Daily Fibonacci Signals

User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

96 pips in a matter of ticks

Post by snailbeard »

The title has a catchy ring to it, but it is rarely good news for us retail traders 'running on Tulsa time' to have a fast deep reverse followed by another fast swing in the trend direction. None of the CAD pairs seemed to work well for SRT. On the other hand it seems that H4 might be a better time frame for trend trading these pairs. However, sometimes there is a paradox: if one trades on the close of the last H4 bar we might miss the price action unless we use pending orders. I have been round this loop before with an earlier version of NanningBob's EA which could produce fantastic profits some of the time based on H4 and pending orders. It was partly the need to have large stop-losses combined with the difficulty of turning off trading (in bad times), which forced me to abandon my version of it.

Lets say you are following GBPUSD on H4 on this chart:
GBPUSD-H4-pull-back-chart-to-1159-14Sep2017.png
So we have the slow moving averages rising and the fast moving averages are falling, therefore we agree that price has pulled back. Also several small doji bars under SMA(20) and under the fast MAs suggest to me that traders are still bearish. Price watchers can wait and see how things pan out, while 'penders' could be interested in catching a move down or a return to the trend. Lets say we are trend followers and we put pending buy price just above the intersection of green and blue MAs. The argument is that we are putting our entry above a resistance level and hope not to be triggered if the price is still going down. In which case, the stop loss might be just under the dashed line of support. However, if we are thinking longer term e.g. riding the trend for a week, then we might consider a deeper stop-loss and we have red and purple MAs to help us choose a bigger stop-loss.

I am quite clear in my mind that I have no idea what is going to happen in the next four hours because I cannot predict the future, only a woolly expectation of bearish before bullish.

Now we can have a look at the next few bars and see how this interpretation actually changes into reality:
GBPUSD-H4-pull-back-chart-to-EndOf-15Sep2017.png
The first thing to acknowledge is that the SRT method fails on these big reverses, which is why it currently won't work on any pair which regularly has these types of bidirectional swings.
The trend based buy entry trader with a pending buy just above the blue MA and a stop-loss just below the red MA might be smiling (CAVEAT: a fast moving price swing is not always kind to pending orders!).

What would the price action trader be doing? Would he be waiting for a break-out at the pending order price?
And what about an EA that works on H4 bars instead of ticks?

To understand this we need to see every tick but best I can do is show you an M1 chart from FxPro:
GBPUSD-M1-96pips-in-a-matter-of-ticks-1401-14Sep2017.png
The main feature is the fast and furious two-way swing of 96 pips which our manual price action trader didn't even see because they just happened to be looking else where for a moment. Then there is my SRT method which operates mostly on M15 and might wake up in another 14 minutes and decide that the day has already swung too much and go back to sleep.

So generally if we could trust pending orders to be honoured, then it would be a good way to catch the trend, unfortunately only live trading over an extended period can give us the true cost of an unlevel playing field

Has anyway actually quantified and documented this as it would very useful to us all?
I could run small live trades using the pending grid method and log the slippage.

However, there is another method I would like to experiment with. It requires H4 bars and every tick and requires multi day trades and significant stop-losses. H4 provides a trend/no-trend framework and the ticks provide ORFB (Open reverse forward break detection) with one addition, which is price has to swing across a significant H4 moving average.

For this scenario I have included H4 SMA(20) which is crossed at least once, whereas the red H4 MA appears to be touched but not necessarily crossed.

Most of the code for H4 is already there but I need to add a new signal for a new H4 based entry method. Since detecting ORFB requires monitoring ticks, back testing is approximate and CPU hungry, so I'll have to be more judicious about testing. So far I have only looked at this one case.

