Steve Hopwood's Fib Target-level retracement auto-trader: compliments Simone | Page 14 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Monster Muncher Manual Trading Discussion: Compliments SH 228 replies

Fib retracement line colors 0 replies

Fib Retracement 13 replies

Fib Retracement 2 replies

Fib Retracement Tool 1 reply

Attachments: Steve Hopwood's Fib Target-level retracement auto-trader: compliments Simone
Exit Attachments

Steve Hopwood's Fib Target-level retracement auto-trader: compliments Simone

  • #261
  • Jul 8, 2010 7:20pm Jul 8, 2010 7:20pm
Quoting scooby-doo
Normally weekly R1 and S1 are respected Monday to Wednesday and usually breakout or trends exceed these areas on Thursday's or if there is very high impact news affecting a currency like interest rate changes etc.

Scoobs.
I have a document I keep. It is called, "Scoobs trading insights".

This has gone into the doc.

The Scoobs Thursday Fib Breakout robot is a while off yet, but it is on my radar.

Cheers, pal.

And now I really am going to bed.

night all. Again.
  • #262
  • Edited Jul 9, 2010 4:59am Jul 8, 2010 8:12pm | Edited Jul 9, 2010 4:59am
Hey Steve-

I've been toying with a system based on weekly periods viewed through the lense of 4H charts. Bringing it up cause I noticed you were taking a look at using periods in some way.

Goes like this:

1)Find the average price of the first 6 four hour bars of the (edit) week.
2)Find the average of the second 6 four hour bars of the week.
3)Find the average of those points on the chart in both time/space.

4)Find the average of the third 6 four hour bars of the week.
5)Find the average of 2nd 6 six bars and 3rd 6 bars.

6)Draw a straight line through point #3 (avg of 1st and 2nd six bar averages) and #5(average of 2nd and 3rd six bar averages)

That gives you a plotted course.

7) Wait for a deviation from that line of approx. half the weekly ATR-in the case of the chart posted G/C half the weekly ATR is 200 points.

8) Buy or sell the deviation and TP when it returns to line, or plus a little.

Why is this a fib system?

Because there are 30 four hour bars in a week. You are taking a ratio of averages of 18 of the 30. This reduces to a ration of 3 of 5, two primary fib numbers. Theoretically then the course plotted by the 3 should become quite magnetic for the rest of the week.

I haven't studied thousands of examples, but quite a few and it seems to hold up pretty well.

Anyway I don't know how programmable any of this is, or if its very relevant to this bot, but can't hurt to mention. Great to see the good progress so far.
Attached Image (click to enlarge)
Click to Enlarge Name: timepricefibdeviation.png Size: 331 KB
  • #263
  • Jul 9, 2010 4:28am Jul 9, 2010 4:28am
Quoting stevefx
Anyway I don't know how programmable any of this is, or if its very relevant to this bot, but can't hurt to mention. Great to see the good progress so far.
It should be programmable, if I get an answer to this question:

Coders, on any given sub-MN1 chart (H4 in this case) I need to find StartOfMonth in this code:
for (int cc = StartOfMonth: cc < x: cc++)

I need the offset from to now back to the first H4 candle in the month. How do I find it?

  • #264
  • Jul 9, 2010 4:34am Jul 9, 2010 4:34am
I know I said I was going to leave my basket of weekly-defined trades running until mid-day. A few minutes ago, the the equity was a satisfying 25% up on when I opened the account and fired off the trades on Tuesday, so I closed them.

As described earlier, I overtraded the account, so similar pips profits in coming weeks will not be the same percentage because I shall not be increasing the lot size.

This is the first week. It may have been a fluke.

For now, I am just sitting back and savouring the flavour of success. Time to

  • #265
  • Jul 9, 2010 5:13am Jul 9, 2010 5:13am
just for fun
10K to 50K in 2 years
10K to 500K in 10 years
With MM but no martingale.
unattended, curve-unfitted & with hard-100pip-SL on all the time

i'll send the crappy code (i can't code, didn't know what mql is 2 wks ago) w/method comments to Steve if any interest there.
Attached File(s)
File Type: zip ftlr mod.zip   304 KB | 498 downloads
:-anything
  • #266
  • Jul 9, 2010 5:19am Jul 9, 2010 5:19am
Quoting SteveHopwood
It should be programmable, if I get an answer to this question:

Coders, on any given sub-MN1 chart (H4 in this case) I need to find StartOfMonth in this code:
for (int cc = StartOfMonth: cc < x: cc++)

I need the offset from to now back to the first H4 candle in the month. How do I find it?

Sorry I had a typo, is edited, first 6 bars of the week, reset each week, so only trading last 12 bars of the week every week.
  • #267
  • Jul 9, 2010 5:42am Jul 9, 2010 5:42am
Quoting SteveHopwood
It should be programmable, if I get an answer to this question:

Coders, on any given sub-MN1 chart (H4 in this case) I need to find StartOfMonth in this code:
for (int cc = StartOfMonth: cc < x: cc++)

I need the offset from to now back to the first H4 candle in the month. How do I find it?

hmmmm....
Try this:

Inserted Code
  int startOfMonthH4Bar = 
iBarShift(NULL,PERIOD_H4,iTime(NULL,PERIOD_MN1,0),false);
?


