scooby-doo BOJ JPY pairs trading robot | Page 402 | 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

  • #8,021
  • Apr 5, 2010 7:59am Apr 5, 2010 7:59am
Hi all,

Have not had a chance to post over the past 2 weeks. I am still getting excellent results but did have a two day period that made me step back and think a bit.

As of markets closing for Easter weekend, my demo account was up 595 pips. The period that caused all of my loses were from orders opening 3-24-10 5:00 to 3-26-10 8:00. During this period, the orders that opened resulted in 10 losers and 5 winners for a net loss of 600 pips.

I believe this is the period Scoobs commented about. Attached you will find the trading log.

Any ideas anyone?

T...
Attached File(s)
File Type: xls The Holy Grail DetailedStatement thru 0404.xls   39 KB | 444 downloads
  • #8,022
  • Apr 5, 2010 10:54am Apr 5, 2010 10:54am
pity it is not in LIVE account, otherwise, i'd say it is really holy grail.

as to the period, nothing strange, this Holy Grail definitely lose when market is doing strong trendy run? NO! i'm kidding.

This holy grail will definitely lose when trend (weekly/daily) reverse. NOTE: it is trend change from bull to bear or bear to bull. As you see, that's what happen for that period, the overall market trend from 2010/01/01 changed from bear to bull.

Such overall trend changes won't happen more than 5 or 6 times per year, so other than these 5 or 6 times, the holy grail will earn money in DEMO, based on your testing.

Now we need to watch it on LIVE.

Quoting tsquad
Hi all,
As of markets closing for Easter weekend, my demo account was up 595 pips. The period that caused all of my loses were from orders opening 3-24-10 5:00 to 3-26-10 8:00. During this period, the orders that opened resulted in 10 losers and 5 winners for a net loss of 600 pips.

T...
  • #8,023
  • Apr 5, 2010 11:07am Apr 5, 2010 11:07am
Quoting charvo
pity it is not in LIVE account, otherwise, i'd say it is really holy grail.

as to the period, nothing strange, this Holy Grail definitely lose when market is doing strong trendy run? NO! i'm kidding.

This holy grail will definitely lose when trend (weekly/daily) reverse. NOTE: it is trend change from bull to bear or bear to bull. As you see, that's what happen for that period, the overall market trend from 2010/01/01 changed from bear to bull.

Such overall trend changes won't happen more than 5 or 6 times per year, so other than these 5 or...
I have it running live, starting its 3rd week, with +200 pips.
  • #8,024
  • Apr 5, 2010 11:26am Apr 5, 2010 11:26am
Yeah, i have been waiting for you man. I have 2 trades last week, 1 trade this morning, all winners.

we are not looking for pips, we are looking at a system with stable, evenly-distributed profit, and acceptable drawdown. such a system with only 10 pips profit per month, will be holy grail.

my pc breaks down, cannot show too much, here is a chart you can draw.

using Tsquad's excel, draw a scatter plot for its rightmost column, it looks like that way i wanted. (sorry, cannot post an image now due to my pc).



Quoting Middleastern
I have it running live, starting its 3rd week, with +200 pips.
  • #8,025
  • Apr 5, 2010 12:39pm Apr 5, 2010 12:39pm
A statistical analysis of the returns:

the p-value for a one-way T-test against a mean of zero gives p=0.26 about, so using this test, doesn't yet give statistical significance for profitability at the conventional 0.05 level.

A randomized bootstrap test shows that about 14% of resamples have a mean return below zero.

All in all it's more probable than not that its profitable but it's not clearly certain.

Also the statistical methods assume independent inputs, but of course there is statistical serial correlation resulting in lower confidence still.

Anyway, this is actually a decent result since most methods are losers confirmed quickly.

This is an example of how hard it is to really show quantitatively convincing returns (and why we should be in awe of Renaissance).
  • #8,026
  • Apr 5, 2010 3:06pm Apr 5, 2010 3:06pm
Quote
(and why we should be in awe of Renaissance).

What is Renaissance in this context?
  • #8,027
  • Apr 5, 2010 4:05pm Apr 5, 2010 4:05pm
Hi guys,

I attached the latest version of TBL onto two separate broker demos this morning. So far no trades.. Anyone else experiencing this or am I doing something wrong?

I did remember to turn it on by setting "TradingSuspended" to false. Everything else is default. No errors in the log either...

I actually haven't had any trades since this robot was transformed from Trend Retracement to The Big Lewinsky. Just thought I'd let you guys know in case theres any errors.

Jason
  • #8,028
  • Apr 5, 2010 4:05pm Apr 5, 2010 4:05pm
Quoting Invisible
What is Renaissance in this context?
James Simons' hedge fund company (referring to Medallion in particular).

It has been generating extreme results for two decades on massive size, and is nearly 100% quant.

You know how most hedge funds are 2 and 20? (fee of two percent of assets, 20% of profits). Medallion was something like 5 and 50 (5% of assets, 50% of profits), but now is nearly 100% employee owned.

Unquestionably the best fund in history (better than Soros/Rogers' original)
  • #8,029
  • Apr 5, 2010 4:28pm Apr 5, 2010 4:28pm
Quoting doc816
I actually haven't had any trades since this robot was transformed from Trend Retracement to The Big Lewinsky. Just thought I'd let you guys know in case theres any errors.

Jason
There will not be a lot. This is not a trading trollop. A lot of initial trades happened because my coding was faulty. I hope you have read about this. If not, then go back a few pages and start again.

Hehe. I am the coder, yet I still got stuff wrong. I forgot to turn trading on and didn't look at the feedback properly, so my robot went the entire day without trading.