The following output from the log file shows what the tick monitor was able to extract from the data:
GBPUSD-H4-every-tick-catch-ORFB-entry-1400-14Sep2017.png
So this test case produced the desired result, however, I have not yet converted this signal into a trade or added stop-loss and target calculations...
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

H4 pennant breakouts

Post by snailbeard »

In the following H4 chart we see typical sideways ranging. It is also a slightly ambigous chart due to the rising purple, red, green moving averages clashing with the yellow and brown MAs. However, we are guided by a breakout from the triangle formed by ABC. The EA has already determined a potential buy at the close of C, but this needs to be confirmed during H4 bar D rather than waiting for the bar D to close.
GBPUSD-H4-tick-monitor-pennant-breakout-1200-5Sep2017.png
There needs to be a price threshold for us, say we are trading a breakout, then this threshold can be a higher high, a cross of an MA or other resistance level. Also, we could have multiple signals for crossing any of several nearby resistance levels. Lets say for this purpose we are using the blue MA as out breakout threshold.

on the following M1 chart we see several possible entry levels:
GBPUSD-M1-every-tick-catch-ORFB-entry-1236-5Sep2017.png
At D we have the opening of a new H4 bar at 12:00, moving along to E at 12:36 price has reached our first threshold for a buy entry signal. There are more higher highs at F and G, therefore, we have multiple chances to spot this breakout, but can I write an efficient tick monitor to catch these events in real time?
And what if I want to see how it performs with 3 years of historical data?

Although I have quite a lot of H4 analysis already, it was necessary to add a few enhancements regarding the relative positions of the MAs to each other, and to know quickly if price is across a threshold. So we already determined in the previous post that the EA can identify these conditions. So the next questions are how far is this trend going to run and what is the smallest stop loss that doesn't result in lots of stop outs, should we use break-even stops, and should we exit on reverse price swings?

The following chart show it was possible to catch 380 pips from the 5th to 13th Sep 2017:
GBPUSD-H4-pennant-ORFB-breakout-380pips-5to13Sep2017.png
Strong 'clean' trends are quite rare, so relying on these kind of weeks would lead to disappointment

However, sometimes, 'how-far' can be gauged by Fibonacci levels as in the following chart:
GBPUSD-H4-Trend-Fibo-based-take-profit-Sep2017.png
In this case August MN1 high is the first significant pause. If it is possible to re-enter the trend after the pull-back there is another push higher up to MN1 Fib 161.8 where the bulls finally seem to have run out of steam.

The EA is now able to trade these signals, early results for H4 bars & every tick for Sept 2017 are encouraging:
GPBUSD-H4Trend-limited-backtest-part-Sept2017.png
This uses a break-even option and fairly dumb take profit, no trailing stop or Fibonacci level exit as yet.
Every tick testing is painfully slow, so I need to add some fudges to skip over places that H4 determines to be uninteresting.
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

Analysis Paralysis

Post by snailbeard »

Have you ever wondered if musicians and composers will run out of options because so many songs have already been composed in the last few hundred years?
Is forex trading similar to analysing a bunch of notes and predicting the next hit song?
Perhaps predicting hit songs is less difficult than predicting good trade set-ups?

I am currently staring at H4 charts from 2014, does this have any relevance to 2017?
Does the code that I am writing today have any relevance to 2018?

Do entry patterns look similar in 2014, 2017 and in future in 2018?

My answer would be that there are a lot of similarities but many variations on a theme.
Which leaves us in a bit of dilemma:
Do we code all variations individually (which would take forever)?
Or do we find a generic analysis which isn't too vague could be also very hard to define?

Lets start with a vague description of an entry system:
If there is a group of moving averages going down and a bar opens below a falling moving average then we want to enter a sell trade.

This is so vague that it doesn't really get us very far, so we can add more rules like: if another group of averages is going up then don't enter a trade. However, now we find that this is not what we want either, we have to qualify that: but if the up MAs meet the down MAs then we do want to enter a sell trade. After more testing we discover actually if the second set of MAs are diverging strongly then we need to ignore the first set of MAs.

After a while we end up tying ourselves up in knots with what is commonly know as 'analysis paralysis'.
This is generally the fate of most of my attempts at automating a particular method of trading.
I use the back-testing to make sure that I am not curve fitting to a narrow window.

