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

Daily Fibonacci Signals
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=4100		 Page 3 of 21
Author:  snailbeard [ Tue Apr 07, 2015 1:45 pm ]
Post subject:  Minuetto Allegretto AUDNZD
Well it is back to the charts after a few days off. I was working through the charts this morning but unfortunately didn't get past A for AUD before something interesting drew my attention. Firstly, do you ever trade AUDNZD? It is not one of my favourites due to variations in spread depending on the broker and type of account and the swap. However, if you are only going with the weekly/monthly direction then the short swap is a bonus. This morning there was an interesting counter trade, but what grabbed my attention was the gap.

Do you see a gap on M1 at 07:30, 7th April 2015 on your chart?
I suppose it depends which data provider you have. Out of 3 brokers the gap is most pronounced on GlobalPrime demo account. It would be very helpful if someone with a real GP account could post a screenshot of AUDNZD M1 before and after 07:30 in case the demo price data is different to the real price data.

In the following side by side comparison price moves quickly and entering a trade during these minutes could be difficult and expensive. I don't have a log of tick by tick price and spread so there is no way of knowing what would have actually happened only that our method of entry must take into account that many pairs have moments each day when the main move of the day can be over and done in a few seconds or minutes:
side-by-side-AUDNZD-M1-0730-7Apr2015.png
Although this is a counter-trend trade, I have observed this on both trend and counter trend, but some pairs more than others. I gave up using real pending orders because of the uncertainty about the spread and slippage when price touches the trigger price.

So whether we are using a manual entry or auto-trader the entry method must have the trade running before the big boys open the flood gates. Sometimes we have several hours price action before the main, but often we also get some reverse price action just before the main move so we cannot rely on pure price action.

It is always good to look at the daily bars first. In the following chart there is such a clear downtrend that it is safer not go against this trend but something else to keep in mind is that after several months of one trend there could be a significant reversal either short lived or for longer:
AUDNZD-D1-March-to-2015-04-07.png
Candle pattern traders would point out the Bullish Harami pattern. Of particular note is that the Doji opened above the previous close and that the close of the Doji was above the open.

H4 is a popular time frame and close of H4 bar in relation to a significant level is often a clue to the strength of support and resistance:
AUDNZD-H4-6Apr2015.png
What I see here is H4 Ema5 rising and H4 bars opening and closing above both blue and red EMAs which to me means don't sell and consider buying although the simple FiboEma would ignore the long term down trend (it follows yesterday's direction) I'm too cautious to trade this manually. Still perhaps I am too cautious and how might we offset the risk with more clues?
Firstly, the green dashed lines corresponds to yesterday's open and close. It is significant that the current open is above the previous close.

Next I'm going to see what the H1 time frame can tell us...
Author:  snailbeard [ Tue Apr 07, 2015 4:10 pm ]
Post subject:  AUDNZD continued
In the next chart the Sma60 and Ema10 form a triangle. The third Doji closes above the Sma60, but more confidence of a breakout is provided by the small pull back of the next bar following breakout of the Sma60:
AUDNZD-H1-Triangular-breakout-7Apr2015.png
Since this is a counter trade we now have the difficulty of choosing an appropriate target and stop loss. Since we might want to limit the target to 30 pips does a stop loss of 15 pips give enough room for the trade to breath?

So far I have hardly mentioned Fibonacci levels. It is becoming a bit tedious to keep using Empty4's built in tool for my preferred settings so I'm thinking of producing an indicator which can plot 3 custom sets of Fibonacci levels. These are Yesterday's high-low (Yellow), Last week's high-low (Wheat colour), Last month's high-low (Magenta).

When we add these to the chart it starts to look a bit crowded but the trick is to focus on the useful bits:

All the interesting stuff happens between A and D, between A and B we are waiting for a strong breakout and at D price moves fast and it might too late to enter a trade. At A the price is pushed back under the Fibs so we hold back but finally there is a big move and price remains above Y.F.100 at 04:45.
AUDNZD-M15-Fibolevels-2015-04-07.png
We can tell from this chart that moving averages are often used for support and resistance. The challenge at B is to believe the breakout will hold. Finally we should take a look at M1 to better understand the timing of the breakout and what might help convince me to take a punt:
AUDNZD-M1-moment-of-entry-0425-07Apr2015.png
At 04:25 We see a complete M1 bar open and close above the second Fibonacci resistance level. There is also increasing divergence of a ribbon of moving averages indicating growing price action. Some traders prefer to use an MA to calculate a stop loss here there are several to choose from.