If it's "week" instead of "month" , then use PERIOD_W1 instead of PERIOD_MN1.

SQ
  • #268
  • Jul 9, 2010 6:48am Jul 9, 2010 6:48am
Quoting tritom
just for fun
10K to 50K in 2 years
10K to 500K in 10 years
With MM but no martingale.
unattended, curve-unfitted & with hard-100pip-SL on all the time

i'll send the crappy code (i can't code, didn't know what mql is 2 wks ago) w/method comments to Steve if any interest there.
Cheers tritom. You can send it to me by email from my profile page. Would you like to post it here so everyone can play with it in backtesting over the weekend?

If we cannot break it, it can become the 'official' version because it will have solved the problem of runaway trades without me needing to do any more coding.

Cheers

  • #269
  • Jul 9, 2010 6:50am Jul 9, 2010 6:50am
Quoting squalou
hmmmm....
Try this:

Inserted Code
  int startOfMonthH4Bar = 
iBarShift(NULL,PERIOD_H4,iTime(NULL,PERIOD_MN1,0),false);
?


If it's "week" instead of "month" , then use PERIOD_W1 instead of PERIOD_MN1.

SQ
Cheers sq. I was going to iterate backwards so this saves me the faff.

I really must investigate iBarShift etc.

Thanks

  • #270
  • Jul 9, 2010 6:58am Jul 9, 2010 6:58am
I promise I won't be cluttering up your thread with any more of this(last one).

Just wanted to point out that profitability could be improved by also buying or selling the breakout away from the average line, and setting a trailing stop on it or otherwise closing at the lower or higher deviation line and reversing the position.

In this ex. not much happening with USD/JPY, but bot would immediately open a sell at 19th bar, and trail it. Also keeps out of buying in steady downtrend.
Attached Image (click to enlarge)
Click to Enlarge Name: usdjpyh4.png Size: 233 KB
  • #271
  • Jul 9, 2010 7:05am Jul 9, 2010 7:05am
Quoting SteveHopwood
Cheers tritom. You can send it to me by email from my profile page. Would you like to post it here so everyone can play with it in backtesting over the weekend?

If we cannot break it, it can become the 'official' version because it will have solved the problem of runaway trades without me needing to do any more coding.

Cheers

ok, here it is it's rather a toy since it works on EURUSD daily chart and nowhere else. But the idea behind it is surely the MOST STUPID TRADING SYSTEM ON EARTH and therefore I'm surprised of its consistence.

Sorry but I removed all the visual routines from the code to have a better "insight" in the code.. sloooowly I'm getting into it.. omg, last time I programmed something it was some 25 yrs ago (Sinclair? Atari? Z80 & MC68000 assembler??)

"Method" explained in code comments.
Attached File(s)
File Type: mq4 Steve Hopwood's Fib Target-level retracement (daily re-break barebone mod) auto-trader.mq4   15 KB | 532 downloads
:-anything
  • #272
  • Jul 9, 2010 7:09am Jul 9, 2010 7:09am
Quoting stevefx
I promise I won't be cluttering up your thread with any more of this(last one).

Just wanted to point out that profitability could be improved by also buying or selling the breakout away from the average line, and setting a trailing stop on it or otherwise closing at the lower or higher deviation line and reversing the position.

In this ex. not much happening with USD/JPY, but bot would immediately open a sell at 19th bar, and trail it. Also keeps out of buying in steady downtrend.
Clutter away Steve. When a trader posts something that is working, I am not going to turn him away. Besides, ftlr hardly needs deep discussion.

You have even graduated to your own folder in my mt4 Experiments/Experts folder. This is a rare achievement.

As is happens, I have spend much of this week coding a robot that requires trendlines. When it comes to making an ea draw trendlines, then I am a diva. Granted, it is sq who told me what to do, as usual, but hey, I can do it now.

How do you calculate the yellow breakout lines?

  • #273
  • Jul 9, 2010 7:32am Jul 9, 2010 7:32am
Quoting SteveHopwood
Clutter away Steve. When a trader posts something that is working, I am not going to turn him away. Besides, ftlr hardly needs deep discussion.

You have even graduated to your own folder in my mt4 Experiments/Experts folder. This is a rare achievement.

As is happens, I have spend much of this week coding a robot that requires trendlines. When it comes to making an ea draw trendlines, then I am a diva. Granted, it is sq who told me what to do, as usual, but hey, I can do it now.

How do you calculate the yellow breakout lines?

Lol, thanks Steve. To answer your question-this is new stuff but I'll try to point you in the right direction.

Generally you want to get the lines away from the mishmash of any up and down action that might be happening right around the average line. So according to the pairs hourly ATR and volatility factor you'd push it further or nearer to the avg line.

UJ has an hourly ATR of 14 points for ex, Weekly ATR of 250. 250/14= about 18 to 1.