To show this graphically in the following chart I see three potential trade set-ups based on the idea of bars opening below a group of falling MAs:
GBPUSD-H4-similar-patterns-Dec2014.png
First if we look at 15Dec14 to 19Dec14 we see that there was ranging across several MAs. Now looking at A, B and C they all have some similarities:
1) 'A' is a group of H4 bars opening under falling MA: totally unambiguous
2) 'B' is a break-out after a pull-back, but short-lived.
3) 'C' is a bit ambiguous because some fast MA's are rising but at the same time there is a sequence of falling bars and what might be a break-out lower.

The interesting thing is that it is possible for each of these entries to return some profit provided that we understand the context.
1) in the case of 'A' we are in a weak trend and using break-even after say 20 pips will cause a stop-out.Also this could be the start of a big move (but it turns out not to be, so a less ambitious target would return a nice profit).
2) 'B' could be the start of a major move down but stops then forms a rising channel.
3) 'C' is at the top of a rising channel, but we have a sequence of falling bars under falling MAs. The final bar opens just under the meeting of two MAs. This should be a signal to sell and could be short range to the bottom of the channel or a bigger move down.

It turns out that 'C' is an area of strong conflict between bulls and bears, notice the long wick and the increased length of the big down bar indicating an increase in volatility. It turns out that if we don't interpret this correctly then we'll miss out on a big move!

Now if I look at the EA' log for this break-out do we pick-up the signal?

Code: Select all

GBPUSD_2015_002_02012015: 00:00 GBPUSD: :  H4/D1 Poor trading conditions: (bD1H4RofChangeStrong:false && bAtr3D1LtAtr100 && bH4Atr14Lt100)
GBPUSD_2015_002_02012015: 04:00 GBPUSD: :  H4/D1 Poor trading conditions: (bD1H4RofChangeStrong:false && bAtr3D1LtAtr100 && bH4Atr14Lt100)
GBPUSD_2015_002_02012015: 08:00 GBPUSD: :  H4/D1 Poor trading conditions: (bD1H4RofChangeStrong:false && bAtr3D1LtAtr100 && bH4Atr14Lt100)
[/size]

Unfortunately, the EA entry is a victim of lagging analysis, having detected weak trading conditions in recent days, it has failed to understand that market volatility is returning. Just one more tweak should fix that... ;)
You do not have the required permissions to view the files attached to this post.
User avatar
RisklessPips
Trader
Posts: 246
Joined: Mon May 09, 2016 2:24 pm
Location: Nairobi, Kenya

Analysis Paralysis

Post by RisklessPips »

snailbeard » Tue Oct 03, 2017 8:48 am wrote:Have you ever wondered if musicians and composers will run out of options because so many songs have already been composed in the last few hundred years?
Is forex trading similar to analysing a bunch of notes and predicting the next hit song?
Perhaps predicting hit songs is less difficult than predicting good trade set-ups?

I am currently staring at H4 charts from 2014, does this have any relevance to 2017?
Does the code that I am writing today have any relevance to 2018?

Do entry patterns look similar in 2014, 2017 and in future in 2018?

My answer would be that there are a lot of similarities but many variations on a theme.
Which leaves us in a bit of dilemma:
Do we code all variations individually (which would take forever)?
Or do we find a generic analysis which isn't too vague could be also very hard to define?

Lets start with a vague description of an entry system:
If there is a group of moving averages going down and a bar opens below a falling moving average then we want to enter a sell trade.

This is so vague that it doesn't really get us very far, so we can add more rules like: if another group of averages is going up then don't enter a trade. However, now we find that this is not what we want either, we have to qualify that: but if the up MAs meet the down MAs then we do want to enter a sell trade. After more testing we discover actually if the second set of MAs are diverging strongly then we need to ignore the first set of MAs.

After a while we end up tying ourselves up in knots with what is commonly know as 'analysis paralysis'.
This is generally the fate of most of my attempts at automating a particular method of trading.
I use the back-testing to make sure that I am not curve fitting to a narrow window.
Wonderful description of a difficult problem. Decision making (or more precisely the difficulties therein) seems to be a theme on posts today at SHF :smile:

I am trying to move beyond this by handing trade entry decision making to price and then manage what price does and is telling me to do as best I can.

