Ramblings of a Data Scientist! | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Ramblings of a Forex Junkie 4,030 replies

Trade Anatomy - ramblings of an old-timer 147 replies

Where to find international lending data, money supply data 0 replies

Mt4 Data Missing - Where to get good data? 0 replies

Attachments: Ramblings of a Data Scientist!
Exit Attachments

Ramblings of a Data Scientist!

  • #21
  • Nov 24, 2017 7:09am Nov 24, 2017 7:09am
oh BTW I trade GBP against AUD,CAD, JPY and NZD. Also I trade GOLD
  • #22
  • Nov 24, 2017 7:24am Nov 24, 2017 7:24am
I have tested this on EURUSD 30/H1 and seems to work a treat, although I am a little confused as to where the message or note telling us to go long !!!

Do you have a similar file I can use for the pairs that I trade?
  • #23
  • Edited 7:48am Nov 24, 2017 7:35am | Edited 7:48am
Quoting Rafaz1
so I have copied your file and it generates a table 1 to 22 and click ok, is something supposed to appear or happen?
That is all you need to do. When I post the line of code overwrite line 22 in the NewGV script then load it on a chart and recompile the EA. Do you follow?

Quoting Rafaz1
oh BTW I trade GBP against AUD,CAD, JPY and NZD. Also I trade GOLD
Subject to request (I think) the next model set will be constructed for Gold. Not sure by when it will be done.

Quoting Rafaz1
I have tested this on EURUSD 30/H1 and seems to work a treat, although I am a little confused as to where the message or note telling us to go long !!! Do you have a similar file I can use for the pairs that I trade?

It won't trade all the time, just when there is an opportunity. Post these numbers into NewGV:

Inserted Code
double D1=1.1857; double D2=1.1823; double D3=1.1753; double D4=1.1745; double D5=1.174; double D6=1.1794; double D7=1.1772; double D8=1.178; double D9=1.1857; double D10=1.1892; double D11=1.1823; double D12=149.2173; double D13=148.8306; double D14=151.4697; double D15=151.1034; double D16=149.4192; double D17=149.3112; double D18=149.2864; double D19=149.781; double D20=149.2173; double D21=149.3352; double D22=149.0994;

Quoting Rafaz1
I have tested this on EURUSD 30/H1 and seems to work a treat, although I am a little confused as to where the message or note telling us to go long !!! Do you have a similar file I can use for the pairs that I trade?
Eventually
  • #24
  • Nov 24, 2017 7:50am Nov 24, 2017 7:50am
Thank you for the prompt responses to my questions (apologies to the naivety of some of those questions).

Could I kindly request files for the pairs that I trade, plus Gold. Apologies in advance as I am not aware of the work this entails by you behind the scenes to construct the relevant files.

May I just ask if you too are UK based?
  • #25
  • Edited 8:05am Nov 24, 2017 7:54am | Edited 8:05am
Quoting Rafaz1
Thank you for the prompt responses to my questions (apologies to the naivety of some of those questions). Could I kindly request files for the pairs that I trade, plus Gold. Apologies in advance as I am not aware of the work this entails by you behind the scenes to construct the relevant files. May I just ask if you too are UK based?

Yes I'm in the UK. There is a poll above the opening page please submit your vote for new model development.


1 HOUR EURUSD MODEL: ENTER LONG...MAY HAVE SOME MORE UPSIDE TO THIS MOVE.

I'm done for the day. Have a great weekend. Don't be a freeloader: don't forget to subscribe and like!
  • #26
  • Nov 24, 2017 8:17am Nov 24, 2017 8:17am
Have a good weekend too

Will make a request for a new model and keep an eye on EURUSD
  • #27
  • Nov 24, 2017 8:30am Nov 24, 2017 8:30am
have a great weekend and let me know when you have some new pitchforks on the menu
MAKE IT RAIN
  • #28
  • Nov 24, 2017 8:48am Nov 24, 2017 8:48am
Quoting mihai1986
have a great weekend and let me know when you have some new pitchforks on the menu

Yes mate...best not forget the bread and butter.
  • #29
  • Nov 24, 2017 10:47am Nov 24, 2017 10:47am
First by reading your winning percentage I reckon this system has an edge that is not random.
Second this model trades rather infrequently compare to most systems, which reduces the trading cost significantly (my single asset system trades around 20 times a year).

I do have few questions though

1. Has the result been achieved by parameter optimization before logic adjustment, or vice versa?
2. What's the general direction of your system design and do you incorporate trailing TP in your system?

