Goal of a successful trader is to make the best trades. Money is secondary
- #1,261
- Edited 2:19pm Oct 5, 2016 1:52pm | Edited 2:19pm
- | Commercial User | Joined Jul 2016 | 1,354 Posts
- #1,262
- Oct 5, 2016 2:22pm Oct 5, 2016 2:22pm
- | Commercial User | Joined Jul 2016 | 1,354 Posts
Goal of a successful trader is to make the best trades. Money is secondary
- #1,263
- Oct 6, 2016 6:24am Oct 6, 2016 6:24am
- Joined Sep 2012 | Status: AlgoTrader | 3,610 Posts
No guts, no glory
- #1,264
- Oct 6, 2016 6:40am Oct 6, 2016 6:40am
- | Commercial User | Joined Jul 2016 | 1,354 Posts
Goal of a successful trader is to make the best trades. Money is secondary
- #1,265
- Oct 7, 2016 12:09pm Oct 7, 2016 12:09pm
- | Commercial User | Joined Jul 2016 | 1,354 Posts
Goal of a successful trader is to make the best trades. Money is secondary
- #1,266
- Oct 7, 2016 1:18pm Oct 7, 2016 1:18pm
- Joined Sep 2012 | Status: AlgoTrader | 3,610 Posts
No guts, no glory
- #1,267
- Oct 8, 2016 7:16am Oct 8, 2016 7:16am
- | Commercial User | Joined Jul 2016 | 1,354 Posts
Goal of a successful trader is to make the best trades. Money is secondary
- #1,269
- Oct 11, 2016 9:29pm Oct 11, 2016 9:29pm
- | Commercial User | Joined Jul 2016 | 1,354 Posts
Goal of a successful trader is to make the best trades. Money is secondary
- #1,271
- Oct 14, 2016 11:39pm Oct 14, 2016 11:39pm
- Joined Sep 2012 | Status: AlgoTrader | 3,610 Posts
No guts, no glory
- #1,272
- Oct 16, 2016 2:14am Oct 16, 2016 2:14am
- | Commercial User | Joined Jul 2016 | 1,354 Posts
Goal of a successful trader is to make the best trades. Money is secondary
- #1,273
- Edited Oct 17, 2016 1:47am Oct 16, 2016 11:29am | Edited Oct 17, 2016 1:47am
Hi Thread,
I think a robot should reflect (as close as possible) a traders way of trading.
So in general I think a Robot should act like you are acting, hence the name, Robot. So when someone has an idea for a Robot that idea should just reflect the way that a person trades live. But in general reading thru Khalil's excellent thread many ideas comes from people "..have great expectations..." or "..I have this feeling it might work..." et cetera. Instead use this approach, tell us how you trade! Simple as that and if you are successful with it or not. And for sure 2 weeks of success is not enough to have a valid trade setup, think 2 years or longer.
I come from the millimeter-bloc-age meaning I waited one day to get stock Open/Close(High/Low and draw my own candles on a paper. From that I then draw trend lines and Zones. So I am not that good with indicators (yet) coz they did not exist for me in the late 80's. I have a introduction of my self here if anyone is interested, it is a sad story
So I trade basically naked, I am not naked but my charts are. How can I write down the rules and later create a Robot for that? Very hard it seems due to a lot of discretion during trading. I will try to do that anyway BUT it might take several posts before I get done. I hope I don't clutter this great thread and if someone think I am I can stop. Just tell me.
Naked Trading-Setup and Problems
**** Setup ****
- (A) Find a Trend (oh no..not moving averages AT ALL)
- (B) Price Action with that trend line
- Time frame
- Stop Loss
- Take Profit
**** Problems ****
(A) MA is not a trend. Forget that. Do you as a person wanna be Average? Follow the moving average and become one. For a Down trend you need 3 components; A + C with a low B look at the image here and I only trade at C. I want A to happen, then wait for B to occur then wait for C to happen. Then look at this image where the 1,2,3 are..it can be so that I have drawn trend lines here live (with Hind Sight we are so good). But I let my Stop Loss take care of that coz when I am in the real trade at C my profit exceeds it by far. Opposite when trading for a Up trend. and then we also have the problem when Trend Change. On this image here I have marked up a area of interest for early next week to not be The Hind Sight Guru.