Once I realised what I had done and corrected it, the bot sent a pending trade immediately - live IBFX. UJ buy that filled at 94.27 - in profit by 4 pips as I write.

Other trades that were taken out erroneously last week, but in the direction of the trend have closed either after a half-close retraced to stop or by hitting tp.

Great start to the trading week despite the holidays.

  • #8,030
  • Apr 5, 2010 4:47pm Apr 5, 2010 4:47pm
Quoting SteveHopwood
There will not be a lot. This is not a trading trollop. A lot of initial trades happened because my coding was faulty. I hope you have read about this. If not, then go back a few pages and start again.
Thanks for confirming Steve. I did read about the coding being faulty, which to my surprise was performing quite well! I almost wish you hadn't found the bug.. lol. But obviously no one here is looking for lucky trades, so I appreciate your continued hard work on making it better.

Jason
  • #8,031
  • Apr 5, 2010 5:14pm Apr 5, 2010 5:14pm
Quoting doc816
I almost wish you hadn't found the bug.. lol. But obviously no one here is looking for lucky trades,
I confess to similar feelings myself. "If it ain't bust, then don't fix it" is a principle I adopt in coding.

This time, I reasoned thus:

 

  1. given that my retrace-spotting did not actually exist, then any trade the system took was relying on there not being a retrace at the moment the trade was taken (for sure, this explains why post-loss trades were so often followed by successful ones - the retrace took out the original trades and set up the post-loss ones).
  2. despite this, the original trades were successful far more often than not.
  3. therefore, measuring any pre-trade retrace would be an advantage.

Ignore me when I post like this, guys. I am thinking aloud; this helps me a lot.

  • #8,032
  • Apr 5, 2010 7:00pm Apr 5, 2010 7:00pm
TBL seems to be working (or operating I should say, at least for now)

Just saw a pending Sell Stop for USDSGD, but it quickly disappeared. I noticed it ended up expiring.

Back to monitoring the demos...
  • #8,033
  • Apr 6, 2010 1:49am Apr 6, 2010 1:49am
modifications:

*) TP/SL configurable, default set back to original scooby settings (not later scooby settings)

*) multiple ways of trend determination

*) (big one), uses LaguerreRSI (internally) to enter trades instead of M05 RSI. This requires a two-stage process of triggering an order with RSI, and then firing it later.

As usual, requires LibGMT.mqh and LibOrderReliable.mqh in 'include' folder.
Attached File(s)
File Type: mq4 TheHolyMackerel_V1.7.mq4   19 KB | 645 downloads
  • #8,034
  • Apr 6, 2010 4:28am Apr 6, 2010 4:28am
Quoting mbkennel
modifications:

*) TP/SL configurable, default set back to original scooby settings (not later scooby settings)

*) multiple ways of trend determination

*) (big one), uses LaguerreRSI (internally) to enter trades instead of M05 RSI. This requires a two-stage process of triggering an order with RSI, and then firing it later.

As usual, requires LibGMT.mqh and LibOrderReliable.mqh in 'include' folder.
As usual, thank you for this. Though I have both Lib in the include folder, I get an error message for the LibOrderReliable, cannot open the program file when I try to compile the V1.7
  • #8,035
  • Apr 6, 2010 5:38am Apr 6, 2010 5:38am
Currently testing 5minmomo (1hr chart), 10 a day, Asian Breakout and TBL.

Had a good week last week with ASian Breakout and 10 a day (despite the errors in coding). I loaded TBL last night and managed to take a couple of trades. Currently getting the message 'Trying to manage trades, but trading not allowed by the criminal'. I have looked through the forum but cannot see reference to this. May be a case of not seeing what is in front of me but then thats hangovers for you. Apologies if this is an obvious mistake. I can't tell when the message first appeared as it happened during the night
  • #8,036
  • Apr 6, 2010 7:36am Apr 6, 2010 7:36am
is anyone else having a bit of a nightmare with TBL (on 5m EURUSD) and the modified basket 508b (running on 5m USDJPY)?

Last week the were ok, but this week has been pretty dire.

Any ideas? ANy luck with any of the other EAs?
  • #8,037
  • Apr 6, 2010 8:01am Apr 6, 2010 8:01am
Quoting Invisible
What is Renaissance in this context?
Renaissance Technologies is Jim Simons' hedge fund, running the uber-successful quant portfolio the Medallion Fund:

http://www.turtletrader.com/trader-simons.html

The fund has averaged 45% per year since inception in 1988.
  • #8,038
  • Apr 6, 2010 8:11am Apr 6, 2010 8:11am
Hi, Just for your information, that 'orderclosereliable' within the lib seems trying to close ANY orders (it should only close orders of a specified magicnumber).

of course, only an observation from my end.

Quoting mbkennel
As usual, requires LibGMT.mqh and LibOrderReliable.mqh in 'include' folder.
  • #8,039
  • Edited 9:03am Apr 6, 2010 8:19am | Edited 9:03am
I have upgraded the History Analysis EA. You can see from the attached pic how much easier it is to read.

The inputs should be pretty easy to grasp; play with the display settings to get the ones you like best.

Attached Image (click to enlarge)
Click to Enlarge Name: chart illustration.gif Size: 9 KB
Attached File(s)
File Type: mq4 Magic Number History Analysis.mq4   11 KB | 291 downloads
  • #8,040
  • Apr 6, 2010 9:06am Apr 6, 2010 9:06am
Quoting SteveHopwood
I have upgraded the History Analysis EA. You can see from the attached pic how much easier it is to read.

The inputs should be pretty easy to grasp; play with the display settings to get the ones you like best.

This is amazing. It'll save so much time trying to figure out performance of each EA manually.

As always, Thanks for your innovations and hard work.
3 traders viewing now