Thanks and I hope there's room for learning and discussion.
No Highs? No Lows? Must be Bose!
  • #30
  • Nov 24, 2017 11:07am Nov 24, 2017 11:07am
Quoting Tardigrade
First by reading your winning percentage I reckon this system has an edge that is not random. Second this model trades rather infrequently compare to most systems, which reduces the trading cost significantly (my single asset system trades around 20 times a year). I do have few questions though 1. Has the result been achieved by parameter optimization before logic adjustment, or vice versa? 2. What's the general direction of your system design and do you incorporate trailing TP in your system? Thanks and I hope there's room for learning and discussion....

Hi Tardigrade,

The system is parameterless apart from the SL. There is no TS or TP although inclusion boosts the model's return.

Hmmm...naively, I didn't expect any genuine Quants to be on FF. Now I have concerns about leaking my edge!
  • #31
  • Nov 24, 2017 11:24am Nov 24, 2017 11:24am
Quoting mbrown
That is all you need to do. When I post the line of code overwrite line 22 in the NewGV script then load it on a chart and recompile the EA. Do you follow?
The EA you have provided on the first page is an ex4 file, so it cannot be recompiled...
I am who I am
  • #32
  • Nov 24, 2017 12:31pm Nov 24, 2017 12:31pm
Quoting imwim
{quote} The EA you have provided on the first page is an ex4 file, so it cannot be recompiled...
Thanks. After loading NewGV reload the EA onto the chart.
  • #33
  • Nov 24, 2017 12:53pm Nov 24, 2017 12:53pm
Quoting mbrown
{quote} Hi Tardigrade, The system is parameterless apart from the SL. There is no TS or TP although inclusion boosts the model's return. Hmmm...naively, I didn't expect any genuine Quants to be on FF. Now I have concerns about leaking my edge!

LOL just friendly exchange. Whenever I hear "parameterless" "non-parametric" I say "good", whenever I hear "Normal" "Log-normal" I call BS.

My focus is on nonparametric and my knowledge in parameterless and pattern recognition is conceptual at best.

Cheers.
No Highs? No Lows? Must be Bose!
  • #34
  • Nov 24, 2017 1:59pm Nov 24, 2017 1:59pm
Quoting Tardigrade
{quote} LOL just friendly exchange. Whenever I hear "parameterless" "non-parametric" I say "good", whenever I hear "Normal" "Log-normal" I call BS. My focus is on nonparametric and my knowledge in parameterless and pattern recognition is conceptual at best. Cheers.

What do you think of mean reverting strategies? I haven't done any testing of this area but my gut tells me it maybe worth exploring away from mean strategies. I guess mean reverting strategies are class as "normal" "log normal" but I guess it depends on the window that is being used.
  • #35
  • Nov 24, 2017 3:00pm Nov 24, 2017 3:00pm
Quoting mbrown
{quote} What do you think of mean reverting strategies? I haven't done any testing of this area but my gut tells me it maybe worth exploring away from mean strategies. I guess mean reverting strategies are class as "normal" "log normal" but I guess it depends on the window that is being used.

I've spent huge amount of time on pairs trading strategy R&D, though not on single-asset mean reversion strategy.
I can only speak from my own experience but tend to agree with what Tyler said on ZeroHedge, "arbitrage is dead".
The first reason being, volatility drives arbitrage(Pair Mean Reversion), hence a low volatility environment such as today may just be stats arb's downfall. See https://www.cfapubs.org/doi/abs/10.2469/faj.v66.n4.1
The second reason pairs trading doesn't work anymore is Alpha factories are prevalent and copying each other's strategy till they milk the very last penny of it.
So a more innovative approach is needed to profit from such dog-eat-dog mean reversion world.

In forex? Yes, away from mean reversion is wise for we have CENTRAL BANKS AND UNHINGED INSIDERS TO DELIVER CONSTANT SURPRISES.
No Highs? No Lows? Must be Bose!
  • #36
  • Nov 24, 2017 3:34pm Nov 24, 2017 3:34pm
Eleven more EURUSD EAs will be made available over the weekend. These EAs past my demo account test and made it to the live account unfortunately they were not a good fit with the other members of the portfolio. The live account exposure was for a very short period (7 weeks) but they are promising as you can see in the live account pip performance chart.

Obviously, I expect you to deploy them on a demo. Live trade at your own risk.
Attached Image (click to enlarge)
Click to Enlarge Name: New Bitmap Image.jpg Size: 47 KB
  • #37
  • Nov 25, 2017 12:34am Nov 25, 2017 12:34am
Hi, I looks to me like you're just trading, and just want to talk about the whole process of trading, cos its your hobby.

Quickly, I don't get one thing, your winrate is 60% of trades are profitable, yet your tag line thing says 90% o the time you are right in your forecasts. Seems like a big discrepancy.