How are you doing it ?

Charles
Trading is a mind game - good job I have a brain
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

Daily Fibonacci Signals

Post by snailbeard »

Well, Charles, I had a good manual trade with Gold yesterday, which would probably have been too difficult for my EA. Firstly, I was buying against strong recent selling but on the basis of strong support and current momentum, but didn't want to get caught out so I went for break even as soon as possible and was checking regularly, then later locked some profit. Without break-even and manual trailing I would have made a loss, but by careful trailing made a good profit. However, if I checked more frequently I could have taken more profit, but wouldn't have got much else done.

So to answer your question, there is a fine line between optimism and delusion. Sometimes we should stop and reflect and ask are we getting better long term results than before or just repeating the same mistakes in different ways. We know that a number of 'best' professional traders have gone from best in their bank to delusional gambler and subsequent prison. So can we learn from this fact? Anyone can make profits in the short term, but how hard is it to be still in the game in 3, or 5 years from now?

It seems that great traders who write books and run courses are often doing it because they had a method which worked but hasn't made money for a long time.

I spent a lot of time paging through Zulu Traders to see if there were any great traders who had a successful history of more than 3 years and I found only one or two making steady profits. The ones on the front page with lots of followers will probably disappear in a couple of months.

Therefore, if it is said that 90% of traders lose their deposit would it be more realistic to expect that the number of long term successful traders is more likely to be 1% or maybe 0.1%?

I don't think I can ever write enough code to have an EA running live for very long, but trying and studying failure is very good for improving ones ability to interpret a chart :D

So in summary
The best indicators for me are
a) candle patterns
b) Fibonacci levels
c) Moving averages
d) The relationships between moving averages
e) The relationships between candles and moving averages

There is an awful lot of code in the EA but if I was starting from scratch the code could be a much smaller collection of generic classes.

Analysis of data results in boolean logic which then helps in making decisions.

However, understanding decisions made by the EA requires sufficient trades or the results are not statistically useful.

In the previous post I wrote about another tweak and here is the initial result:
GBPUSD_H4Trend_waiting_for_MA_corss_late2014.png
So the back test is running on H4 bars only not every tick, which is quick. I have high lighted 02.12.2015 which was previously blocked by lagging falling ATR filters.

From the log file we can see the change:

Code: Select all

00:00 GBPUSD: :  H4/D1 Poor trading conditions: (bD1H4RofChangeStrong:false && bAtr3D1LtAtr100 && bH4Atr14Lt100)

04:00 GBPUSD: :  (Atr3D1LtAtr100 && bH4Atr14Lt100) override by H4IgnoreAtr3D1LtAtr100 (H4HasGrowingMomentum)
04:00 GBPUSD: H4: bH4TEM_PBO_WaitingMaCrossBuy: 0 bH4TEM_PBO_WaitingMaCrossSell: 1

08:00 GBPUSD: :  H4/D1 Poor trading conditions: (bD1H4RofChangeStrong:false && bAtr3D1LtAtr100 && bH4Atr14Lt100)
[/size]

Do you see the gap in entries between 22Dec14 and 2Jan15 - it is both the Western holiday period and a lull in momentum, so that appears to be an improvement in the accuracy of filtering weak volatility.

I also added a fudge to skip tick processing if neither of these flags are true.
According to the back test report over 6 million ticks were processed but according internal tracking:

Code: Select all

GBPUSD_2015_019_19012015.txt:800:20:00 GBPUSD: Ticks:   Cumulative st_countCalls: 3018316
[/size]
the EA analysed about 3 million, which is more than I was expecting but would have reduced the back test time by half.

