a simple indi with these three lines?? (If you can't have three separate lines on one indi pane, just code it once, and we can apply the same indicator 3 times on the same pane, just each having different parameter lengths). No problem with scaling here because we would be comparing apples with apples (ADR to ADR).

I know Larry Connors does these sorts of calculations with historical volatility (HV).

Just a crazy professor idea.... Just ignore if apropos, LOL.
  • #6,892
  • Edited 10:47pm Feb 27, 2010 10:33pm | Edited 10:47pm
Hi, Steve,

I'm such a friggin doorknob, I missed the obvious....

Your exit on a SAR breakout of the +/- 0.7 ADR level should be an hourly CLOSE back inside the hourly 20 Bollinger band used as a trailing stop

Why?

1) it looks reasonably safe and should give lots of scratch or small loser trades (hey, scratches are big winners; anything to prevent repeated sizeable DD trades).

2) and very, very important: on a real runner, big trend day smashing the 1.0 ADR level by a country mile, the hourly outer BB will trail the whole trend to it's denouement for a big RR win

And #2 is exactly what we need: a couple of big outlier big ADR trend days to more than pay for the many scratches/small losers in #1. This will produce a positive expectancy SAR breakout component to any of our failed 0.5 ADR mean reversion trades.

Take a look at a chart of those Shinigami ADR indi lines I posted and a 20 BB trail on the hourly charts.

Example:

red boxes = entry points of failed mean reversion trades, folllowed by...

ellipses = SAR breakout trades...our attempts at self-insuring/gaming this bot's failures....

green ellipses = successful SAR breakout trade (entry and exit)

red ellipse = small loss or breakeven SAR breakout trade

blue boxes = successful mean reversion trades

So what we have: small net winner on far left; loser in middle; two modest winners on right. And a dual mean reversion/breakout system bot.

Note: I'm not biasing the charts here. It's simply the chart I happened to have up. There are no huge runner days here, but the SAR breakouts do succeed or only have a small loss/scratch. And on yen crosses, there can be huge breakout days once in a while. Hourly trail Bollinger Band will grab a good portion.

**Double click on chart to read the text

P.S. I'll be back Monday to check in (nothing tomorrow)....

http://i47.tinypic.com/vrdf8x.gif
  • #6,893
  • Feb 28, 2010 5:19am Feb 28, 2010 5:19am
Quoting bit9ret
I'm such a friggin doorknob, I missed the obvious....
Reminds me of the time a piano pupil turned to me and muttered, "Heck Steve, I am such a spoon."

The good news is that none of this is difficult to code; I shall do it tomorrow.

  • #6,894
  • Feb 28, 2010 5:27am Feb 28, 2010 5:27am
Steve....

"I do until every one here does the sensible thing and abandons them for non-US criminal accounts.

Hint hint, everyone."

I've tried to sign up with a few brokers but I have to prove my residency and soon as I'm from USA they won't let my registration go through. If any member out there living in the US is using a good hedge-able (is that a word) broker please list here or pm me I'm interested. Some one mentioned FXDD allows it which I use I have not tried it cause I thought it would not work here(US)

Thanks
  • #6,895
  • Feb 28, 2010 6:52am Feb 28, 2010 6:52am
Quoting dryclean
Steve....

"I do until every one here does the sensible thing and abandons them for non-US criminal accounts.

Hint hint, everyone."

I've tried to sign up with a few brokers but I have to prove my residency and soon as I'm from USA they won't let my registration go through. If any member out there living in the US is using a good hedge-able (is that a word) broker please list here or pm me I'm interested. Some one mentioned FXDD allows it which I use I have not tried it cause I thought it would not work here(US)

Thanks
Yup... FXDD still allows hedging in all forms.
"The only successful substitute for brains is silence."
  • #6,896
  • Feb 28, 2010 7:19am Feb 28, 2010 7:19am
Hello all,

Amazing amount of information in this thread. Steve, Scoobs and all - thank you for sharing. Steve a special thanks to you for the extra effort to keep this thread focused...

Anyway - worked a lot with EA's over the years and have had very limited success. After reading about Scoob's Holy Grail, thought I would give it a run.

My results...

3 days trading 6 orders 206 PIPS !!! I killed the last two orders manually since I don't like leaving anything open over the weekend.

I might be dangerous when I figure out what I am doing...


T...
  • #6,897
  • Feb 28, 2010 7:31am Feb 28, 2010 7:31am
Quoting dryclean
Steve....

"I do until every one here does the sensible thing and abandons them for non-US criminal accounts.

Hint hint, everyone."

I've tried to sign up with a few brokers but I have to prove my residency and soon as I'm from USA they won't let my registration go through. If any member out there living in the US is using a good hedge-able (is that a word) broker please list here or pm me I'm interested. Some one mentioned FXDD allows it which I use I have not tried it cause I thought it would not work here(US)

Thanks

fxdd -us does allow hedging. But there is all the NFA talk about reducing leverage from 100:1 to 10:1. An option for US citizens is fxdd-Malta. They will accept applications from US citizens and their leverage goes to 200:1 I believe. There are other brokers who willaccept US citizens though. I've heard good things about GO Markets. Goodluck.
Sandy
  • #6,898
  • Feb 28, 2010 9:02am Feb 28, 2010 9:02am
Steve - Does it make sense to move this ADR mumbo-jumbo to its own thread?
  • #6,899
  • Feb 28, 2010 10:20am Feb 28, 2010 10:20am
Quoting dryclean
Steve....

"I do until every one here does the sensible thing and abandons them for non-US criminal accounts.

Hint hint, everyone."

I've tried to sign up with a few brokers but I have to prove my residency and soon as I'm from USA they won't let my registration go through. If any member out there living in the US is using a good hedge-able (is that a word) broker please list here or pm me I'm interested. Some one mentioned FXDD allows it which I use I have not tried it cause I thought it would not work here(US)

Thanks
Good Morning,
I am also a US resident who keeps his forex accounts across the pond because of our pre-school govt. Enclosed is a list of the demo accounts on my system now. All are non-ECN, some are USA types. This list was compiled over years of digging about. I will not promote one over the other....

whip
Attached File(s)
File Type: txt brokers.txt   < 1 KB | 348 downloads
  • #6,900
  • Feb 28, 2010 10:40am Feb 28, 2010 10:40am
Quoting dryclean
Steve....

"I do until every one here does the sensible thing and abandons them for non-US criminal accounts.

Hint hint, everyone."

I've tried to sign up with a few brokers but I have to prove my residency and soon as I'm from USA they won't let my registration go through. If any member out there living in the US is using a good hedge-able (is that a word) broker please list here or pm me I'm interested. Some one mentioned FXDD allows it which I use I have not tried it cause I thought it would not work here(US)

Thanks
Try forex.com with U.K. server.
14 traders viewing now