GC has an hourly ATR of 36, weekly ATR of 400. 400/36= 11 to 1.

So, not only is GC wider ranging, its more volatile on an hourly basis. Therefore you might set GC lines Hourly ATR x2=72. UJ lines Hourly ATR x1.5=21. That would be how I'd approach it, then with some tinkering you could find best fit.

Thanks for the folder!
  • #274
  • Jul 9, 2010 9:13am Jul 9, 2010 9:13am
Quoting squalou
hmmmm....
Try this:

Inserted Code
  int startOfMonthH4Bar = 
iBarShift(NULL,PERIOD_H4,iTime(NULL,PERIOD_MN1,0),false);
?


If it's "week" instead of "month" , then use PERIOD_W1 instead of PERIOD_MN1.

SQ
It turns out this returns the no of bars to exactly a week ago. I can always iterate forward to reach the start of the week.

Cheers

  • #275
  • Jul 9, 2010 10:45am Jul 9, 2010 10:45am
@ Tritom

Just downloaded your bot and am looking forward to seeing what is the stupidest trading system on Earth. Could you have stumbled across my proprietary manual always lose trading method? Oh no, it seems your one makes profit!

Thanks again!
  • #276
  • Jul 9, 2010 11:10am Jul 9, 2010 11:10am
Quoting SteveHopwood
It turns out this returns the no of bars to exactly a week ago. I can always iterate forward to reach the start of the week.

Cheers

Yes, i found the iTime() function to be highly unreliable when used with higher TFs...
This must have to do with the W1 history not being updated perhaps.
Even with PERIOD_D1 i get crazy readings.

Better use the current chart's TF bars as the calculation basis.
At least the current chart's TF should be up-to-date, right ?

For getting today's and week start time and bar count, i have these formula working fine (it seems...) when used on a chart with a TF lower than or equal to W1:

Inserted Code
  datetime today_start_time =
   Time[0]-(TimeHour(Time[0])*3600+TimeMinute(Time[0])*60+TimeSeconds(Time[0]));

  int today_start_bar = iBarShift(NULL,0,today_start_time,true);
  int today_start_H4_bar = iBarShift(NULL,PERIOD_H4,today_start_time,true);

  datetime week_start_time = today_start_time - (TimeDayOfWeek(Time[0])-1)*86400;

  int week_start_H4_bar = iBarShift(NULL,PERIOD_H4,week_start_time,true);
I would not use iTime() anymore on higher TFs...


SQ
  • #277
  • Jul 9, 2010 11:25am Jul 9, 2010 11:25am
Quoting tritom
ok, here it is it's rather a toy since it works on EURUSD daily chart and nowhere else. But the idea behind it is surely the MOST STUPID TRADING SYSTEM ON EARTH and therefore I'm surprised of its consistence.

Sorry but I removed all the visual routines from the code to have a better "insight" in the code.. sloooowly I'm getting into it.. omg, last time I programmed something it was some 25 yrs ago (Sinclair? Atari? Z80 & MC68000 assembler??)

"Method" explained in code comments.
No wonder I've never been able to make money in this market. Now I know why!!! Thanks for explaining that only 'stupid people' make money. 'Smart' as I am how can I make money?
  • #278
  • Jul 9, 2010 11:31am Jul 9, 2010 11:31am
Quoting Invisible
@ Tritom

Just downloaded your bot and am looking forward to seeing what is the stupidest trading system on Earth. Could you have stumbled across my proprietary manual always lose trading method? Oh no, it seems your one makes profit!

Thanks again!
When I first saw what it does, I searched through all possible threads here on FF.. Found something a bit similar - some EURUSD weirdness described by tdion some 2 yrs ago (http://www.forexfactory.com/showthread.php?t=95895) makes the bot profitable (although in another manner - but consistency is the same). It's only a question of time when it stops.. or maybe it lasts for a while.

I'm playing with other pairs - with no remarkable success yet. I'm able to get profit over a limited backtesting period w/certain trading hours, days, parameters etc.. but this is worthless - with EU you can set anything and it works all the time. Maybe other markets ?
:-anything
  • #279
  • Jul 9, 2010 12:54pm Jul 9, 2010 12:54pm
@ tritom

Having looked at that thread, it seems that this is a quirk of EURUSD. It is interesting that the same method seems to work two years later, after all the market changes that have occurred in the meantime. As you say, without understanding why it works as it does, it is difficult to say how it would continue to work in the future also.
  • #280
  • Jul 9, 2010 1:17pm Jul 9, 2010 1:17pm
Quoting Invisible
@ tritom

Having looked at that thread, it seems that this is a quirk of EURUSD. It is interesting that the same method seems to work two years later, after all the market changes that have occurred in the meantime. As you say, without understanding why it works as it does, it is difficult to say how it would continue to work in the future also.
.. maybe I've got a clue, will hook on some serious backtesting platform with reliable data.. (not only fx) ..the method is lazy and doesn't use tight sl/tp so ticks are not an issue. But metaquotes data before 2007 is absolutely worthless I think.
:-anything
1 trader viewing now