Code: Select all

2017.10.04 19:13:17.249	GBPUSD,H4: 6208695 tick events (1306 bars, 6209695 bar states) processed in 0:59:48.657 (total time 0:59:59.297)
[/size]
This forces me to approach changes and testing differently. I now have to think about starting tick mode just before lunch or dinner, or having something else to work on while the test is running.
I am still optimistic that results will improve and hopefully it is not a delusion :?
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

H4 Trend 3Year growth

Post by snailbeard »

It has been most members experience here that leaving an EA running unattended leads to losses. Running over such a long period is a great way to uncover bugs as well as dodgy assumptions. Sometimes losses are a necessary part of the method and sometimes we can see an obvious issue with the logic.

This current H4 method is the best so far in terms of stability over a long period of time, it has periods of losses, but the profits accrue over time and outweigh the losses. Gains are modest compared to M15-SRT which can make a lot more in a short time if used carefully.

It would be useful to piggy back some SRTs on the back of the H4 entries to suck up some nice profits. I am currently chewing the cud on this.

In the meantime, I have been, trying to improve pattern recognition (without being too specific). Here is an example of an entry which is initially analysed to be a buy set-up:
GBPUSD_H4_dubious_buy_0000_15Dec2015.png
There are several signs of an up-trend with red and purple MAs rising. However, the current bar is under falling LWMA MA's, and the recent bars are between converging MAs which is often a sign of treading water.

Initially the EA interprets this as a potential buy and a rising breakout triggers a buy entry:

Code: Select all

74	2015.12.15	1	17	buy	27	0.06	1.51725	1.48595	1.57115	0	4988.09
75	2015.12.15	18	32	close	27	0.06	1.50575	1.48595	1.57115	-45.82	4942.27
[/size]

The next chart adds a bit more information and gives us an alternative view:
GBPUSD_H4_Sma200_Sell_1600_15Dec2015.png
So now we see brown Sma200 acting as resistance. In addition we can see triangular merging between the slower and faster moving averages. So now we see a longer term down trend after a pull back, therefore we need the EA to stop looking for a buy entry until all the faster MAs are above SMA200.
In addition, we can see that at 16:00 a dramatic price drop occurs within the H4 bar and we know that this is not a random price move, it is clearly a market mover with expensive software making a decision to move significant funds. This means my EA is pretty stupid if it is looking for a buy entry when market movers are looking for a sell entry. So I need the EA to perform a similar kind of analysis, although I don't have a team of programmers, so I'll have to hack it.

Something I have recently added is the moving average merge matrix which looks a bit like this:
GBPUSD-H4Trend-mergedMAs-5Sep2017.png
This helps in the interpretation of a pull back, breakouts tend to happen after a group of moving averages have merged and this is equivalent to finding the pinch of a pennant flag.

If my analysis was as good as the MMs I would get my sell entry after 16:00 (which would mean being able to use a smaller SL and hence bigger profit), but it isn't: this is how crude analysis switches from a buy to a sell set-up:
GBPUSD_H4-was-bad-buy-now-good-sell-1028-15Dec2015.png
Although this seems like a genuine improvement, I need to rule out accidental curve fitting by trying out the latest version on multiple pairs. In addition I also added Alerts for these set-ups to help with manual trading.
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

EURUSD entry patterns

Post by snailbeard »

H4Trend spotting worked really well on GBPUSD but failed to find good entries on EURUSD and several other pairs. Therefore, without damaging the GPBUSD pattern detection I need to spot additional patterns for other currencies. These patterns are similar but require additional conditions. The following chart shows a long duration sell entry and several inter-day entries:
eurusd-H4Trend-mixed-entry-types-Oct2014.png
In generic terms, we are looking for a triangle followed by a break-out. However, there are an infinite number of variations on this and some combinations work better than others. At A we see brown Sma200 providing resistance for several bars. The Yellow Sma100 is rising and converging towards Sma200 but they never meet, however the purple MA appears to be close to A along with the fast MAs. So we might say these MAs converge with each other and Sma200.

'C' shows a rapid pull-back followed by a rapid cross down, so price is moving against the fast-ma, so we might want to skip 'C', whereas 'D' follows a smaller pull-back and returns to the trend but is only good for a day trade. A few days later there is a potential short at E but this time even less 'pipage' than 'D'.

