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

MNB 10.4 Streamlined by Snailbeard
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=2443
Page 8 of 9
Author:  rosst [ Thu Jul 25, 2013 7:09 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

Just reading about all that analysis was giving me a headache! LOL...
Author:  snailbeard [ Thu Jul 25, 2013 7:55 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

Perhaps talking about trends and end of trends is inappropriate.

In nature the 'trend' is towards balance and so in economics there should be a natural balance except we have governments and powerful organisations who believe they have power over nature. Their actions create imbalance and when that imbalance cannot be sustained something must give.

So it is with a pendulum which 'trends' backwards and forwards, so lets drop the word trend and talk about limits, the pendulum ranges between limits past it's natural state of rest.

In the financial world there are groups with supercomputers busy analysing economic data and they need to change their outlook if value(x) > y - nobody can quite agree on these things so 'x' can only move so far before flags start waving. We are told that forex is both fractal and a follower or generator of Fibonacci sequences.

I have tried to use Fibs to predict significant levels but the difficulty is that there are overlapping Fib. sequences.
It is possible to find the right one historically and this also led me to the conclusion that the pivot levels are sometimes real and sometimes not, but there is a real level often near a pivot level. Real pivots are the ones which correspond with real Fib.s - which are easy to find historically.

Perhaps I have been looking in the wrong place, instead of riding along with the pendulum, thinking that we are following a trend perhaps we should move to the top of the pendulum. In the fractal world we should see the same patterns at a higher level.

So far I have neglected to include any data from the weekly bars, but H4 and D1 Bollingers are useful so what about the W1 Bollinger? And what about Fib. on W1? Can we predict 'no go zones'?
Author:  snailbeard [ Thu Jul 25, 2013 5:17 pm ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

In the Weekly screen shot we see the price ranging. If we apply Captain Jack's approach:
from April to May we see 'railroad tracks' at the extremes of the Bollinger and we would simply trade back into the Bollinger. Referring back to the pendulum concept the home is the centre of the Bollinger. However, this pendulum is inside another bigger pendulum, and we must travel up another level to see another story on the monthly chart where physical forces take back control from intelligent idiots with PhD's.
According to the monthly: GU is going down down down. It recently bounced off a Fib. 161.8 (1.48) but it is under the monthly Bollinger and has come back into a zone of uncertainty. In 2008 GU touched 1.3.
If support fails at Fib:161.8 then it could be down down down to Fib:261.8 (1.4).
May 2012: Big fall from long term channel line to bottom Bollinger
Feedback to EA: reduce targets below Bollinger
June 2012: further fall then fast retrace up of 50%
Feedback to EA: strong resistance at Fib 50% of May 2012: reduce targets
July 2012: Doji :
Feedback to EA: Keep modest targets
August 2012: Crossed the 61.8 but closed just beneath it. We have a sequence of three candles top wick smaller than bottom wicks: bias is still up. According to the zig-zag we are 600 pips from the low and 400 pips from the previous high and 250 pips from a long term channel boundary
Start Sept 2012: We need an array to list significant levels:
There is good chance that some of these levels will be reflected in monthly pivots.
Feedback to EA: consider these levels as S/R
Start Oct2012: Sept had a long top wick, Candle closed inside the channel boundary, breakout or reverse?
Feedback to EA: reduce targets
Oct : Doji, closed outside channel, could be breakout, but zone of resistance above doji
Feedback to EA: keep targets close
Start Nov: gets blocked at the zone of resistance and plunges through retrace targets (23.6, 38.1, 50) and fails to reach 61.8 due to strong support as September's low (this forms a W pattern) - therefore suggests an upwards bias, price goes back above mid-Bollinger, November closes above mid-Bollinger,
so we have both an W and a close above mid-Bollinger BUT the close is just below the channel:
The trend line and mid-B (& 38.2) creates a triangle from which to observe a break-out.
Simplify:
bias neutral
Important levels:
Up: 23.6, !!! 0.0(100, 261.8) near 1.63 !!!
December resumes up trend passes 23.6, hits strong support just above 1.63, pulls back, and closes in a zone of support..
GU-M1-Dec2008-to-2013-natural-forces.png
Back to the weekly chart:

Aug2012 penetrates above the Bollinger which almost horizontal therefore a typical ranging turning point.
Therefore feeding back into D1/H4 trading: reduce targets for uptrends: take profits early
The next month is a Doji which closes above the Bollinger: reduce targets, take profits early, anticipate a reversal into the Bollinger.
Further information coming from past candles and higher time frames gives a zone of strong resistance from 1.628 to 1.630. Therefore, the end of trend is not a surprise at all!