Although this post was mostly about a counter trend the same principles can be applied to the main trend and with more confidence than a counter trend.
Author:  snailbeard [ Thu May 07, 2015 9:28 am ]
Post subject:  Interesting Gold setup
I have recently been experimenting with trading indexes and gold, and commodities. It took a while to fix issues with risk, lot size, spread and converting between raw prices, pips and points. The manually tailored look-up table inherited from old code had to be rewritten to work on any broker for any type of trade. This part now seems to be reliable. Futures are a bit of a headache because of additional issues with times and dates.

In addition to the usual methods I have also been experimenting with volume indicators and their derivatives.
Most of the time they are just another lagging indicator which follows the price action but sometimes they can complement ATR or diverge from the price action.

The time that I have found these most useful is when (M15, H1) ATR is lagging behind price action which sometimes delays entries. Generally trades entered when ATR is falling or very low tend to fail but then price action can happen very quickly, but then we need to corroborate price action with other clues to avoid false moves.

To get anything useful out of a lagging indicator the smaller the time-frame the better, so on this M1 screen shot of Gold we see clues to sentiment while the price action picks-up. OnBalance Volume (Lime-green) has a steady march down while M1-ATR(14 & ATR(120) both increase. The slow moving averages show a preference to going down, but price action was moving up, so first we have a false breakout, followed shortly afterwards by a sharp move down. OBV initially goes up with the false breakout but exposes the true sentiment as ATR continues to increase and OBV reverses to the sell side.

Note that yesterday's low is today's upper region and take profit is consistent with Fib 161 relative to yesterday's top and bottom:
Gold-M1-interesting-sell-setup-7May2015.png
Author:  snailbeard [ Tue May 12, 2015 8:56 am ]
Post subject:  AUDJPY breakfast pips
While half the world was still asleep there was a buy breakout for AUDJPY. Unfortunately ZB5 dropped it. Analysis paralysis doesn't just happen to humans, ZB5 often has this problem, which is the cause of many good entries being delayed by hours and a cause of a potential win turning into a loss.

In the following M15 chart it is easy to spot the ideal time to enter a buy trade:
At 04:45 CET there was a strong breakout in the dominant direction shown by the thicker green moving average. The problem for ZB5 is that it generally follows H1 Sma60 which of course is lagging. Often pullbacks followed by breakouts are too fast for H1 Sma60 so some magic is required to tell ZB5 to follow a different Sma when this situation arises:
AUDJPY-M15-Missed-Breakout-Signal-0445-12May2015.png
On the H1 chart the thick purple Sma60 is falling and converging with the thick orange 240 Sma (rising).
Since the candle tops and bottoms are moving up there is no doubt that price is generally moving up. Difficult to see on this crowded chart is that Sma20 is moving up and diverging from Sma240:
AUDJPY-H1-Missed-Breakout-Signal-0445-12May2015.png
Finally delving into the log file helps me understand ZB5's confused state at the time of the missed opportunity:

Code: Select all

04:45 AUDJPY: :   *** bM15BreakH2High ***
04:45 AUDJPY: :  st_bM15OC_CrossUpH4MBB: sc_M15OC_CrossUpH4MBB: 4
04:45 AUDJPY: :  M15 Price action good for BUY
04:45 AUDJPY: :  combineD1H4Trend: set UP trend!  bW1DownPullbackOver;
04:45 AUDJPY: :   bM15BreakH2High: 1 bM15BreakH2Low: 0 bM15BodyLenGtAtr: 1 bH1BodyLenGtAtr: 0 bM15H1PriceHasMomentum: 0 bM1Atr100Doubled: 0 bPriceHasMovedEnough: 1 bAdxH1FallenToWeak: 0 bIgnoreH1AdxWeak: 0
04:45 AUDJPY: :   i3M15CandleBias: 1, i3H1CandleBias: 1, i3H4CandleBias: 1
, bM15OC_AboveM15Ema5: 1, bM15OC_BelowM15Ema5: 0
, bM15BollSlopeDirUp: 0, bM15BollSlopeDirDown: 1
, bH1BollSlopeDirUp: 1, bH1BollSlopeDirDown: 0
, bH4BollSlopeDirUp: 0, bH4BollSlopeDirDown: 1
04:45 AUDJPY: :   bLosingInterest:bH4AndH1CandleConflict  , H4Dir:0, H4:1, H1:1 , M15:1
04:45 AUDJPY: :   iFollowDirection:  -1
04:45 AUDJPY: :   CAUTION: bH1MaCnvgBarRangeOnSma60
[/size]

