EurAnalysis | Page 3133 | Forex Factory

Options

Go to
Permalinks
Print Thread

Related Instruments

EUR/USD

Similar Threads

EurAnalysis Kindergarten 24 replies

Attachments: EurAnalysis
Exit Attachments

EurAnalysis

  • #62,641
  • Dec 9, 2012 2:08pm Dec 9, 2012 2:08pm
Week of 10 Dec 2012 projections (pl see prev posts for context):

1.2874
1.2952

Note: experimental - do not trade based solely on these projections
  • #62,642
  • Dec 9, 2012 3:04pm Dec 9, 2012 3:04pm
Quoting msood
Week of 10 Dec 2012 projections (pl see prev posts for context):

1.2874
1.2952

Note: experimental - do not trade based solely on these projections


I wonder how you came about your projection.....that 1.2874 seems very much like my mean for the next 16 days.....I am not sure however we come about targets the same way
AVT INVENIAM VIAM AVT FACIAM
  • #62,643
  • Dec 9, 2012 3:05pm Dec 9, 2012 3:05pm
...and the Berlusconi circus starts again...
  • #62,644
  • Dec 9, 2012 3:17pm Dec 9, 2012 3:17pm
Quoting the redlion
I wonder how reliable is extrapolating data from oanda's costumers......I mean in real terms they are a very small portion of retail....and retail doesn't move the market.

i suppose the question is how representative are they of the market?
Here you go..
Look what happen when everybody & Pavlov's Dogs are long here..
Attached Image (click to enlarge)
Click to Enlarge Name: euro longshorts @11022012-horz.jpg Size: 232 KB


and here..
Attached Image (click to enlarge)
Click to Enlarge Name: euro longshorts @07082012-horz.jpg Size: 206 KB
  • #62,645
  • Dec 9, 2012 3:21pm Dec 9, 2012 3:21pm
Quoting the redlion


I wonder how you came about your projection.....that 1.2874 seems very much like my mean for the next 16 days.....I am not sure however we come about targets the same way
Hi - I don't quite know myself either

As I said in the start posts, this is a black box neural net based projection system that is "supposed" to provide a safe trading range for the specified period, which has a "v high" chance of being met. The figures I post are raw numbers out of the system - it does not provide any reasoning or analysis, just these sets of figures.

The lo/ hi projections are meant to represent "a" safe trading zone (there will be others during the period, I am sure). These are not targets that won't be exceeded. In fact the tests show that these projections are often exceeded.

I suppose, one could look at these numbers as something additional to consider along with their principle TA style.

Interesting re the mean for next 16 days - how's that done?
  • #62,646
  • Dec 9, 2012 3:25pm Dec 9, 2012 3:25pm
Quoting cdndollar
Here you go..
Look what happen when everybody & Pavlov's Dogs are long here..
Attachment

and here..
Attachment
oanda traders are usually 200 ticks behind the most recent breakout/selloff. just do the opposite of what the net oanda retail position is and youll probably do pretty decent
Any prices quoted are futures
  • #62,647
  • Dec 9, 2012 3:38pm Dec 9, 2012 3:38pm
Quoting cdndollar
Here you go..
Look what happen when everybody & Pavlov's Dogs are long here..
Attachment

and here..
Attachment

well, at least be careful with those...i attached a chart showing that it's not always wise to be contrary to retail...It all depends.



PS. i used to focus a lot on those and had the same mindset as you do. In time, i figured out that a lot of things can be evened out. Trading's not that simple...The book works like everything else...until it doesn't...and then it does again...
Attached Image (click to enlarge)
Click to Enlarge Name: oanda book.png Size: 41 KB
  • #62,648
  • Dec 9, 2012 3:40pm Dec 9, 2012 3:40pm
Quoting msood
Hi - I don't quite know myself either

As I said in the start posts, this is a black box neural net based projection system that is "supposed" to provide a safe trading range for the specified period, which has a "v high" chance of being met. The figures I post are raw numbers out of the system - it does not provide any reasoning or analysis, just these sets of figures.

