Potential new machine learning style software. | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Machine Learning with algoTraderJo 941 replies

Algorithmic Quant Trading (Machine Learning + Stat-Arb) 25 replies

Forex with Machine Learning Software Project 1 reply

Machine Learning + Retail Forex = Profitable? (Quant) 1 reply

My most recent advancements into machine learning 16 replies

Attachments: Potential new machine learning style software.
Exit Attachments

Potential new machine learning style software.

  • #41
  • Dec 8, 2009 7:54am Dec 8, 2009 7:54am
Quoting sloof
The formula is in the pic
Formula is not that important, please tell us which data you put to Eureqa?
  • #42
  • Dec 8, 2009 7:59am Dec 8, 2009 7:59am
Quoting sloof
Because the information in that image is just waaaay too good =P I'll put again, just make sure you guys don't take credit for it.

So I used the info and applied it to my GU chart and tried live trading off of it... so far it works!
You are simply predicting the values of some kind of oscillator (stoch, macd or something, am i right?) x is simply the bar number (or something else proportional to time)?
  • #43
  • Dec 8, 2009 8:02am Dec 8, 2009 8:02am
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
Quoting toddanderson
Could you explain how you applied this to the chart.
I applied this by creating a custom indicator with just the formula. (I did not need to shift it like you guys will probably have to because the formula is already based on my data)

Quoting Tenax
Formula is not that important, please tell us which data you put to Eureqa?
I fed it data from an indicator I programmed. That I can't give, but I'll tell you it's based off of the MA values with a period of 120. There's a bit more to it than just that though.

Quoting 7bit
You are simply predicting the values of some kind of oscillator (stoch, macd or something, am i right?) x is simply the bar number (or something else proportional to time)?
Nope, I did not put in values of any kind of oscillator, that would be too obvious and probably would not work. It's based on a few indicators actually (MA being one of them), none of which oscillate though.
  • #44
  • Dec 8, 2009 8:14am Dec 8, 2009 8:14am
Sloof, thank you for answers and congratulate you with achieve that results.
Maybe you can post some recent trades?
  • #45
  • Edited 3:22pm Dec 8, 2009 8:24am | Edited 3:22pm
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
Rather than showing trades... since it's still up to interpretation, I'll show you what I see. Using this you guys should be able to replicate what I see using the formula.

EDIT: Sorry, I gotta remove some important images.
  • #46
  • Dec 8, 2009 8:33am Dec 8, 2009 8:33am
Quoting sloof
Rather than showing trades... since it's still up to interpretation, I'll show you what I see. Using this you guys should be able to replicate what I see using the formula.
Hmm... for me it's look like some kind of cycle oscillator, thank for posting
  • #47
  • Dec 8, 2009 9:13am Dec 8, 2009 9:13am
Quoting Tenax
Hmm... for me it's look like some kind of cycle oscillator, thank for posting
Yes, this is equivalent to extracting the dominant cycle(s) and their phase(s) and extrapolating them into the future.
  • #48
  • Dec 8, 2009 9:21am Dec 8, 2009 9:21am
You know... Sloof's discovery could possibly mean the Delta Phenomenon could actually hold some weight.
  • #49
  • Dec 8, 2009 9:27am Dec 8, 2009 9:27am
Or it could be that the particular pair in question has spent most of it's last few months locked in a range.
  • #50
  • Dec 8, 2009 9:32am Dec 8, 2009 9:32am
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
Quoting Rabid
Or it could be that the particular pair in question has spent most of it's last few months locked in a range.
I've never heard of the delta phenomenon ... hmmm I'll look it up