So we can see that it spotted the breakout and that the candles are going up but " iFollowDirection: -1" means it is looking to sell instead of buy. The iFollowDirection needs to be recomputed to understand that H1-Sma240 is acting as support in this particular case.
Author:  snailbeard [ Fri May 15, 2015 9:39 am ]
Post subject:  AUDJPY earlier entry point
After adjusting the current price direction by price action, further testing uncovered several bugs in other entry filters. Debugging a 6000 line function has it's challenges and even the logging conspired against me to produce dangerously large log files. So I had to fix a number of side issues before I could get this to work.

In the following log extract we can now see that everything is flagging a buy signal:

Code: Select all

05:00 AUDJPY: :  , bH1Ma60Up: 1, bH1Ma60Down: 0
05:00 AUDJPY: :  bBuyConfirmedByPrcAct
05:00 AUDJPY: :   bM1AdxAtrBuySignal: 1, bM1AdxAtrSellSignal: 0
05:00 AUDJPY: :   iH4Direction: 1, i3M15CandleBias: 1
05:00 AUDJPY: :   iH1Ma60Dir:  1
05:00 AUDJPY: :   IsM15Y1Breakout():  bY1M15BreakoutFlag:  1 
bM15Y1BreakoutUpFlag:  1 bM15Y1BreakoutDownFlag:  0
05:00 AUDJPY: :   iFollowDirection:  1
05:00 AUDJPY: :   CAUTION: bH1MaCnvgBarRangeOnSma60
05:00 AUDJPY: :  sbEnterByM1AdxAtrDirection, iM1AdxDirection: 1, bBuyConfirmedByPrcAct: 1
[/size]

and shortly afterwards a buy entry is triggered:
AUDJPY-M15-PriceActionBreakout-0506-12May2015.png
Author:  snailbeard [ Fri May 22, 2015 4:05 pm ]
Post subject:  A Brief History of Time
Sorry this is not a contribution by Prof. Hawkings. This is just another rant about issues with MetaCrap particularly SlackTester.

A friend asked me to test out an idea that he'd got from Jesse Livermore, presumably a book and not his ghost. The essence of which is to enter trades at (D1 breakout) higher-highs or lower-lows using a wide stop loss of say 3x ATR. However, the hard part would be adding a feature to stack trades due to a limitation which I need to fix. Now that MetaCrap supports classes this should be more straight forward than managing arrays of arrays.

The method sounds similar to Zorro's breakout script which is just a few lines of code in their first example.
I was so annoyed with SlackTester's limitations with historical data, i.e. all the ways it can be missing, incomplete or wrong and the hurdles one has to jump through to fix it. So somehow I was reminded of Zorro which I looked at just over two years ago when it was the new kid on the block. It appeared to be a bit flaky at that time but after two years I am wondering how things have progressed. Now some members here are making good use of it. However, there are not that many posts here about Zorro the most recent is by Dewey.

Dewey is very positive about Zorro's back testing and optimization capabilities and historical data just happens by magic. However, it looks like most of us are still locked into running our existing code in MetaCrap for the time being...

So my monster EA has quite a few added checks which look for missing bars and I have to set the simulation start date one week before the time the first trades can open, so that various arrays get filled with long term data.
If I start the simulation then expect the EA to look for past bars, those bars might not exist, which is why I let the EA itself work out when it has accumulated enough passed bars to start trading.