In the next post I'll show different examples of potential trades and ways to spot them.
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

Tripple H4 Trend entry

Post by snailbeard »

It has been a while since my last post but after a lull in coding its time to get the old grey cells working again.

Did I mention a triple entry system? Perhaps I hinted at it?

After experimenting with CrappyTester for multiple pairs and long periods the usual issue arose of lean periods with little trading or worse steady incremental losses. However, the problem with a single entry is the all or nothing outcome. Looking at some of these barren stretches I noticed that the trades would have made profits if they had been less ambitious. However, that would be less good for better trading periods, which led me too thinking about 3 entries per signal each aiming for a different profit. Perhaps it is the same as having one initial trade and taking some profit after 'a' pips then some more after 'b' and finally 'c' pips. From an implementation point of view they are different but the single entry is less flexible and requires active monitoring whereas the triple entry is more robust in case there is a problem somewhere between the EA and the market. It does not rule out active stealth monitoring of trades, it just means the results should not be significantly different if Empty4 hangs.

The idea follows on from the close proximity stop loss and moving to break even after a certain number of profit pips. The idea of 3 entries for one alert sounds simple but it did conflict with previous assumptions about monitoring open trades and active exits. When starting from scratch there are easier ways to implement a triple entry system but I found it hard to work around some of the dynamic stealth trading functionality and to think of a way that didn't involve starting from scratch.

Lets now see how a working triple entry system could work during a H4 trend using the existing alerts:
The following chart for H4 EURUSD shows the familiar rise in the EURO despite the fragility of the EU:
H4Trend_Buy_precursors_H4_12Jan2018.png
With the convenience of hindsight can we see that at around 8:00, 16:00, 20:00 there could have been entries with no or small reversals. From this I need to know what the EA was doing: does it find the potential entry points?

Code: Select all

08:00 EURUSD:
3H4 Bias:   1 , Priority:  2H4BreakHigher  H4HaramiAtRes
gb3H1BarPriceMoving: 1 gbH1Has3BarBkOutMomentum: 1
[/size]
So from the above at 8 am there are some positive precursors to a buy entry, so now we can have a look at an M15 chart to have a better idea about timing an entry:
H4Trend_Buy_precursors_M15_12Jan2018.png
At 'A' around 02:00 CET there is a breakout and the set-up is quite similar (we can see the merging of faster MAs with support above the red MA. The Asia session jump in volume suggests that a majority of manual/ automated systems are bullish and continuing MA support for price reinforces this view. Now at 08:00 it is the start of the European session and nothing happens immediately but we have an ABC triangle formation followed by continuing MA support along with a pinch of moving averages, which means our fingers should be resting on the trigger and we just need to wait for the signal...
Another look at the log file:

Code: Select all

08:00 EURUSD: :   check H1MaCrossPotential:  gbH1MaCrossPotentialBuy: 1 
[/size]
However, this was a H4 only test and to to see if the EA produces the signal I have to run the much slower tick by tick test.

From the log file I found the precursor flags for 16:00
16:00 EURUSD: H4:
bH4TEM_PBO_WaitingMaCrossBuy: 1 bH4TEM_PBO_WaitingMaCrossBuy2: 1

However analysis seems to have failed to spot a similar pattern at 08:00
so there is no point having a triple entry system when there is low hanging fruit not being picked.
Hopefully its just a small change and then we can get back to the triple entry system...
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

Triple Wedge Breakout Alert

Post by snailbeard »

This is a slight diversion from my main task and fell out of some existing analysis. However, it is an extension of the H4 triangle breakout analysis, where the same pattern occurs on H1 and M15 all at the same time. In this case the 'triple' is not to do with the number of simultaneous entries but the number of time frames exhibiting the same pattern.

If the two methods are combined then it will be quite a mouthful:
"The triple entry triple triangle breakout method" or those of you who prefer abstraction and object oriented programming might call it the 'multiple entry coincident temporal wedge breakout system'.