(B) Price Action is (the look and behaviour of a Candle/s) is easier to imitate for a Robot (I think). This is how I think when Candles touches Trend Lines should look when we can take a Put trade. Image is here.
The main problem is the trend line. Is it possible to solve? I found this Machine learning website here and I am not sure how to incorporate that into a Robot. Maybe it can be done otherwise.
So this could be the start of the "Wendy" - Naked Chart, PA-Robot.
I think a robot should reflect (as close as possible) a traders way of trading.
So in general I think a Robot should act like you are acting, hence the name, Robot. So when someone has an idea for a Robot that idea should just reflect the way that a person trades live. But in general reading thru Khalil's excellent thread many ideas comes from people "..have great expectations..." or "..I have this feeling it might work..." et cetera. Instead use this approach, tell us how you trade! Simple as that and if you are successful with it or not. And for sure 2 weeks of success is not enough to have a valid trade setup, think 2 years or longer.
I come from the millimeter-bloc-age meaning I waited one day to get stock Open/Close(High/Low and draw my own candles on a paper. From that I then draw trend lines and Zones. So I am not that good with indicators (yet) coz they did not exist for me in the late 80's. I have a introduction of my self here if anyone is interested, it is a sad story
So I trade basically naked, I am not naked but my charts are. How can I write down the rules and later create a Robot for that? Very hard it seems due to a lot of discretion during trading. I will try to do that anyway BUT it might take several posts before I get done. I hope I don't clutter this great thread and if someone think I am I can stop. Just tell me.
Naked Trading-Setup and Problems
**** Setup ****
- (A) Find a Trend (oh no..not moving averages AT ALL)
- (B) Price Action with that trend line
- Time frame
- Stop Loss
- Take Profit
**** Problems ****
(A) MA is not a trend. Forget that. Do you as a person wanna be Average? Follow the moving average and become one. For a Down trend you need 3 components; A + C with a low B look at the image here and I only trade at C. I want A to happen, then wait for B to occur then wait for C to happen. Then look at this image where the 1,2,3 are..it can be so that I have drawn trend lines here live (with Hind Sight we are so good). But I let my Stop Loss take care of that coz when I am in the real trade at C my profit exceeds it by far. Opposite when trading for a Up trend. and then we also have the problem when Trend Change. On this image here I have marked up a area of interest for early next week to not be The Hind Sight Guru.
(B) Price Action is (the look and behaviour of a Candle/s) is easier to imitate for a Robot (I think). This is how I think when Candles touches Trend Lines should look when we can take a Put trade. Image is here.
The main problem is the trend line. Is it possible to solve? I found this Machine learning website here and I am not sure how to incorporate that into a Robot. Maybe it can be done otherwise.
So this could be the start of the "Wendy" - Naked Chart, PA-Robot.
- #1,274
- Oct 16, 2016 11:41pm Oct 16, 2016 11:41pm
Good Morning,
To me it is clearly wise to build an indicator taking the signals you want to trade from/on. So I did a simple Highest/Lowest (I know there are drizillions out there) and I get spot on the A-B-C drawn on my image. But I also get a lot of other unwanted signals. So my plan is to add every Area-of-Interest into an Array and then check if pattern is U-D-U last U is our C-Area in this example.
I am not old in Mql so this is my approach to reach the final goal - Wendy - but most likely not the smartest way. However proceeding is better then be at stop.
