Slope Trend EA Thread...... | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Looking for the slope of a slope histogram? 0 replies

T3 Slope Indicator 5 replies

Trend Slope retracement 19 replies

how to draw mutiple trend lines with same slope 7 replies

I want to understand the term "SLOPE" 5 replies

Attachments: Slope Trend EA Thread......
Exit Attachments

Slope Trend EA Thread......

  • #21
  • Mar 19, 2009 9:12pm Mar 19, 2009 9:12pm
Quoting damonl
Hey Ron -

If you want to handle the cross filter for GbpUsd for H1, that would be great. Since it sounds like you may want to do an additional filter or pair, I will leave that up to you. We still have the cross n-sync filter for both GbpUsd H1 and M30. And since you have experience, anyother pair you would like to optimize using either the cross filter or the cross n-sync filter would be fine.

...........

P.S. I was thinking this would also work well with exotics like GbpChf, EurAud and EurCad. If you have time and want to check those in...
Hi Damon,

Is there some redundancy with assignments I saw on Page 1 of thread? And would you be specific about the slope crossings you would like an individual to do, unless we do all 3 possibilities?

Damon, in optimizing the first full protocol, at the point of doing the SL, it would not post *any* data.

I went on to try the TS, and same thing.

So I have not proceeded past this point.

On other clarification:
at the instruction for the first use of the slow slope, you wrote to use 4 for Fast Slope stop, but the # is missing when you meant (surely) to say to use *5* for the Slow Slope Value there.

I look to hear from you about why Optimization does not post for SL & TS.

As far as my double checking tells me, I have it set up correctly.

Best to all,

Ron
  • #22
  • Mar 19, 2009 9:21pm Mar 19, 2009 9:21pm
You got it... I'll be ready for this coming week then - I have the time. My wife is out of town till Monday...

Quoting damonl
Hey Tom.

Why don't you focus on the Crossing pair for the GbpUsd M30 TF. If you find some of my instructions unclear, please let me know and I will clarify them ASAP.

Damon
  • #23
  • Mar 19, 2009 10:28pm Mar 19, 2009 10:28pm
Hey Guys-

There is an error for the default setting in the EA.

For all testing you need to set the BreakEvenAtProfit = 0. I think this will solve your problems. I will get this corrected and post a new version ASAP, but might not be until the AM.

Hope this helps.

Damon
"Keep your eyes on the helpers" - Mr. Rogers
  • #24
  • Mar 19, 2009 10:50pm Mar 19, 2009 10:50pm
Hey Ron -

Hopefully in the above post you will find your answer to the testing.

Regarding the filters, there are three, and they are the last 3 options in the EA's properties area.

The last 2 are the most complicated and that is why I had you do one of them because of your experience and they have to deal with optimizing two slope settings working in tandom at one time. I gave the 1st one out to a few people for GbpUsd because they deal with only optimizaing the settings for 1 slope value.

I believe I gave you the 2nd filter for the H1 TF, but if you would like to do more you can also do the M30 and/or H1 for the 3rd filter as well. Just let me know so I don't start giving out the same task to a bunch of people.

Damon



Quoting ronrhodes
Hi Damon,

Is there some redundancy with assignments I saw on Page 1 of thread? And would you be specific about the slope crossings you would like an individual to do, unless we do all 3 possibilities?

Damon, in optimizing the first full protocol, at the point of doing the SL, it would not post *any* data.

I went on to try the TS, and same thing.

So I have not proceeded past this point.

On other clarification:
at the instruction for the first use of the slow slope, you wrote to use 4 for Fast Slope stop, but the # is missing when you meant...
"Keep your eyes on the helpers" - Mr. Rogers
  • #25
  • Mar 19, 2009 10:58pm Mar 19, 2009 10:58pm
Hi Damon,

okay!

Would you confirm or adjust me then:

G/U H1 Slopes set: 1 & 2, 1 & 3, 2 &3 ? (to start with?)

And should we wait for the corrected EA in the AM and then optimize 3 / 19-21?

Or if you see me doing other than above to start let me know.

Regards,

Ron
  • #26
  • Mar 19, 2009 11:07pm Mar 19, 2009 11:07pm
SLF Filter = true
M30 TF
Single Slope Value = 14
S/L =27
T/S = 22
T/P = 55
B/E = 22

SLF Filter = True
H1 TF
Single Slope Value = 8
S/L = 30
T/S = 0
B/E = 14
T/P = 80
"Keep your eyes on the helpers" - Mr. Rogers
  • #27
  • Mar 19, 2009 11:13pm Mar 19, 2009 11:13pm
Hey Ron, you are correct with your sequence and to make it easier we thought that we would just use Fib #'s for the Slow Slope Value....

3,5,8,13,21. I think you will not have to go higher than 21 on that one.

As for waiting, Gary will have family thing to deal with tomorrow, but nothing will change. When you test, just change B/E to 0 and you should be good to go.

Damon

Quoting ronrhodes
Hi Damon,

okay!

Would you confirm or adjust me then:

G/U H1 Slopes set: 1 & 2, 1 & 3, 2 &3 ? (to start with?)