Also, yeah I noticed that ranging could explain this so I've already applied 1M data since 2001 on my MT4 and it's already extracting my custom indicator data. The whole process from start to finish could take a while (12Hrs +)

so until then... I've got final exams to study for =( good luck to all you other college kids!
  • #51
  • Dec 8, 2009 9:42am Dec 8, 2009 9:42am
Quoting sloof
I've never heard of the delta phenomenon ... hmmm I'll
so until then... I've got final exams to study for =( good luck to all you other college kids!
Well... if your discovery can predict the future, I'd forget about them exams...
  • #52
  • Dec 8, 2009 9:44am Dec 8, 2009 9:44am
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
Quoting blogzr3
Well... if your discovery can predict the future, I'd forget about them exams...
hahaha =P well still gotta study... just in case
  • #53
  • Dec 8, 2009 10:24am Dec 8, 2009 10:24am
Quoting sloof
Some more info on the data...

it's based on the 5M data of G/U from 8/20/2009 to 12/8/2009

On the graph (and equation) X represents time. Time b/w each point is 5Min... using this you can graph out the oscillation on a G/U chart +/- some time depending on your time zone, data stored in your MT, etc.


what should be substitute into X and solution is .....?

thanks sloof
  • #54
  • Dec 8, 2009 10:25am Dec 8, 2009 10:25am
Quoting sloof
I've never heard of the delta phenomenon ... hmmm I'll look it up

Also, yeah I noticed that ranging could explain this so I've already applied 1M data since 2001 on my MT4 and it's already extracting my custom indicator data. The whole process from start to finish could take a while (12Hrs +)

so until then... I've got final exams to study for =( good luck to all you other college kids!

hehe...stay tuned for new result^^,

thanks again, man~
  • #55
  • Dec 8, 2009 12:24pm Dec 8, 2009 12:24pm
Quoting sloof
Because the information in that image is just waaaay too good =P I'll put again, just make sure you guys don't take credit for it.

So I used the info and applied it to my GU chart and tried live trading off of it... so far it works!
Your Data Points are obviously no Close Prices. So why do you evaluate a formular that shows the turning points of you data when your "Wave Length" is constant anyway?
  • #56
  • Dec 8, 2009 12:28pm Dec 8, 2009 12:28pm
Quoting sloof
Rather than showing trades... since it's still up to interpretation, I'll show you what I see. Using this you guys should be able to replicate what I see using the formula.
i think i see it. nicely done.
  • #57
  • Dec 8, 2009 3:01pm Dec 8, 2009 3:01pm
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
Quoting Tjpld
Your Data Points are obviously no Close Prices. So why do you evaluate a formular that shows the turning points of you data when your "Wave Length" is constant anyway?
Although the wave length is constant I still made an indicator out of it to test it out visually against my custom indicators (not shown on any image) and the correlations are amazing!

BTW, just an update... I took a peek at the data so far. All I can say is...



I'm just looking at the raw data and graphing it out on excel. I still have to put it through Eureqa or something to get a formula out of it.
  • #58
  • Dec 8, 2009 3:12pm Dec 8, 2009 3:12pm
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
To give you an idea of what I see I attached an image of it...

Oh and if anyone steals this, claiming it as their's, I will sue lol
The only reason why I decided to post here is because of the posts being dated so it can be proven I came up with it first =)
Attached Image (click to enlarge)
Click to Enlarge Name: Untitled.png Size: 124 KB
  • #59
  • Dec 8, 2009 3:49pm Dec 8, 2009 3:49pm
Quoting sloof
To give you an idea of what I see I attached an image of it...

Oh and if anyone steals this, claiming it as their's, I will sue lol
The only reason why I decided to post here is because of the posts being dated so it can be proven I came up with it first =)
so do you actually need the curve fitter?
  • #60
  • Dec 8, 2009 3:57pm Dec 8, 2009 3:57pm
  •  sloof
  • | Joined Apr 2009 | Status: Birejji (ビレッジ) ►FOREX◄ | 141 Posts
Quoting ABoulfoul
so do you actually need the curve fitter?
Well not really, but I'm using it to figure out how the markets "should" behave and see what happens when they don't fit the curves (if there is a difference).
0 traders viewing now