So what needs to be coded?
Author:  snailbeard [ Fri Jul 26, 2013 4:00 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

Firstly, we need the ability to allow each pair control over TP and SL

We are currently limited to:
extern bool UseWMidSR1 = false; //MR1/MS1
extern bool UseWSR1 = false; //R1/S1 default
extern bool UseWMidSR2 = false; //MR2/MS2
extern bool UseWSR2 = true; //R2/S2
And we need individual pair adjustments:

int arrTargetMethod[] // Target price method for each pair
int arrStopMethod[] // Stop distance method for each pair
int arrEntryMethod[] // How to choose entry point

double arrSRLevelsP[][] // Sorted List of S/R levels (price) per pair
double arrSRLevelsW[][] // Sorted List of S/R level (weight) per pair
// Variation on above:
double arrMonthlySRLevelsP[][] // Sorted List of S/R levels (price) per pair
double arrMonthlySRLevelsW[][] // Sorted List of S/R level (weight) per pair
double arrWeeklySRLevelsP[][] // Sorted List of S/R levels (price) per pair
double arrWeeklySRLevelsW[][] // Sorted List of S/R level (weight) per pair

// Bollinger extremes:
double arrMN1HighBll[] // Open month upper Bollinger per pair
double arrMN1MidBll[] // middle Bollinger
double arrMN1LowBll[] // lower Bollinger

double arrW1HighBll[] // Open week upper Bollinger per pair
double arrW1MidBll[] // middle Bollinger
double arrW1LowBll[] // lower Bollinger
Looking at GU current W1:
a.i) price left from lower B 2.8 weeks ago (20% of this week left)
a.ii) price has moved 500 pips
b) price crossed Fib. 100 and returned to Fib. 100
c) range since last pull back almost 300 pips:
bias: expect a pull back/ short range target

Looking at GU current MN1:
a) There are some trend lines creating a triangle
b) Price has stopped just inside this triangle
c) There is a super-support & Fib.50, 100 pips down from current price
d) count the number of MN points on the upper triangle TL: [[[ 6 ]]]
bias: retrace, target around Fib.50
e) price is inside a super SR zone centred on Fib.50
possible scenario:
price move down just beyond Fib.50 to support near 1.53
price stops at Fib.50 near 1.53

(W1 chart - not shown) classic CJ 3 stage move up from BB - ripe for pull back
upward outlook: none : if any upward H1H4 trend develops then keep short range targets

(MN1 Chart ) A confirmed breakout could change this view:
Author:  snailbeard [ Fri Jul 26, 2013 5:38 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

Stop-losses, Drawdown and profit

Running the EA with a 300 pip SL over a long period:
Due to the reduced lot size, profits were small but losses were much smaller:
Conclusion:
Better than leaving cash in the bank but not very exciting
39 2013.03.07 07:00 close 12 0.06 1.50003 1.54222 1.48433 72.17 10431.97

Running the EA with 100 pip SL, increases the lot size, the winners are bigger
and the losses are numerous
Total net profit: £28.56 (0.28%)
About same as my bank

In this particular race the snail wins while the hair runs hard and fast but keeps falling into potholes!

Now this is just one pair, but theres quite a bit of correlation between pairs
How many pairs can be trading at the same time without correlation?
Or what if we can tune each pair and find 30 uncorrelated instruments to trade, then we would be quids in.

In fact if we could find 10 unrelated instruments of similar liquidity
10431 would rise to 14310
For a slow and steady growth pension plan this looks fantastic :D
Areas for optimisation
1) individual pair parameters
2) find the optimum stoploss
3) switch from real pending orders to stealth pending orders - adjust the entry price of the stealth pending orders according to better understanding of the current pull back - convert to real pending orders or instant orders at better entry points.
When the price crosses a virtual pending order price: check candles on TF(t) to confirm the breakout, if this works we can have smaller stop-losses and bigger lot sizes!
4) Better trailing profit management, per pair. Using higher time frame information know the end of the range (there is no such thing as a trend ;) )
5) Better entry methods: multi-lot entry/scaling out, individual pair options.

Although, like others I was close to moving on to a fast and furious EA,
This snail can be turbo charged :lol:
Author:  snailbeard [ Fri Jul 26, 2013 4:50 pm ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

EurJpy
According to NB10 on H4 we have the typical reentry set up:

However the weekly view with zig-zag shows a 3-phase completion.

The monthly is even more interesting...
Why do we get a perfect trend line going all the way back to 1992?
Are we about hit some stormy seas?
Author:  snailbeard [ Sat Jul 27, 2013 1:34 pm ]
Post subject:  Better entries and exits

In attached screenshot, we can see how to get a good 4 level counter swing trade, followed by a re-entry of the main swing.