OT from my own two posts: We should somehow setup VPS so that the most promising versions of the Robots already built can be forward tested. Does anyone have a rack for free over to let us use? For instance this thread is of some age already imagine if the testing already had been done from the start we would have some meat on our bones. I will ask my company of they can donate a rack, not sure they will but asking is cheap
To me it is clearly wise to build an indicator taking the signals you want to trade from/on. So I did a simple Highest/Lowest (I know there are drizillions out there) and I get spot on the A-B-C drawn on my image. But I also get a lot of other unwanted signals. So my plan is to add every Area-of-Interest into an Array and then check if pattern is U-D-U last U is our C-Area in this example.
I am not old in Mql so this is my approach to reach the final goal - Wendy - but most likely not the smartest way. However proceeding is better then be at stop.
OT from my own two posts: We should somehow setup VPS so that the most promising versions of the Robots already built can be forward tested. Does anyone have a rack for free over to let us use? For instance this thread is of some age already imagine if the testing already had been done from the start we would have some meat on our bones. I will ask my company of they can donate a rack, not sure they will but asking is cheap
- #1,276
- Oct 18, 2016 12:09am Oct 18, 2016 12:09am
{quote} I can donate by current VPS, going to open a forex account with a broker and they provide a free VPS. I don't mind paying for this current one for you guys to forward test the bots if something promising gets developed.
) Right now I am in the process finding good setups for Stoch-Trend robot. I am using 15 yrs of data and so far it can't even survive that.Has anyone already processed the steps A and B below please come forward with Robot and .set file
So we can hook it up on gselsidi12's VPS.Thx gselsidi12
--- ******************* ---
Let me explain how I think of Robot-phases
- A) Find a good strat (hard enough) and then build a Robot around it
- B) Can it survive backtesting (I agree to Khalil). Backtesting is crap BUT it should survive it at least.
- C) 1 yr of Demo trading and Survive
- D) 1 yr of Live trading small lots and Survive
- E) Robot-Gradudate
So stuff like this takes time for a person like me. False hope does not give me any pleasure.
Backtesting part I divide it into parts. I start with Time of day, then day of weeks to see if there is any angle to a simple setup like that. Still it takes time with so much data to process. The data I have come from GDMFX history. If I still can't make it profitable with time and day then I must let optimization run. And if Robot does not have Time and Day variables I must add it. So on and so forth
But really the hard part is already done by Abokwaik. It is time we do our part to this.
- #1,277
- Oct 18, 2016 12:15am Oct 18, 2016 12:15am
Regarding "Wendy" from my post above. I have hard time program the trend Who hasn't, so I'm not the only one.
If you all could look here where I try to explain (mostly to myself) how I pick a trend. And that is how I want the first part (A) to function in the Robot. Trend-spotting. Maybe a friendly soul here have know how to develop trend-spotting. Maybe I am the only one finding trends like I do too. After trend-spotting comes candle-patterns but that is another story.
Who hasn't, so I'm not the only one.If you all could look here where I try to explain (mostly to myself) how I pick a trend. And that is how I want the first part (A) to function in the Robot. Trend-spotting. Maybe a friendly soul here have know how to develop trend-spotting. Maybe I am the only one finding trends like I do too. After trend-spotting comes candle-patterns but that is another story.
- #1,279
- Oct 18, 2016 1:12pm Oct 18, 2016 1:12pm
- Joined Sep 2012 | Status: AlgoTrader | 3,610 Posts
Hi Thread, I updated Stoch_trend with two things - Time/Day filters - BeltNSuspenders BeltNSuspenders, places pending orders opposite directions of the order in case it does not go as we want to. After this I have been trying for the whole day profitable setup that can survive decades. Right now I can't find it..not saying it does not exists. If you guys uses my set file attached as .zip and load the robot with it and run 2015-01-01 to 2016-10-01 you get huuuuuge profits (ca 150% and 15% drawdown). But when I run the same settings fr 2014-01-01...
No guts, no glory
- #1,280
- Oct 19, 2016 11:32am Oct 19, 2016 11:32am
{quote} Not at all Sandberg, I appreciate your effort, and might join you when I have time.
I am conducting optimization runs and it is slooooooooooowly going forward.