However, we are only looking at one particular manifestation of this, so in the following extract from the log file we can see that patterns were detected and both alerts for manual trading and signals for auto-trading occurred:
EURUSD_M15_Log_TripleWedgeBreakout_0900_12Jan2018.png
From this we can see that the main MA crossing method reaches stage 3 at 0800 and we get the first H4 based pattern Ma2BarWedge based alert and signal. However, the auto-trade signal is cancelled by a filter which has calculated that price has moved too far already. Most of the time this filter is correct but it struggles to understand strong moves. I have been doing more work on this analysis but have not yet had chance to test the effectiveness. The triple entry method could also mean that this filter is less important than it is for a single entry method.

At 09:00 and 09:34:15 we get alerts and signals for Triple wedge method, again overruled by the filter.
The M15 Chart shows in more detail how the two methods are in some ways variants of the same idea:
EURUSD_M15_TripleWedgeBreakout_0900_12Jan2018.png
From A to C we have the H4 precursor wedge and we also see the merging of the moving averages which are in the ideal arrangement between 08:00 and 0900 and this give better entries than 09:34 in terms of minimal price reversals.

So from this it would be interesting to know if the triple wedge breakout is generally useful over a longer period of time or does it produce a lot of false entries?

I also need to complete the work on improving the signal filter.
You do not have the required permissions to view the files attached to this post.
User avatar
snailbeard
Trader
Posts: 615
Joined: Mon Dec 24, 2012 10:54 am
Location: Just above water somewhere between Oxford & Cambridge

Trading the news

Post by snailbeard »

Are you a successful news trader? I believe some members here trade the news, but I have not spent much time trying to do this. Perhaps, it was because I was caught out by my greedy broker of the time, manipulating spreads and prices. Also, testing such methods is more difficult unless you have all the tick data.

However, there is an interesting event in January 2018 which coincides with an ECB new release.
In the following chart the expectation is that if nothing changes the trend will continue:
EURUSD_D1_FiboTriangle_Dec2017Jan2018.png
The great thing about this chart is the simplicity, unlike my usual charts which are overloaded with analysis this one follows a nice simple ABC triangle plus C,D shows rising bars above a supporting trend line. The strong Doji at D is interesting because it was the day before the ECB announcement and ignores potential bearish news from the ECB and perhaps that’s because there were no hints of a change before hand. So was the strong bullish spike due to the simplicity of an entry method based on this Fibonacci analysis, i.e. D had a higher low then went beyond Fib. 61.8 with momentum? Those traders entering near the crossing could make a nice profit.
However, my EA generally does not like to enter opposite trades immediately after a strong move.

Again the triple entry method is more robust for such an entry, if the timing is right.
I would like to understand the thinking and timing behind this clever pip snatch.
The following M15 chart shows a bit more information about the timing:
EURUSD_M15_wedge_breakout_1230_10Jan2018.png
Although, it is clear that there are rising lows and rising tops, I am not clear about the trigger for the big move at 12:30. However, the clue is in the time 12:30, so I looked at the 30 minute bars and noticed a beautiful wedge with a final doji:
EURUSD_M30_wedge_doji_breakout_1230_10Jan2018.png
So at E we have the final bar of a wedge which is also a doji,
then during the following bar at F we see ORFB (Open Reverse Forward Break)
Note that the open is higher than the close of the previous bar and that the low of F is higher than the low of E
and the price crosses the high of E and the Fibonacci resistance line.

That is just one speculative interpretation and this might not be the real basis for the bullish spike.
Perhaps M1 might reveal more information:
EURUSD_M1_wedge_breakout_with_price_action_1235_10Jan2018.png
This chart reveals the detailed timing:
Firstly there is increasing bar size, followed by shrinking bar sizes to a small doji, which appears to be our final wedge. Immediately after the doji we get bullish break out bar, followed by an even bigger breakout bar of 7 pips which closes above the Fibonacci resistance.

This is the kind of price action trigger I have used in trading methods before and works well some of the time.
You do not have the required permissions to view the files attached to this post.
Post Reply

Return to “Automated trading systems”