And should we wait for the corrected EA in the AM and then optimize 3 / 19-21?

Or if you see me doing other than above to start let me know.

Regards,

Ron
"Keep your eyes on the helpers" - Mr. Rogers
  • #28
  • Mar 20, 2009 2:28am Mar 20, 2009 2:28am
Single Slope Filter (SLF)
Tested: GBPUSD
T/F: 30 M
T/P: 169
S/L: 10
B/E: 32
T/S: 25
S/P: 17

More than one value typically showed up during optimizing leading to choices as to what to input. For example, three slope periods occurred, namely 17-18-19. I used 17. Break Even displayed figures from 21-32. I used 32. Take profit figures ranged, at the low end, from 0 to 170 to 180. Trailing stops ranged from 0 to 100. I arbitrarily input 25.

Steve
  • #29
  • Mar 20, 2009 7:56am Mar 20, 2009 7:56am
On optimizing for the days 19 and 18 the GBPUSD 1H reported 0 trades so I decided to optimize for the period 03/12/2009 till 03/19/2009 (one week worth of data).
I optimized the crossing filter pair (CSF/SSF).
The best results were obtained with:

S/L 57
T/S 62
B/E 11
T/P 157
SlowSlopePeriod 3
FastSlopePeriod 2
TradeUsingCSF True
TradeUsingSSF True

Hope this helps everybody!
Stay on the PIPest side of life!
  • #30
  • Mar 20, 2009 8:00am Mar 20, 2009 8:00am
Quoting damonl
SLF Filter = true
M30 TF
Single Slope Value = 14
S/L =27
T/S = 22
T/P = 55
B/E = 22

SLF Filter = True
H1 TF
Single Slope Value = 8
S/L = 30
T/S = 0
B/E = 14
T/P = 80
Which pair do these results relate to?
Stay on the PIPest side of life!
  • #31
  • Mar 20, 2009 9:35am Mar 20, 2009 9:35am
Hey PIPer -

It will not work quite right if you had both filters set to true at the same time. They really should be tested individually.

And you bring up a good point, and one that I have been thinking out since we started this thread. The plad originally was nightly optimizations, but really very trades get placed sometimes and I think the last 2 days would be skewed a bit because of the G-10 meeting and the huge sudden decrease in the USD.

I am beginning to think that it may be best to optimize and a larger area of time and yet find a way to stay fairly current. I am wondering about maybe a 2 week period of time done over the weekend so we are ready for Monday. And at the very least a week. This will help filter out those chaotic days and smooth things out a little bit.

Do you have any thoughts on this? What about anyone else reading this thread and following along. I mean somedays I would optimize for the the 2 days and would get 3 trades, how can you get good optimation stats for 3 trades.

Also, it appears that you had to make some decisions based on the results being equal for different settings. This will happen a lot, but less if we use a longer testing period, which I think we should think about.

Whatcha think?

Quoting PIPer
On optimizing for the days 19 and 18 the GBPUSD 1H reported 0 trades so I decided to optimize for the period 03/12/2009 till 03/19/2009 (one week worth of data).
I optimized the crossing filter pair (CSF/SSF).
The best results were obtained with:

S/L 57
T/S 62
B/E 11
T/P 157
SlowSlopePeriod 3
FastSlopePeriod 2
TradeUsingCSF True
TradeUsingSSF True

Hope this helps everybody!
"Keep your eyes on the helpers" - Mr. Rogers
  • #32
  • Mar 20, 2009 9:38am Mar 20, 2009 9:38am
Hey Steve -

Thanks for your input. please see my reply to PIPer and give your response to that if you would please.

Also, maybe it is that way in the booklet, but you would probably only need to optimize the T/S feature for 50 pips max.

Damon

Quoting Steveg100
Single Slope Filter (SLF)
Tested: GBPUSD
T/F: 30 M
T/P: 169
S/L: 10
B/E: 32
T/S: 25
S/P: 17

More than one value typically showed up during optimizing leading to choices as to what to input. For example, three slope periods occurred, namely 17-18-19. I used 17. Break Even displayed figures from 21-32. I used 32. Take profit figures ranged, at the low end, from 0 to 170 to 180. Trailing stops ranged from 0 to 100. I arbitrarily input 25.

Steve
"Keep your eyes on the helpers" - Mr. Rogers
  • #33
  • Mar 20, 2009 10:19am Mar 20, 2009 10:19am
Damon: In that this project is still in a ramp-up period, and initial concepts are subject to being re-thought, I have no problem with extending the time frame for optimizing. I think it is at the discretion of the chief scientist, cook and bottlewasher. However, perhaps this should be an empirical issue. The initial concept of frequent, short, up-to-date testing made sense. If experience proves that one gets no added benefit, and a dollup of random artifacts from short-interval optimizing, then extending the time frame would be the logical conclusion. But you have the latitude to make that judgment without a lot of cumbersome ado.

Steve
  • #34
  • Mar 20, 2009 10:25am Mar 20, 2009 10:25am
Thanks Steve -

Gary is out for a while today and will run it by him when he gets in, but I really see no benefit from 2 days of backtesting with few trades taken. On average we should have at minimum 10 - 15 trades, and I think we can get that from the previous week.