One thing that is missing is a check for holes in the historical data after trading has started, not just a few M1 bars here and there but a lot of bars.

Perhaps Jeremy can explain why even GlobalPrime has this issue with historical data. Does Empty4 purge M1 bars more than about 3 months in the past? How come we can't load these historical bars when we need them for back-testing magically like some of these new modern alternatives to MetaCrap?

Here is some example code for spotting holes in MetaCrap's history data, it only checks M1 bars but could be generalized for other time frames:

Code: Select all

//+------------------------------------------------------------------+
void CheckBackBarAndTime()
{
	if( IsTesting() == false ) 
		return;

	static bool bStartHoleInM1Data = false;
	static datetime dtStartTimeM1Hole = 0;
	bool bFoundCorrectM1Bar = false;
	string strM1HoleDetails;
	
	datetime dtNewM1BarTime = iTime( NULL, PERIOD_M1, 0);
	datetime dtNewM5BarTime = iTime( NULL, PERIOD_M5, 0);
	datetime dtNewM15BarTime = iTime( NULL, PERIOD_M15, 0);
	datetime dtNewM30BarTime = iTime( NULL, PERIOD_M30, 0);
	datetime dtNewH1BarTime = iTime( NULL, PERIOD_H1, 0);
	datetime dtNewH4BarTime = iTime( NULL, PERIOD_H4, 0);
	datetime dtNewD1BarTime = iTime( NULL, PERIOD_D1, 0);
	
	int  iM1BarDayOfYear = TimeDayOfYear( dtNewM1BarTime );
	int iDayOfyear = DayOfYear();
	
	int minuteCurrent = TimeMinute(TimeCurrent());
	int hourCurrent = TimeHour(TimeCurrent());
	int iNearestH4 = hourCurrent / 4;
	int iSubH4Hour = hourCurrent % 4;

	int iSubM15Minutes = minuteCurrent % 15;

	/* Don't check minutes if simulating M15 or higher */
	if( Period() >= PERIOD_M15 && (iSubM15Minutes != 0) )
		return;
	
	bFoundCorrectM1Bar = true;
	if(  (iM1BarDayOfYear != iDayOfyear) 
		|| (TimeHour( dtNewM1BarTime )  != hourCurrent) 
			|| (TimeMinute( dtNewM1BarTime )  != minuteCurrent) )
	{
		bFoundCorrectM1Bar = false;

		if( bStartHoleInM1Data == false )
		{
			bStartHoleInM1Data = true;
			dtStartTimeM1Hole = dtNewM1BarTime;
			
			strM1HoleDetails = "Start M1 Hole: " + TimeToStr( dtNewM1BarTime, TIME_DATE|TIME_MINUTES|TIME_SECONDS)
			+ ", ServerTime: DofY: "+ iDayOfyear + ", HH:MM:SS: " + TimeToStr( TimeCurrent(), TIME_MINUTES|TIME_SECONDS)
			;
			Print( Symbol()+"_hole: ", strM1HoleDetails );
		}
	}
	
	if( bFoundCorrectM1Bar && bStartHoleInM1Data )
	{
		bStartHoleInM1Data = false;
		dtStartTimeM1Hole = 0;
		
		strM1HoleDetails = "End M1 Hole: " + TimeToStr( dtNewM1BarTime, TIME_DATE|TIME_MINUTES|TIME_SECONDS)
		+ ", ServerTime: DofY: "+ iDayOfyear + ", HH:MM:SS: " + TimeToStr( TimeCurrent(), TIME_MINUTES|TIME_SECONDS)
		;
		Print( Symbol()+"_hole: ", strM1HoleDetails );
	}
	//------------------------------------------------------------------
}
[/size]

After running the EA, I initially found the following lines by search the log file:

Code: Select all

20150521.log:287: 0	16:56:00	2014.10.21 00:15  testtimeEA GBPUSD,M15: GBPUSD_hole: Start M1 Hole: 2014.10.21 00:13:00, ServerTime: DofY: 294, HH:MM:SS: 00:15:00
20150521.log:288: 0	16:56:00	2014.10.21 00:30  testtimeEA GBPUSD,M15: GBPUSD_hole: End M1 Hole: 2014.10.21 00:30:00, ServerTime: DofY: 294, HH:MM:SS: 00:30:00
20150521.log:373: 0	16:56:02	2014.11.04 00:45  testtimeEA GBPUSD,M15: GBPUSD_hole: Start M1 Hole: 2014.11.04 00:44:00, ServerTime: DofY: 308, HH:MM:SS: 00:45:00
20150521.log:374: 0	16:56:02	2014.11.04 01:00  testtimeEA GBPUSD,M15: GBPUSD_hole: End M1 Hole: 2014.11.04 01:00:00, ServerTime: DofY: 308, HH:MM:SS: 01:00:00
20150521.log:375: 0	16:56:02	2014.11.04 07:30  testtimeEA GBPUSD,M15: GBPUSD_hole: Start M1 Hole: 2014.11.04 07:29:00, ServerTime: DofY: 308, HH:MM:SS: 07:30:00
20150521.log:376: 0	16:56:02	2014.11.04 07:45  testtimeEA GBPUSD,M15: GBPUSD_hole: End M1 Hole: 2014.11.04 07:45:00, ServerTime: DofY: 308, HH:MM:SS: 07:45:00
20150521.log:502: 0	16:56:06	2014.11.21 14:30  testtimeEA GBPUSD,M15: GBPUSD_hole: Start M1 Hole: 2014.11.21 14:24:00, ServerTime: DofY: 325, HH:MM:SS: 14:30:00
20150521.log:503: 0	16:56:06	2014.11.21 14:45  testtimeEA GBPUSD,M15: GBPUSD_hole: End M1 Hole: 2014.11.21 14:45:00, ServerTime: DofY: 325, HH:MM:SS: 14:45:00
20150521.log:716: 0	16:56:11	2014.12.29 00:00  testtimeEA GBPUSD,M15: GBPUSD_hole: Start M1 Hole: 2014.12.26 23:59:00, ServerTime: DofY: 363, HH:MM:SS: 00:00:00
20150521.log:1040: 0	16:56:21	2015.03.12 17:15  testtimeEA GBPUSD,M15: GBPUSD_hole: End M1 Hole: 2015.03.12 17:15:00, ServerTime: DofY: 71, HH:MM:SS: 17:15:00
[/size]

This unfortunately also picks up even just a few missing bars. A few lost minutes here and there is not the problem, it is the big hole at the end:

Code: Select all

20150521.log:716: 0	16:56:11	2014.12.29 00:00  testtimeEA GBPUSD,M15: GBPUSD_hole: Start M1 Hole: 2014.12.26 23:59:00, ServerTime: DofY: 363, HH:MM:SS: 00:00:00
20150521.log:1040: 0	16:56:21	2015.03.12 17:15  testtimeEA GBPUSD,M15: GBPUSD_hole: End M1 Hole: 2015.03.12 17:15:00, ServerTime: DofY: 71, HH:MM:SS: 17:15:00
[/size]

Empty4 is missing just over two months of M1 bars!

You might be wondering why I need those M1 bars at all for testing a D1 breakout method.
The method doesn't need them but I have my other unrelated calculations running which require the M1 bars and I had not planned to run without them :oops:

If the trading logic was confined to a DLL it would be possible to use it with both Zorro and MetaCrap although I have not thought this through so perhaps it is not that simple?
Author:  snailbeard [ Mon May 25, 2015 3:40 pm ]
Post subject:  OBV: more examples
The following posts are a detailed walk through of several time-frames over a period of time - I'll be going into quite a bit of tedious detail about On-Balance-Volume and Price: so you might find reading a book on vector analysis or Ron Suskind's One Percent Doctrine more interesting...