The lo/ hi projections are meant to represent "a" safe trading zone (there will be others during the period, I am sure). These are not targets that won't...
ok I will try as simple as I can

when you take the euro returns it resembles a normal distribution correct with heavy tails. usually on a normal distribution , you get the mean and median at 0.

to make price simulation you need to turn the normal distribution curve into a log normal that is done with this formula

price*e^( drift*time + standard deviation*sqrt(time)* random normally distributed numbers)

this is how you simulate potential closing price for the next 16 days....why 16 because the trading month has about 21 days so we have already traded 5 days that is 21-5 = 16

if you believe the market is random your mean would be 0
if you take the historical drift= mean in my opinion you are always looking back and no projections are being made reliably same as taking the historical volatility

so i use the GARCH 1,1 method to project future volatility that goes something like this ..... historical variance* gamma+ last period log return^2*alpha+ last period variance*beta . Now you take the squared root of that to turn it into volatility.

to account for mean regression to historical variance as it has been empirically shown

historical variance+(alpha+beta)^time*(last period projected variance- historical variance)

I do this for 21 periods then use the squared root rule to project into a desired period. by averaging n-1+ t0 / 2 all the way down to 21 periods.

at the end I get a reading which I use to base my assumptions.
the projected drift I use a calculation I came up with ......... absolute value last period's return - squared root of ( last period return^2- projected variance)/ projected daily volatility * squared root of time. if last period return was negative i multiply by -1 if positive by 1

this gives me the annual drift which i divide by 252 (252 trading days in the year) this gives me the daily drift

then generate 100,000 trials and take the mean and the standard deviations to arrive at my targets.
AVT INVENIAM VIAM AVT FACIAM
  • #62,649
  • Dec 9, 2012 3:43pm Dec 9, 2012 3:43pm
Quoting the redlion
ok I will try as simple as I can

when you take the euro returns it resembles a normal distribution correct with heavy tails. usually on a normal distribution , you get the mean and median at 0.

to make price simulation you need to turn the normal distribution curve into a log normal that is done with this formula

price*e^( drift*time + standard deviation*sqrt(time)* random normally distributed numbers)

this is how you simulate potential closing price for the next 16 days....why 16 because the trading month has about 21 days so we...
what beta are you using? are you throwing out outliers?
Any prices quoted are futures
  • #62,650
  • Dec 9, 2012 3:49pm Dec 9, 2012 3:49pm
Quoting redteamgo
what beta are you using? that is a subjective variable with a potentially huge impact
I do a MLE test on my data..... and also corroborate that with a GARCH fit via R
AVT INVENIAM VIAM AVT FACIAM
  • #62,651
  • Dec 9, 2012 4:05pm Dec 9, 2012 4:05pm
Quoting lumesh
well, at least be careful with those...i attached a chart showing that it's not always wise to be contrary to retail...It all depends.

PS. i used to focus a lot on those and had the same mindset as you do. In time, i figured out that a lot of things can be evened out. Trading's not that simple...The book works like everything else...until it doesn't...and then it does again...
Hey lumesh..
Of course, things don't always work out that way & everything we see is subjective.. it all depends on how we use & interpret what we see as all of us have different mindsets & methods to skin the cat..
but one can't expect to make a dime in this business thinking & doing things like the conventional crowd..

"black cat white cat as long as it catches mice it is a good cat"..Deng Xiaoping
http://ih1.redbubble.net/work.185145...ite-cat-v1.png
  • #62,652
  • Dec 9, 2012 4:12pm Dec 9, 2012 4:12pm
Quoting the redlion
ok I will try as simple as I can

when you take the euro returns it resembles a normal distribution correct with heavy tails. usually on a normal distribution , you get the mean and median at 0.

to make price simulation you need to turn the normal distribution curve into a log normal that is done with this formula

price*e^( drift*time + standard deviation*sqrt(time)* random normally distributed numbers)

this is how you simulate potential closing price for the next 16 days....why 16 because the trading month has about 21 days so we...

Thanks for the explain - thats very well thought out and based on solid analysis. Unlike the neural approach. I am sure with your stat backgrnd you already know this - neural nets, in general, use cyclic error propogation between some input data set and desired outcome that needs to be mapped, and try to 'model' a function that will minimize the global error between the desired output and the input. This 'function' can be thought to be stored within the system state after desired number of training iterations are done and/ or global error is at or below the target minimum. The function that is so mapped can often be 'explained' in terms of some statistical paradigm - so I wouldn't be surprised if the similarity of the numbers is not entirely coincidental.