Will see what Gary chimes in when he gets back.

I am going to upload the EA with the fixed B/E default feature in the first post now.

Enjoy....Damon

Quoting Steveg100
Damon: In that this project is still in a ramp-up period, and initial concepts are subject to being re-thought, I have no problem with extending the time frame for optimizing. I think it is at the discretion of the chief scientist, cook and bottlewasher. However, perhaps this should be an empirical issue. The initial concept of frequent, short, up-to-date testing made sense. If experience proves that one gets no added benefit, and a dollup of random artifacts from short-interval optimizing, then extending the time frame would be the logical conclusion....
"Keep your eyes on the helpers" - Mr. Rogers
  • #35
  • Mar 20, 2009 11:14am Mar 20, 2009 11:14am
Count me in on cable 30!
  • #36
  • Mar 20, 2009 11:21am Mar 20, 2009 11:21am
Single Slope Filter (SLF)
GBP/USD
TF-H1
SPF=10
SL =21
TS =0
BE =10
TP =0

I agree on the greater number of days to be tested. I had 10 to 15 of the same high profit trades with many different SL values, also the same was true for TS and TP, all the way from a low of 0 to highs over 100. I recorded the lowest number even if it was 0. I also ran my settings on strategy tester and it came up with the same highest profit figure. 90% Modelling quality and 0 Mismatched charts errors. 5 Winners 0 loosers.

Damon : Sorry I didn't get back to you on verifying my assignment.I was out for awhile with my 89 year old Father who fell and broke his hip...hes in good hands now so will be able to continue with testing.

Richard
  • #37
  • Mar 20, 2009 1:00pm Mar 20, 2009 1:00pm
I have been reading and following the threat since the slingshot days. Finally i decided to join last night and just now received posting privileges.

I will be testing and optimizing EURUSD and GBPUSD.

I am at work now, but will start posting results tonight.

Just be patient with me since I do most of my manual trading in the night. When I get home, besides spending time with family, school work and trading FX for profit, it leaves very little time to play with, but i am up for the challenge.

Take care all.
  • #38
  • Mar 20, 2009 1:42pm Mar 20, 2009 1:42pm
Quoting damonl
Hey PIPer -

It will not work quite right if you had both filters set to true at the same time. They really should be tested individually.

And you bring up a good point, and one that I have been thinking out since we started this thread. The plad originally was nightly optimizations, but really very trades get placed sometimes and I think the last 2 days would be skewed a bit because of the G-10 meeting and the huge sudden decrease in the USD.

I am beginning to think that it may be best to optimize and a larger area of time and yet find a...
Damon,
Please note that I Tested 3 possible combinations with the filters:

  1. CSF true & SSF false
  2. CSF false & SSF true
  3. CSF True & SSF true

and the best results were obtained with the 3rd option. Please see the spreadsheet where I recorded the test results.

I think that a 2 week period for the optimization is long enough. To keep the results fairly current we could do the optimization nightly for the last 14 days. So, for example, on the 20 of every month we would use the period between days 5 and 19, onthe 22 of every month we would use the period between days 7 and 21, and so on...
What do you think?

Attached File(s)
File Type: xls Optimização EA.xls   14 KB | 407 downloads
Stay on the PIPest side of life!
  • #39
  • Mar 20, 2009 2:18pm Mar 20, 2009 2:18pm
hahahahaha

Hey PIper -

If you add up the totals for the results of using each one individually you will see they come very close to the totals for using them both together.

They are each different entries, and by putting both to true, all you did was optimize two seperate entries at the same time.

This is not a good thing because they are two seperate entries, and by optimizing both together you are getting an average of what worked well for the both of them rather than what worked best for each one individually. Although they may be close most of the time, may times they may not be.

does that make sense to you?

As for the optimizing, I think it should be done every week at least before monday's open, but it is just a matter of how far back we go each time. 1 week or 2. I agree that no more than 2 is probaby good, but I also like the idea of going back to the previous week.

Any more thoughts on what I just said?

Thanks, and by the way, it looks like you are optimizing correctly, great job!

Damon

I guess they don't all add up, but the best results for both individually sure make up the best results for both together. Optimizzing both seperate will give you the best settings to trade each one individaully on your chart for the best overall results.



Quoting PIPer
Damon,
Please note that I Tested 3 possible combinations with the filters:

  1. CSF true & SSF false
  2. CSF false & SSF true
  3. CSF True & SSF true

and the best results were obtained with the 3rd option. Please see the spreadsheet where I recorded the test results.

I think that a 2 week period for the optimization is long enough. To keep the results fairly current we could do the optimization nightly for the last 14 days. So, for example, on the 20 of every month we would use the period between days 5 and 19, onthe 22 of every month...

"Keep your eyes on the helpers" - Mr. Rogers
  • #40
  • Mar 20, 2009 2:19pm Mar 20, 2009 2:19pm
will do

Quoting silverngold
Count me in on cable 30!
"Keep your eyes on the helpers" - Mr. Rogers
0 traders viewing now