stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Boulder's Grind
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=4743		 Page 2 of 6
Author:  tomele [ Sat Jun 04, 2016 5:31 pm ]
Post subject:  Boulder's Grind
Very interesting approach. I would love to help you with testing. I could set up a terminal on one of my computers and let it run next week 24/5. Please let me know if you are interested and which settings and pairs to use for the test.

Does GP allow more than one test account? I am also testing HGBnS on GP as well as MADTrader and SwissArmyKnife on FXPro and dont want to mix it all up.
Author:  Boulder [ Sat Jun 04, 2016 7:29 pm ]
Post subject:  Boulder's Grind
tomele wrote:Very interesting approach. I would love to help you with testing. I could set up a terminal on one of my computers and let it run next week 24/5. Please let me know if you are interested and which settings and pairs to use for the test.

Does GP allow more than one test account? I am also testing HGBnS on GP as well as MADTrader and SwissArmyKnife on FXPro and dont want to mix it all up.
Hey Tomele,

Sure thing would be great :)! The more setups and combinations we can test, the faster we shall advance. I'm planning on having 3 - 4 terminals ready for next week.

Terminal 1:
Standard settings.

Terminal 2:
Ceteris paribus apart from TP increments set to 8 pips, starting off with a BasisTakeProfitPips set to 8 pips.

Terminal 3:
Ceteris paribus apart from the trade levels. MinTradeLevel set to 1, following TradeLevels set to 2 and 3 respectively.

Terminal 4:
Ceteris paribus apart from trade levels set to 1, 2 and 3 respectively with TP increments set to 8 pips, starting off with a BasisTakeProfitPips set to 8 pips.

What also could be interesting are the following setting combinations:
1. Ceteris paribus with Trade Levels set to 0.8, 2 and 3 respectively
2. Ceteris paribus with Trade Levels set to 0.8, 2 and 3 respectively with TP increments set to 8 pips, starting off with a BasisTakeProfitPips set to 8 pips.

I'm planning on running BG 4 hours after the markets have opened on Monday and on having the last trade made on Friday 3 hours prior to the weekly close.
Author:  jal811 [ Sat Jun 04, 2016 7:31 pm ]
Post subject:  Boulder's Grind
The inherent nature of using a higher TF DMI (M30 in this case) , is that you would likely be able to enter the trade only +7 bars later on the 5 min chart after visual confirmation of the +/- delta i.e. we need to wait for the M30 bar to close to get the static value of the preceding six 5 min bars.

I am sorry if I come off as a negative poster here. Notwithstanding that I may enter the sin bin after this post, I just wanted to point this out since Steve has spent time coding the EA when it appears exemplary results could be due to the higher TF nature of the DMI indicator used for the visual hypothetical trade opportunity screenshots.

A good weekend to all.

Cheers!
Author:  SteveHopwood [ Sat Jun 04, 2016 8:11 pm ]
Post subject:  Boulder's Grind
jal811 » Sat Jun 04, 2016 7:31 pm wrote:The inherent nature of using a higher TF DMI (M30 in this case) , is that you would likely be able to enter the trade only +7 bars later on the 5 min chart after visual confirmation of the +/- delta i.e. we need to wait for the M30 bar to close to get the static value of the preceding six 5 min bars.

I am sorry if I come off as a negative poster here. Notwithstanding that I may enter the sin bin after this post, I just wanted to point this out since Steve has spent time coding the EA when it appears exemplary results could be due to the higher TF nature of the DMI indicator used for the visual hypothetical trade opportunity screenshots.

A good weekend to all.

Cheers!
The point about the M5 chart thingy is how long to leave the trade open. Boulder wants it left for 15 minutes, which is 3 x 5 minute candles, which is easier to see when trading manually and is how this thingy started.

It would be just as easy to make TradeCandleLife a minutes input. I just coded what Boulder placed in post 1 and subsequently amended.

Boulder used a few key phrases in this from post 1:
I found it remarkably effective once my eye caught the pattern and have been trading it successfully for a few days now. The actual method is very straightforward and mechanical. Therefore it is perfectly suited for automation and could reach its full potential that way. Today I traded NFP with it successfully as well which I consider a powerful test for any strategy. Passing it prompted me to share it here.
I would have ignored his post without these key phrases. Ok, so a few days is not long but I also knew that it would take me longer to post a description of the inputs that it would to code the EA.

For now, let's assume he knows what he is doing.

:xm:
Author:  Boulder [ Sat Jun 04, 2016 8:21 pm ]
Post subject:  Boulder's Grind
jal811 wrote:The inherent nature of using a higher TF DMI (M30 in this case) , is that you would likely be able to enter the trade only +7 bars later on the 5 min chart after visual confirmation of the +/- delta i.e. we need to wait for the M30 bar to close to get the static value of the preceding six 5 min bars.

