Best Thread Debunking Fibonacci's Code...

Saw this research paper commented on in my newspaper's business pages today, which used it to stick the boot into TA of all descriptions. Basically concludes that Fibonnaci type retracements occur no more frequently than would be expected by chance.

(by the way, the thread title comes from the University's press release announcing the publication of the research)

http://www.cass.city.ac.uk/media/stories/resources/Magic_Numbers_in_the_Dow.pdf

thats pretty interesting.

confirms my view that if you draw enough lines on a chart, some of them are going to line up with something,

Jack o'Clubs said:

Saw this research paper commented on in my newspaper's business pages today, which used it to stick the boot into TA of all descriptions. Basically concludes that Fibonnaci type retracements occur no more frequently than would be expected by chance.

(by the way, the thread title comes from the University's press release announcing the publication of the research)

http://www.cass.city.ac.uk/media/stories/resources/Magic_Numbers_in_the_Dow.pdf

Click to expand...

a typical academic research using static prognosis in a dynamic enviroment

Andy

i should have mentioned their study of Fibonacci is actually accurate fib numbers do not always appear if one looks carefully you will see that many ratios derive from the square and the circle so you wont get 0.786 but 0.707
0.786 derives from square root of 0.618 whist 0.707 is the square root of 1.414 which in turn is the square root of 2 (sacred geometry)
point being time and price is working via a muti-dimensional plane e.g
in a moment of time you may see the face of a square if you have 2 maybe 3 points mapped within the square your odds of forecasting the next point ie a retracement or an extension is vastly increased

hence what they should have looked into is if its not fib what is it and back track to how it relates to the square and the circle


Andy

andycan said:

hence what they should have looked into is if its not fib what is it and back track to how it relates to the square and the circle
Andy

Click to expand...

Why ? Why not a triangle, parabola or a hyberbola or more generally an nth order polynomial, or any other geometric shape or function you care to name.

And by the way, 0.707 is not the square root of 1.414.

Sorry I don't believe a word of it.

I havn't read the paper, but what I'd much rather see is an analysis of market profile levels, VWAP and pivots whose basis is eaily understood without appeals to numerology. And I'd like to see it done over several markets - especially some of the asian markets. Just choosing the dow is a bit limiting considering the interrelation of the major US indices.

yep sorry inverse of 1.414 = 0.707
oh yes the triangle plays a major part as for the other shapes simply after thousands of hours analysing measuring and calculating this is my conclusion
the circle and square is at its most simplistic from them you can derive at all other shape related ratios so why complicate it life
as for believing i have no interest in converting you im expressing an opinion which in this particular subject i know what im talking about
certainly the theoretical and how i apply it to my trading trading


Andy

square circle and triangle

From a circle, square and triangle if you play around a little you get the following numbers:-

1.414 (sqr2)
1.618 (phi or the fib number)
1.272 (sqr phi)

and their inverses
0.618
0.707
0.786

These together with 0.5 (1/2) and 0.382 (1-0.618) do seem to recur a large amount in the mkts esp if one allows a few percent tolerance either way.

Reading the economist this week (page 89) they quote a study which disproves the validity of fib numbers for market trading but again how are these studies done?)

Personally suspect that success has far more to do with psychology (sticking with your trading plan and not taking emotional decisions) and this is very hard.

I'm studying using fib ratios per above based around certain patterns to give an "edge", but fully believe that I and others using this will only suceed if we get the application right

cheers
P

Coincidence or causal relationship

Paddington said:

From a circle, square and triangle if you play around a little you get the following numbers:-

1.414 (sqr2)
1.618 (phi or the fib number)
1.272 (sqr phi)

and their inverses
0.618
0.707
0.786

These together with 0.5 (1/2) and 0.382 (1-0.618) do seem to recur a large amount in the mkts esp if one allows a few percent tolerance either way.

Reading the economist this week (page 89) they quote a study which disproves the validity of fib numbers for market trading but again how are these studies done?)

Personally suspect that success has far more to do with psychology (sticking with your trading plan and not taking emotional decisions) and this is very hard.

I'm studying using fib ratios per above based around certain patterns to give an "edge", but fully believe that I and others using this will only suceed if we get the application right

cheers
P

Click to expand...

Paddington

They do appear to be prevalent, but this does not mean that there is any causal relationship.

The table below shows the numbers you quote. You also mentioned a tolerance of a few percent, lets say 2%.

• 98%

• 100%

• 102%

• 0.37436

• 0.382

• 0.38964

• 0.49

• 0.5

• 0.51

• 0.60564

• 0.618

• 0.63036

• 0.69286

• 0.707

• 0.72114

• 0.77028

• 0.786

• 0.80172

• 1.24656

• 1.272

• 1.29744

• 1.38572

• 1.414

• 1.44228

• 1.58564

• 1.618

• 1.65036

We now have a fairly large set of ratios with which to curve-fit. Furthermore attempting to do this on a chart is a somewhat hit-and-miss affair. A price bar or candle has a length from the high to the low, with opening and closing prices also. Which ones does one attempt to correlate the approximate fib levels with ? Also from what starting point do you apply the ratios ?