As background info, I do neural net modelling for forecasting and imaging problem domains (been doing since early 90's) - and became interested in applying nnets to forex around 2-3 years back.

The variables I play with are in terms of input data organization and normalization, trying out different neurodynamics, etc.
  • #62,653
  • Dec 9, 2012 4:15pm Dec 9, 2012 4:15pm
Quoting cdndollar
Hey lumesh..
Of course, things don't always work out that way & everything we see is subjective.. it all depends on how we use & interpret what we see as all of us have different mindsets & methods to skin the cat..
but one can't expect to make a dime in this business thinking & doing things like the conventional crowd..

"black cat white cat as long as it catches mice it is a good cat"..Deng Xiaoping
http://ih1.redbubble.net/work.185145...ite-cat-v1.png
Take a look at the Oanda NZD/USD 1 Year Historical Position Ratio.

We are at EXTREME divergence between long positions and market price. This tells me that price could be near a reversal. This is also in confluence with market price being at strong weekly resistance. There were only 5 weekly candles in HISTORY that closed above the level we are currently at.

If you see more room to the upside like you stated yesterday where do you see the actual top for NZD/USD?
  • #62,654
  • Dec 9, 2012 4:19pm Dec 9, 2012 4:19pm
Quoting msood
Thanks for the explain - thats very well thought out and based on solid analysis. Unlike the neural approach. I am sure with your stat backgrnd you already know this - neural nets, in general, use cyclic error propogation between some input data set and desired outcome that needs to be mapped, and try to 'model' a function that will minimize the global error between the desired output and the input. This 'function' can be thought to be stored within the system state after desired number of training iterations are done and/ or global error is at...
the problem with statistics is the use of past data and one MUST NEVER assume that it is predictive in any sense

although my drift weights the current momentum into account... the NUMBER ONE RULE
is that price is moved by orders and hell if I know the magnitude the motive of those orders.

I can ONLY extrapolate ceteris paribus according to statistical data what can I expect. the chart even tick data always lags ipso facto. Because every tick means orders executed eat up liquidity, and the net inclination of these orders is what moves price.....that is reflected in the charts but one NEVER KNOWS what the next tick will be.
AVT INVENIAM VIAM AVT FACIAM
  • #62,655
  • Dec 9, 2012 4:25pm Dec 9, 2012 4:25pm



Quoting pawpaw1000
It's not a dream brother, he already has one :-)

I'm all for dreaming, but one has to be able to admit your mistakes and take criticism...
"Millionaires don't use astrology, billionaires do"
  • #62,656
  • Dec 9, 2012 4:34pm Dec 9, 2012 4:34pm
"Any intelligent fool can make things bigger and more complex.... it takes a touch of genius - and a lot of courage to move in the opposite direction." Albert Einstein
When the Joker is in the deck.. fear not and play it well
  • #62,657
  • Dec 9, 2012 4:41pm Dec 9, 2012 4:41pm
If you cant explain it simply, you don't understand it well enough." Albert Einstein


I'm done now.
When the Joker is in the deck.. fear not and play it well
  • #62,658
  • Dec 9, 2012 4:58pm Dec 9, 2012 4:58pm
Quoting Slim Buffett
If you cant explain it simply, you don't understand it well enough." Albert Einstein


I'm done now.

hahahaha I give too much detail because I am used to corroboration and replication by peers.

here it is simply

standard deviations are are how much the closing price can vary from the average. because of the nature of random like movement things can be below or above the average and generally cancel out, but cluster around the average this is called a standard deviation and 68% of observations are found here around the average.

what I do is I take the current reading and use it to answer this question. all thing being equal what do i expect the average to be in a day, 3 days, a month etc. but due to the way price moves it will not be EXACTLY the average but will vary around it.


so in summary: from where to where can price close in 16 days given that the average will be 1.2877 , and how do I know it will be 1.2877 because if price continues in the same direction at the same rate that is what it will be in 16 days.


are you happy now Slim Buffet lol
AVT INVENIAM VIAM AVT FACIAM
  • #62,659
  • Dec 9, 2012 5:17pm Dec 9, 2012 5:17pm
Quoting the redlion


are you happy now Slim Buffet lol
NO !!

Please refer to....
"Any intelligent fool can make things bigger and more complex.... it takes a touch of genius - and a lot of courage to move in the opposite direction." Albert Einstein

Then I might add...... "You have to learn the rules of the game. And then you have to play better than anyone else" Albert Einstein

OK... I'm just on an Einstein roll for a minute. But really... just establish your trading rules and trade it ????