First, price is below the mid Bollinger
Then H1 goes from red to green
And price crosses mid-Bollinger
This is enough to grab a few pips with an instant order, short range TP & SL
Several full candles completely above mid Bollinger: bias towards up
Then H4 goes from red to green (after recent H1): indicates a good move
in case of spikes we need a SL under WS1 (35 to 40 pips)
Note the next five price levels up
But note last weeks high achieved WR1 bit not WMR2
This leaves 4 levels
Consider each level to be a place for
a) take some profit (say .25% ) on crossing each level
b) adjust stop loss - move to break even after crossing first level
So on reaching WR1 we have completed the trade in parts
We are now free to revert to NB10: RSI2 is at the top
price is above WR1 so a pending order can be put just below WMR1
We take several approaches to taking profits
The difficulty in following the 'trend' is the reversals
Using MTF MA & mid-Bollinger we can catch 1 to 5 levels at each reversal.
Once we bail out of an NB trend we can be locked out of a good move either
a) because of no more pivots
b) RSI not reversing
This is why it is suggested to hold on and ride the reversals (this makes sense if you can tell the difference between a retracement and a full blown reversal)
Author:  snailbeard [ Tue Jul 30, 2013 10:12 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

One filter is removing orders because of H1/H4 pull back,
this filter can dramatically cut losses and reduces the required stoploss (which means we get bigger lot sizes - fewer losses and more on the winners :D ),
but then there are also lost opportunities.
However, if the main swing has not quite finished (or still has 100's of pips),it would be good to get back in after the pull back. We can wait for a sign that the pull-back is over and go for a direct order on re-entering the Bollinger at the edge or at the middle,
or in some cases RSI2 is still pinned at the bottom.
Author:  snailbeard [ Tue Jul 30, 2013 11:26 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

The Hunt for Red October

I have been experimenting with some code to work out the pip swings between moves:

H1 Flip:
2013.06.13 07:00
price: 1.56776
PrevPrevH1Flip: 1.55946
PreviousH1Flip: 1.56597
CurrentH1Flip: 1.56776
Pips swing: -18.00000
Prev swing: -65.00000
SRnearPrevPrev: 1.56041
SRnearPreviou: 1.56041
SRnearCurrent: 1.57404

It is not really effective as yet because the crossovers are lagging, so we need to hunt backwards, for the recent high/low (zig zag), the amount moved is more important when both ends are near a significant level - even better if we find that the ZZ is outside the Bollinger in proximity to a pivot.
If we have enough clues then we can inhibit an entry until there has been a pull back or look for a counter-swing trade.

The screenshot shows what we are looking for...
Author:  snailbeard [ Tue Jul 30, 2013 11:36 am ]
Post subject:  Re: MNB 10.4 Streamlined by Snailbeard

Comparing results for H1+H4 pull-back filter:
H1/H4 trend filter ON

1 2013.06.09 21:00 buy stop 1 0.19 1.56097 1.55097 1.59561 0.00 10000.00
2 2013.06.10 04:00 delete 1 0.19 1.56097 1.55097 1.59561 0.00 10000.00
3 2013.06.11 00:00 buy stop 2 0.19 1.56097 1.55097 1.59561 0.00 10000.00
4 2013.06.11 15:00 buy 2 0.19 1.56097 1.55097 1.59561 0.00 10000.00
5 2013.06.14 08:00 close 2 0.19 1.56534 1.55097 1.59561 82.08 10082.08
6 2013.06.18 04:00 buy stop 3 0.20 1.57400 1.56400 1.58973 0.00 10082.08
7 2013.06.18 08:00 delete 3 0.20 1.57400 1.56400 1.58973 0.00 10082.08

H1/H4 trend filter OFF
useMTF4MA_H1H4DeletePending = false;

1 2013.06.09 21:00 buy stop 1 0.19 1.56097 1.55097 1.59561 0.00 10000.00
2 2013.06.11 15:00 buy 1 0.19 1.56097 1.55097 1.59561 0.00 10000.00
3 2013.06.14 08:00 close 1 0.19 1.56534 1.55097 1.59561 82.08 10082.08
4 2013.06.18 04:00 buy stop 2 0.20 1.57400 1.56400 1.58973 0.00 10082.08
5 2013.06.18 13:00 modify 2 0.20 1.56594 1.55594 1.58973 0.00 10082.08
6 2013.06.19 07:00 buy 2 0.20 1.56594 1.55594 1.58973 0.00 10082.08
7 2013.06.19 19:00 s/l 2 0.20 1.55594 1.55594 1.58973 -201.40 9880.68
All times are UTC Page 8 of 9