That said, 5% per month win rate is respectable, and 10 percent max drawdown, seems stable, perhaps if you have strong confidence in your strategy, you may want to double or triple your trade size and extend that drawdown to 20-30 percent which is still a safe region IF you are confident in your strat. Speaking of are you discussing strategies on this thread?
HMSR
  • #38
  • Edited 10:51am Nov 25, 2017 3:12am | Edited 10:51am
Quoting Hmsr
Hi, I looks to me like you're just trading, and just want to talk about the whole process of trading, cos its your hobby. Quickly, I don't get one thing, your winrate is 60% of trades are profitable, yet your tag line thing says 90% o the time you are right in your forecasts. Seems like a big discrepancy. That said, 5% per month win rate is respectable, and 10 percent max drawdown, seems stable, perhaps if you have strong confidence in your strategy, you may want to double or triple your trade size and extend that drawdown to 20-30 percent which...
Hi,
Thanks for the post.

Forecasts and trades are different things; I traded four times yesterday with 50% success; the forecast had a 100% success rate...enter long. You all may take the thread where ever in terms of discussion, I will merely ramble, post forecasts, signals and EAs etc.

P.S. Regarding hobby:

Millions of trader-wannabes rush into the most competitive game called financial speculation every year, only around 1% of them come out on top.Attachment 2575275 (https://www.forexfactory.com/attachment.php?attachmentid=2575275)

Thousands of entrepreneur-wannabes in the US rush into creating their own businesses every year, around 25% of them fail every year.
https://www.statisticbrain.com/start...e-by-industry/ (https://www.statisticbrain.com/startup-failure-by-industry/)

Ten males chasing a beautiful and intelligent woman at the same time, at least 10% of them get what they intended.(No citation for this one, it could be 0% if she's a lesbian )


See the difference?

Instead of thinking about "How can I become a consistently profitable trader?", you should be asking yourself, "How the hell do you win a highly competitive game?"

Borrowing some wisdom from the Animal Planet, to survive and thrive in the wild one should be either

1. Faster, like the cheetah
2. Stronger, like the elephant
3. Stealthier, like the chameleon
4. Smarter & slyer, like the Homo sapiens
5. Work in groups, like ants

Who is faster, stronger, stealthier, smarter than you but also work in groups?
Hedge fund, central banks, insiders

How do you grab meat from their hand?
You don't. You eat their leftovers. Find those leftovers by read, trial & error, rinse, repeat. Don't you ever commit a fucking penny on live account before successfully trading a refined strategy on demo which is usually getting better after 500 iterations with the assistance of detailed journal recording every adjustment, either that or find a patron kind enough to support your gambling habit.


Also, start with a less competitive market will help, for instance, anything but forex?!

https://www.forexfactory.com/showthread.php?p=10527826#post10527826
The dude who wrote this post is like the articulate version of me. He's even reached the same conclusions about data and forecasting. His posts are worth reading!
  • #39
  • Nov 25, 2017 4:18am Nov 25, 2017 4:18am
Quoting mbrown
{quote} That is all you need to do. When I post the line of code overwrite line 22 in the NewGV script then load it on a chart and recompile the EA. Do you follow? {quote} Eventually
I believe you mean "Compile the script"? The file with the name NewGV.mq4? Sorry if I am wrong here.
  • #40
  • Edited Nov 26, 2017 3:26am Nov 25, 2017 5:16am | Edited Nov 26, 2017 3:26am
Quoting shiva
{quote} I believe you mean "Compile the script"? The file with the name NewGV.mq4? Sorry if I am wrong here.
Correct. Sorry for the confusion instructions (I'm doing far too many things at once) but they will all iron out if demo is used first.

Upload tomorrow as Jack Daniels is visiting.


261117 9:44:
@ SUBSCRIBERS - UPLOAD YOUR FAVOURITE INDICATORS (THE SIMPLER THE BETTER) AND I'LL MAKE A PARDO ROBUST EA STRATEGY FROM IT.

@ POLL: USDJPY RECEIVED THE HIGHEST VOTES. CURRENTLY COMPLETING THE MODEL FOR DEPLOYMENT BY TOMORROW. ADDITIONAL EAs WILL BE UPLOADED THIS EVENING.

USDJPY: 500 TO 1
POSITION SIZE: 1% OF ACCOUNT
ASSUMES AN ACCOUNT SIZE OF $10000
IDEAL SL IS NO SL. FOR THOSE WHO MUST HAVE A SL, SET IT TO 150 PIPS.

So the USDJPY is a new model so this is now at the forward testing stage. It has past all my robustness tests but follow signals with caution.

PLEASE NOTE: ALL DAILY SIGNALS WILL BE POSTED TO THE OPENING PAGE (FIRST POST).
Attached Image (click to enlarge)
Click to Enlarge Name: SCREEN SHOT.jpg Size: 171 KB
1 trader viewing now