It is all too easy to work around a chart trying to find a suitable start and end point from which to derive the levels and also a temptation to adjust by a few percent. Coincidence or genuine causal relationship ?

Charlton

About a year or so ago, I tried to see if Fibs had any significance, but sadly, couldnt.

I started with the proposition:
a: what is the highest point in the month of January 1998.
b; what is the lowest point in January 1998.
c: what are the major Fib extensions/retracments from Jan-High.
d: what are the major Fib extensions/retracements from Jan-Low.
e: are any lines met, (Highs or Lows) in the next 90 days (Feb/Mar/Apr) that were swing-highs or swing-lows? The purpose of selecting swing-highs/lows was so that the hit had to identify a true local high or low (potential turning point), rather than a hit in a mass of other bars. Swing High = a bar where the preceding and subsequent day did not breach the high. (and vice-versa for swing-lows)
e1: with a 5 pip tolerance
e2: with a 10 pip tolerance
e3: with a 20 pip tolerance
Repeat, this time taking Feb High/Low looking at Mar/Apr/May; then March High/Low, etc.

NB:
01: also made allowances for pip tolerance, such that, I was wary of <5 pip hits when monthly range was narrow.
02: had to restrict checks to next 90 days, as the lines were everywhere!!!
03: also did same for Quarterly High/Low, looking forward 3 quarters.
04: also did same for Yearly High/Low fib extensions/retracements, looking 3 years ahead.
05: Rather than try to "find" patterns and thus "hits", which the human eye is very good at, even when there may be no pattern, I decided to use maths to establish how many hits there were, thus attempting to rule my preconceptions out of the equation.

I try to be methodical and analytical as possible, but found no significance.
I am not a trained scientist or mathematician.

.

.

Actually, the "real point" is that there is no proof that Fib levels "work" for anybody without a controlled study, including control of the variables. As with so many quick fixes of this kind, the individual who is "successful" with it believes that it is the approach that works when it might be something else entirely, as with, for example, trendlines.

The conflict here has to do with belief and faith on the one hand and science on the other. As long as one believes that Fib "works" (or ADX, or stochastics, or RSI, or MACD, or CCI, or Elliott, or Gann, or candles, or pivot points, or whatever), then any discussion centered on "proof" is doomed. Which is why these arguments/discussions regarding Fib/etc in particular or technical analysis in general go on and on.

Paddington glad to see an aspiring geometer
dont forget square root 3 and 5. they are critical in a harmonic markets
the problem with academics trying to prove or disprove fib is simply because they are exploring fib in the wrong way everyone is too busy trying to find a retracement related to fib or an extension and forecasting,
most academics have no idea of market structure so their thesis is based on an up move and a down move then look at another down move and an up move and so on and try to correlate it.this is where they go wrong a market will have a structure more often than not they measure the substructure which is incorrect. in the pdf file from Jack O Clubs they show a chart with their ideas and its completely incorrect thats when i stopped reading any further.
also to point out there are ratios which are similar to other ratios for example its not uncommon for these targets to be a tad shy or for it to slightly overshoot, and the dilemma is what is it is it 0.786 or is it 0.764, 0.7236 or 0.75 and so on and to the untrained eye this can be daunting, the clue is when ever there is a measure you have to find the common denominator (ratios relating to the appropriate geometric shape) now this can be hidden and some times it is crystal clear. what this means is one has to determine if the market is working via the square ie root 2 or the triangle root3 and the harder one to find root5, but it does not stop there most make the mistake of just looking at the harmonic numbers there is also sub harmonic that can get a little messy. With root 2 most will use 0.707 or 1.414 but 0.841 which is the sq root of 0.707 crops up a lot 0.354 is out there too most who see it would related it to 0.382 but its not 0.354 + 0.354 =0.707 these are the hidden ratios most ignore or assume it to be something else.thats is why many can make it work.
this maybe difficult for some to grasp but believe me if you do the work you will start to see it. but most importantly geometry is a mathematical representation of the human psyche, the masses will do the same thing over and over and over again hence why these ratios always crop up


Andy

I've been using fibos since I first read ‘Elliott Wave Principle : Key to Market Behavior’ by Frost and Prechter in 1981, fibos work, consistently, over and over again and again.

Fibos are The simplest most accurate price 'targeting' method available.

As Jimbo57 points out, their use is tied to EW, so one 'has to' have an understanding of PriceTime Movement in terms of Waves.

The application of the fibo tool on price bars varies, just as any workman/artist varies the tool being used in order to produce a result, so too 'with experience' the fibo user will draw the fibo from for instance the HC/LC rather than HH/LL (or draw both) on long timeframes when drawing a Retracement fibo.