I am sorry if I come off as a negative poster here. Notwithstanding that I may enter the sin bin after this post, I just wanted to point this out since Steve has spent time coding the EA when it appears exemplary results could be due to the higher TF nature of the DMI indicator used for the visual hypothetical trade opportunity screenshots.

A good weekend to all.

Cheers!
Yes Jal you make a very viable point here. Thanks for putting it out since it is an important one I failed but should have mentioned from the onset, for which I do apologise. This was my prime and only concern as well I had when I started to play around with this. The trades I took survived however. I always went in as soon as the 1st and 2nd M5 histogram bar had sufficient distance and it seemed to work out practically every time within 10 to 15 minutes (even during NFP). Hopefully it proves not to be a stupid fluke :/ or else I'd be feeling incredibly lousy.. I was also and am as we speak working on a lower M15 timeframe regarding the DMI in order to conduct tests with. Thanks for your comment once again to remind us :)
Author:  marcoFF [ Sun Jun 05, 2016 9:56 am ]
Post subject:  Boulder's Grind
This night I start a demo account.
Thanks Boulder and as always thanks Steve :hi:
Author:  Boulder [ Sun Jun 05, 2016 11:28 am ]
Post subject:  Boulder's Grind
I couldn't stop thinking about the potential problems and inconveniences the inherent nature of using the M30 TF would bring during automation of the system as I presented it. With this bothersome thought in the back of my head and in the hope of finding something in order not to have waisted everybody's time :|, I immediately resumed looking for similar patterns applying a lower TF Wilder's DMI - more specifically on the M5 in order to eliminate this effect entirely. It turns out the strategy could very well hold ground here. It seems in order to make it work we'll have to use rather conservative settings and turn it into a true scalper by capitalising fairly quickly and keeping market exposure tight. I'm planning on experimenting with the following configurations.

Terminal 1:
  • Timeframe M5
    Wilder's DMI on M5
    TradeLevel 1 = 2.4
    TradeLevel 2 = 2.8
    TradeLevel 3 = 3.2
    BasisTakeProfitPips = 7
    TakeProfitIncrementPips = 1
    TradeCandleLife = 3 (being 15 minutes)
    StopLossPips = 35
Terminal 2:
Same as Terminal 1 apart from TakeProfitIncrementPips = 2.

Terminal 3:
Same as Terminal 1 apart from TradeCandleLife = 4 (being 20 minutes tops).

Terminal 4:
Same as Terminal 1 apart from BasisTakeProfitPips = 6.

Terminal 5:
Same as Terminal 1 apart from BasisTakeProfitPips = 6 and TradeCandleLife = 4

Terminal 6:
Same as Terminal 1 apart from BasisTakeProfitPips = 6, TakeProfitIncrementPips = 2 and TradeCandleLife = 4

Feel free to experiment with the settings as well. I feel there's much more to discover here in terms of robust patterns and filtering mechanisms, though those are for later stages when this thing's basic model actually proves to works. Will keep working on it :good: .
Author:  Boulder [ Sun Jun 05, 2016 11:51 am ]
Post subject:  Boulder's Grind
marcoFF » Sun Jun 05, 2016 11:56 am wrote:This night I start a demo account.
Thanks Boulder and as always thanks Steve :hi:
Thanks Marco I do appreciate it! :good:
Author:  SteveHopwood [ Sun Jun 05, 2016 1:24 pm ]
Post subject:  Boulder's Grind
Boulder just pointed out a couple of typos in my TP calculation description in post 3. I guess most of you made a mental translation, but I have corrected the typos.

:xm:
Author:  Boulder [ Sun Jun 05, 2016 4:45 pm ]
Post subject:  Boulder's Grind
Guys more excitement! I may have found something more interesting. Have a look at this:
Method 2.JPG
Wilder's DMI set on M5. As you can see there's an interesting pattern hidden there. The pattern consists of 3 subsequent histograms declining or rising. Again the lifetime of a trade at 15 minutes tops. I've indicated with blue the trades that would've been profitable provided that we would be using a feasible pip target, say 7 - 8 pips, and entered right at the open of the 4th candle. Black lines represent either losses or break evens. I've done this exercise for quite a number of days and it seems to hold ground. Losses will be incurred as well but it seems that the wins outnumber the losses over the long run. In combination with the first and original method I'd suspect this to have higher chances to be profitable over the longer run. One particular constraint that could be set up is to forego e.g. triggering a 5th or 6th sell (buy) after 4 or 5 subsequent shorts (longs). There seem to be very useful combinations to be found as well as to how the 3 subsequent histograms are positioned in relation to each other. Or in other words the relationship/pattern between 2 Diffs of 3 falling or 3 rising histograms. I'll be researching this further and keep you posted!
All times are UTC Page 2 of 6

Powered by phpBB® Forum Software © phpBB Limited