So, if you are still reading this... the first thing that I have noticed is that divergence (on D1) hints at a price reversal, but price can continue to diverge for days, weeks and even months. This then presents another difficulty: when is price reversal just a pull-back or a real long term reverse? Further, to this is the issue of our timescale, mine is generally less than one day to several days. I don't like large stop-losses and long pull-backs as it requires a deeper draw-down and at the same time there is a high chance of a smaller SL of less than 30 pips being hit, especially by non-ECN brokers. However, there are arguments in favour of trading the macro trends. The advantage is less chance of criminal manipulation hitting your stops or just missing your target price. However, this approach fails when the timing of smaller price cycles within the the larger trend mean the entries are too late in the long trend. If you are familiar with spectral analysis and filter design you will know that it is referred to as phase lag. This happens on all time-frames and OBV has been helping me to see pre-price-action money flow. Phase lag happens in my EA a lot. The Fibonacci levels are very significant places where interesting things happen but don't tell us anything about future price direction. When there is a good long strong trend every thing is simple and our EAs are apparently very profitable, especially in forward testing but after a while the links to MyFxBook mysteriously disappears and it is off again to the next exciting idea of the month. So this month's exciting idea (for me) is volume and OBV I have also looked at AD and MFI. For the moment I am focusing on OBV. We can link OBV with ATR and ADX. However, ATR is perhaps one of the most important companions of price action and OBV. I find ATR is often the biggest clue to the start or end of a cycle on a particular time-frame.

In an earlier post I discussed long term trading, holding a position for weeks or months and stacking trades. Radar and others in this forum have done quite a bit of work on stacking trades. The long-term trades are pointless without stacking more trades on the journey to the take-profit signal. However, if we start stacking entries late into the trend we could end up swept along by the market-juggernaut into heavy draw-down or worse:

The following chart is an example of long term fall on D1 for GBPUSD:
gbpusd-D1-short-term-reverse-JanFeb15.png
The OBV is clearly working its way down from Jan 2015 to March 2015. However, Price and ATR follow their own paths. We are looking at this chart with the benefit of hind sight but we need to image that we have just reached the 2nd Feb: We are now in a dilemma: price is rising - how far will it go: should we exit or hold? When a D1 bar opens and closes above the fast blue and red EMAs it confirms price is moving up.
It is a matter of optimization whether or not hold or close.
So 3 typical options could be:
a) a fixed (n x ATR) pips away from the low
b) price crosses MA(n)
c) bar closes above MA(n)

The above chart is a bit squashed and it looks like a minor pullback but actually at 570 pips low to high it is a trade-able counter swing.
So we could have had the following outcomes
a) 3 x D1-ATR results in a exit giving back about 300 pips
b) price crosses D1-EMA(60) has a similar outcome to (a)
c) price crosses D1-SMA(60) is equivalent to 4 x D1-ATR

In addition to the above we might look at alternative or hybrid exit conditions based on OBV.
We can certainly see OBV move up on D1 but compared to the overall long term falling OBV it does not look that significant. Looking at H4 can give a more accurate shorter term outlook. In following chart the strength of the reverse is more apparent:
GBPUSD-H4-OBVandPrice-Rising-26Feb2015.png
OBV has moved up above the previous highest value and the H4 bars are strongly trending
On this time frame it looks like we have a trade-able reverse or a fairly strong sell exit signal.

So my next task is to see if I can turn these combined conditions into meaningful signals...

Since this topic is mainly about Daily Fibonacci signals, how do we apply the above in the context of MN1 and W1 Fibonacci levels?
In the following chart I have added MN1 and W1 but left out the D1 Fibs.
Firstly, MN1 Fib.61.8 suggests a likely take profit.
There are several crossings of W1 Fibs which all produce a good run (in hindsight)
However, since it is a counter trend trade we might prefer closer targets based on W1 Fibonacci levels:
GBPUSD-H4-Fibonacci-context-26Jan2015.png
Author:  snailbeard [ Wed May 27, 2015 4:43 pm ]
Post subject:  The Big Easy
There might be plenty of manipulation and noise on the smaller time scales, but some things are much easier on a daily timescale. I can run back tests on higher time frames with missing M1 data and it is now possible to stack trades (but not manage them as a group). This is a method that works really well on long term trends:
GBPUSD-D1-JLStkMethod-Oct-Dec2014-perfect-conditions_run27May2015.png
This is only achieved under ideal conditions like the following:
GBPUSD-D1-easy-downtrend-Oct-toDec-2014.png
The draw down is reported as 6% but this is taken from running profit which means more free margin and scope for running other trades in parallel.