On all Projection fibos I use the L-H/H-L of the 0-1 first Wave of the move.
Levels used: 1.146 1.236 1.382 1.618 1.764 1.854 2.00 and up.
I may 'step' (copy paste) the fibo onto the LC/HC of the W2/3 end/start, and may re-set a new H/L of the original fibo as the move progresses.
All my charts have an MA S-1-C applied since the C rather than HL is often significant when interacting with fibo levels.
The application of linked parallel lines (MetaStock's Standard Error Channel for me, others may prefer Andrews Pitchfork etc) provide an 'envelope' where the intersection of SEC/fibo lines and Price usually signify a trend change.

I've never found fibos applied to Time to be as accurate for me as Price fibos so use 'Bar Counts' for Time targets.

Trust the attached FTSE Weekly chart proves my points.
.

I thought the paper read like a fairly average Masters thesis, where the writer is trying to demonstrate to his tutor a knowledge of whatever statisical analysis he needs in order to graduate in his subject. The fact that this happened to be an analysis of Fib levels in the Dow was probably largely irrelevant to the project until the professor guiding it realised that with a quick top-and-tailing from him, and a snazzy and a knowingly 'controversial' press release from the university it would get another publication to his name and a few column inches in the press. I may be wrong, but this seems to be more about self-serving academics than any definitive proof of the rights or wrongs of Fib / Elliot analysis.

One really good way to debunk or not to debunk, would be to post an analysis, and see how it pans out over the coming days and weeks.

Shall we start with me giving you all a laugh and post my Elliott Wave counts ?

EDIT: brain melt-down!! Having read Elliott Wave Principle, and Fibonacci Trader by Fischer, I now think of Fibs in relation to Elliott, as per jimbo57s post.
This thread is about debunking Fibs, not about EW.!! Sorry.

Jack o'Clubs
you are absolutely right!! in the real world one tries to find solutions not create more questions i guess thats easier said than done

Trendie wave structure has its place in geometry, if geometry is a mathematical representation of human emotion then it goes without saying wave structure will come into it. i have never studied Elliott Wave so i cant say how relevant it would be but my interpretation of wave structure works for me

Andy

A good discussion going here! Thanks for the welcome Andycan. Never historically thought of applying geometry to markets, but have been doing so alot over recent months. Absolutely fib is only one method alongside many others, and the proof or otherwise of its relevance is whether one can determine a set of trading rules based around it which generate a consistent profit of course! I expect to take some months to discover this.

I've recently been mentoring with a fib tutor which explains my interest and I'm fairly impressed by their approach. Alot of what I'm learning though is about the psychology of trading which is probably even more relevant - what seperates those who succeed from those who don't.

I certainly know people who use fib techniques successfully. The approach these guys take (you can read at tradingtutor.com) relies on pattern recognition around fibonacci ratios, specifically patterns called gartleys and butterflies and to a lesser extend variants of AB=CD. These patterns have more geometry in them than just the one fib ratio, and as an extension of that I'm looking to see where I can find multiple fib ratios at a point and seeing whether those points give higher probability trades. Will take some time to discover though.

The other key elements they use are (1) warning signs such as gaps/ long range bars which lead to ABCD extensions, (2) waiting for a completion then confirmation candles and (3) allowing some wriggle room to allow the trade to work. For example if trading a 0.707 pattern are you able to set your stop so that it also works for a 0.786 pattern if that develops within your risk management plan? I.e. risk management and money management techniques as we know count for alot.

If I find out how to post charts I'll post some that I've been looking at on the FTSE 60 minute charts which appear to have alot of fib ratios on over the last few months (not all exact but close enough to trade).

And absolutely I'm aware that there are lots more numbers as you say sqr(0.707) = 0.841 which comes up a bit as does 0.886 (sqr (0.786). I've yet to see these sufficiently often.

Two that I'm watching which people may find interesting are daily chart patterns on

1) AXP (american express) - taking the 2nd June high as X and 24th July low as A it has just completed a bear butterfly at the 1.272 extension on Tuesday. Yesterday confirmed so today would be a short entry with the stop just above Tuesday's high.

2) BBY (better buy) has just completing a bearish gartley and confirmation day at the 0.786 retrace to the high of 7th April.

Again if I learn how I'd be happy to post and follow up to see how these did. I understand that these patterns have an approx 70% probability but have yet to get enough history of my own to have a belief of my own around

cheers
P

Paddington said:

.....................If I find out how to post charts............................. P

Click to expand...

paddington

when you write a post scroll down to "manage attachments" (you'll need to "go advanced" if you're using quick reply) which shows the valid file file extensions. You need to save your chart as one of those type of files - jpeg works well - and then upload. You may need to put your chart into paint first and save it from there if you're working directly from the web.

good trading

jon