"The only thing that intereferes with my learning is my education" yup.... Albert Einstein

Cheers
When the Joker is in the deck.. fear not and play it well
  • #62,660
  • Dec 9, 2012 5:32pm Dec 9, 2012 5:32pm
Quoting Slim Buffett
NO !!

Please refer to....
"Any intelligent fool can make things bigger and more complex.... it takes a touch of genius - and a lot of courage to move in the opposite direction." Albert Einstein

Then I might add...... "You have to learn the rules of the game. And then you have to play better than anyone else" Albert Einstein

OK... I'm just on an Einstein roll for a minute. But really... just establish your trading rules and trade it ????...
For My Friend Slim

My middle name is Warren and now to an intelligent article

http://news.goldseek.com/GoldSeek/1355060877.php

Keep Posting...

A golden renaissance?
No wonder gold is enjoying a renaissance as a monetary instrument and store of value.
Over the last forty-two years gold has reacted sharply to every significant change in the oil market. The oil shocks of the 1970s both resulted in huge moves in the gold price that were not capped until significant new supply came on stream in the form of discoveries in Alaska, offshore Gulf of Mexico and the North Sea.
Gold sank for some twenty years until production from those three provinces peaked. Gold bottomed in 2001 and then climbed by some 65% by 2005 when global conventional oil production peaked.
Since peak oil production in mid-2005, gold has risen strongly, rising over 360% in seven years (see chart above).
Given all this, the development of shale oil and gas and of tar deposits in Canada, Venezuela and elsewhere is of great interest.
Were it to be exploitable on a par with say, Saudi Arabia as Saudi Arabia’s oil endowment was fifty years ago, it would be a clear signal that the bull market in gold would be near an end.
This is unlikely, for the simple reason that neither source of fuel is of a quality commensurable with conventional light crude and the impact of both on the market has been over-hyped.

Recoverable shale oil and gas is, after ten years of experience, likely much lower than promised, and it depletes at a rapid rate. A typical well produces 75-80% of its ultimate production in just two years. Unlike conventional oil and gas wells that are drilled and produce for years, this means that drilling has to accelerate exponentially to maintain production in an established production province, never mind increase. It is a massively energy and capital intensive process, on a par with the mining and processing of bitumen for fuel.
Indeed, it appears that the initial boom in the shale oil and gas industry appears to have run its course. Shale oil requires a WTI price of around $80 a barrel to make financial sense for the producers; shale gas requires a price of around $8 per billion cubic feet.

The greater global integration of the oil industry has kept prices high enough, but an intensification of the global recession will change that.

The isolation of the North American gas market has meant that the price there has collapsed with the influx of shale gas, and may put many of the companies that rushed in, out of business.
Peak oil and peak energy means that gold is an attractive diversification for individuals, institutions, corporations and central banks – and will become increasingly so.

It also means that the sharp price fall that gold experienced in 1980 are less likely. Prices may be volatile and there will be sharp corrections but gold is likely to settle at a higher plateau in the coming years rather than a price collapse.
http://67.19.64.18/news/2012/12-9gc/image008.gif
But there is another reason why gold remains attractive, and that is simply that it, too, is a depleting resource and a unique one. The once dominant role of South Africa has declined significantly with gold production having peaked in the 1970’s and falling sharply ever since. Other producers such as China have developed to fill the gap between supply and demand – but there are question marks over some of the figures coming out of China. Despite the increase in Chinese production, world gold production has been slow to respond while the gold price has risen.

This is a physical problem as much as a financial one. The rising energy price of primary energy production translates into much higher energy costs for miners.

They are being forced to extract and separate increasingly hard-to-get and dilute fractions of ore. This raises serious questions about the outlook for their long term profitability. Increasingly, production is a by-product of mining for other minerals from increasingly dilute fractions of gold.
None of this, though, seems to bother the navigators of our ship, who are stubbornly steering us farther from care than danger on a course across an ocean of debt to follow the great white whale of business as usual. These are interesting times alright, and historically interesting times have been the best times to own real, tangible, physical gold.
Somewhere, of course, where the navigators can’t steal it.
Chris Sanders
25 traders viewing now