Although I have added more OBV analysis it hasn't been pulled into an entry filter or exit signal. At the moment it is just a collection of true and false flags.

If that was the result of 3 month forward test we might claim that this is a tried, tested & proven method and tempted to run it live. Unfortunately, the next three months are a roller coaster ride:
GBPUSD-D1-ranging-reversing-Feb-Apr2015.png
The results for this period are of course negative:

Code: Select all

1	2015.03.16 01:00	sell	       1	0.04	1.47365	0.00000	0.00000	0.00	10000.00
2	2015.03.16 01:00	modify	1	0.04	1.47365	1.51041	1.29003	0.00	10000.00
3	2015.03.18 13:00	sell	       2	0.04	1.46735	0.00000	0.00000	0.00	10000.00
4	2015.03.18 13:00	modify	2	0.04	1.46735	1.50620	1.27520	0.00	10000.00
5	2015.03.18 23:15	s/l	       1	0.04	1.51041	1.51041	1.29003	-98.34	9901.66
6	2015.03.18 23:15	s/l	       2	0.04	1.50620	1.50620	1.27520	-103.65	9798.01
7	2015.04.13 01:00	sell	       3	0.04	1.46273	0.00000	0.00000	0.00	9798.01
8	2015.04.13 01:00	modify	3	0.04	1.46273	1.50438	1.25472	0.00	9798.01
9	2015.04.17 13:00	s/l	       3	0.04	1.50438	1.50438	1.25472	-111.78	9686.23
[/size]

So that is about 3% down, which would not be so bad if it came after the previous 10% up, we would still be about 7% in profit. So this method is suitable for anything which has good long trends with limited amounts of ranging periods.

Is there scope for improvement, i.e. earlier detection of ranging and reversing?
Looking at OBV at the beginning of February shows a temporary upward move. So perhaps some hedge entries would be better than hard stop-losses. That would get us over the first hump.

In March OBV starts to drift upwards, by the end of March there is a clear upswing. Perhaps, it is possible to spot this in time to prevent more sells and also start closing running trades....
Author:  snailbeard [ Thu May 28, 2015 10:13 am ]
Post subject:  BoE/FOMC swallows pips
Did I write 'longer time frame trades are less vulnerable to manipulation' ? - oops - that was a bit misleading. Central Banks are the ones mostly likely to instantly wreck your trades or even your accounts after uttering a few words.

Muddle muddle, toil and trouble let the base rate be double!

On the 18th March the BOE and later FOMC released an update which resulted in a swing of about 540 pips. Which was more than 4x ATR - therefore no chance of the trade entered two days earlier surviving up-swing:
GBPUSD-D1-BoE-statements-swings540pips-18Mar2015.png
Firstly, the BoE news causes a temporary spike lower:
gbpusd-M15-BoE-down-spike-1230-18Mar2015.png
Later-on the FOMC has a much bigger but temporary up-swing:
GBPUSD-M15-FOMC-440-pip-news-spike-1200-18Mar2015.png
Could one use a 3 pair hedge instead of a stop-loss to cope with news spikes?

I have just been looking at the Zorro manual and it provides an example of back-testing multiple pairs at the same time.
Author:  snailbeard [ Thu May 28, 2015 3:56 pm ]
Post subject:  OBV rescue (long term trades)
Going back to the long term trend following method
Without OBV filtering there is a sell entry right at the end of a trend just before it reverses:
An order was placed on 13th April and hit the (3 * ATR) stop-loss on 17th April 2015
GBPUSD-D1-OBV-filters-end-of-trend-sell-13Apr2015.png
From the log files:

Code: Select all

Without OBV filtering:

01:00 GBPUSD: :  call IsVpoTriggered(): for  trigger: 1.46277 dM1Open: 1.46273, ask: 1.46283, bid: 1.46273
01:00 GBPUSD: :  IsVpoTriggered: true

=======================================================================
With OBV filtering:

01:00 GBPUSD: LTAsset Entry: Drop Sell: OBV is rising!
[/size]

So the next question is how well does this work in the long term and for other pairs?
All times are UTC Page 3 of 21

Powered by phpBB® Forum Software © phpBB Limited