Why can Fibonacci ratio become the Support or Resistance?

[valentinestock]

Hi, I try to find out "why Fibonacci ratio can become the Support or Resistance" in our forum, as links below
Trading with Fibonacci
http://www.trade2win.com/boards/training-education/16387-why-successful-traders-use-fibonacci-golden-ratio.html

Here is the link which says that "reasons that are unclear"
What is Fibonacci retracement, and where do the ratios that are used come from?

They dont mention the reasons why Fibonacci ratio works well. Is there any scientific proof or explainations by behavioral finance?

Thank in advance!

 

[Liquid validity]

http://www.trade2win.com/boards/analysis/21333-debunking-fibonaccis-code.html

http://www.trade2win.com/boards/analysis/165852-useful-fibonacci-levels-3.html

A Leveraged Trading Strategy with Fibonacci Approach: Empirical Evidence for DJI 30 and ISE 100 by Oral Erdogan, Taner Doguc :: SSRN

http://www.trade2win.com/boards/analysis/165852-useful-fibonacci-levels-3.html#post2066538

 

[cablemonster]

They are just levels that a lot of participants look at and therefore there is likely to be a fight between buyers and sellers there. It might be a big or small fight. As with any level it may or may not hold.

Hi, I try to find out "why Fibonacci ratio can become the Support or Resistance" in our forum, as links below
Trading with Fibonacci
http://www.trade2win.com/boards/training-education/16387-why-successful-traders-use-fibonacci-golden-ratio.html

Here is the link which says that "reasons that are unclear"
What is Fibonacci retracement, and where do the ratios that are used come from?

They dont mention the reasons why Fibonacci ratio works well. Is there any scientific proof or explainations by behavioral finance?

Thank in advance!

 

[valentinestock]

Thank Liquid validity. I dont find the right answer for my question.

Cablemonster. Yup, I agree that there are some levels that buyers and sellers have a fight on price. But I want to know why they fight at 0.618, 0.382... Why they always fight at those numbers. How about 0.617, 0.616, 0.615... because they are not the lucky numbers?

They are just levels that a lot of participants look at and therefore there is likely to be a fight between buyers and sellers there. It might be a big or small fight. As with any level it may or may not hold.

 

[chaselancaster]

In answer;

1. Buyers and sellers don't always 'fight' at fib levels-whatever that level is...as often as not if not more times than not - price ignores them, or at least that is how it seems.

2. Why not the other numbers you mention? - because they are not what the market generally recognises as fib retracement ratios - these are 23.6%, 38.2%, 50% (although this one not strictly a fib ratio,) 61.8%, and either 78.6% or 76.4% I have seen some also use 85.4% and 14.6% but the ratios 23.6% - 61.8% inclusive are I would say the main ones.

In general - fibs are to a large extent subjective in that - yes there are obvious swings on any given t/f but less obvious ones too where some market participants draw fibs from.

Thank Liquid validity. I dont find the right answer for my question.

Cablemonster. Yup, I agree that there are some levels that buyers and sellers have a fight on price. But I want to know why they fight at 0.618, 0.382... Why they always fight at those numbers. How about 0.617, 0.616, 0.615... because they are not the lucky numbers?

 

[Shakone]

The only level I've found any use for is the one that has nothing to do with Fibonacci, the 50%.

 

[Liquid validity]

Thank Liquid validity. I dont find the right answer for my question.

In what way?
Not what you wanted to hear?
At best most TA is a supporting crutch to reading price action or the tape.
In isolation, most of it falls apart at the seams.

 

[Traderallen]

The ratios work only because several years ago someone starting price patterns noticed that occasionally prices would reverse at these ratios. They then proceeded to write a book or paper on and publish it to the masses. It is now a self-fulfilling prophecy they are robots programmed to look for these retracements, also individual traders are looking for these retracements there is no magic behind the numbers it simply this. Price goes up trader start taking profits price pulls back when it reaches a fib level robots kick in and start to buy price stalls individual traders kick in and start to buy more robots start to buy as price reverses a heads up swing traders jump in they help ride price of as price goes up momentum traders jump in they help drive the price up. And the pattern repeats itself again and again. It often fails when price reaches the levels and as the robots kick in there's still too much selling pressure causing the levels to fail as they fail breakout traders jump in adding more selling pressure then the swing traders and momentum traders follow suit and there again the pattern repeats itself. So basically you end up with a 50-50 chance of a fib level turning the market. The key to making money would be recognizing order flow early and jumping in quickly and out quickly.

 

[neil]

The only level I've found any use for is the one that has nothing to do with Fibonacci, the 50%.

An interesting level used by many

 

[Purple Brain]

I'm sorry I can't remember the source, but a paper written a few years back basically said if you were extremely retrospectively artful in selecting your high and low points, you could show that fib levels were statistically valid. But even then, you wouldn't know which of the levels was was going to be a valid support or resistance level.

It then went on to show that a genuinely random application from genuine - rather than artfully selected - highs & lows showed a negative correlation against all fib levels. That is, worse than random chance.

 

[valentinestock]

Thank chaselancaster, Shakone, Liquid validity for your answers

I'm sorry I can't remember the source, but a paper written a few years back basically said if you were extremely retrospectively artful in selecting your high and low points, you could show that fib levels were statistically valid. But even then, you wouldn't know which of the levels was was going to be a valid support or resistance level.

It then went on to show that a genuinely random application from genuine - rather than artfully selected - highs & lows showed a negative correlation against all fib levels. That is, worse than random chance.

Thank Traderallen . Your reason also makes Fibo value happens more often than the usual

Thank Purple Brain :smart:, this is what I want. I found some similar things your answer at the link below.
Each Fibo value has its p value. The smaller p value, the more significant in statistics the Fibo value has. So we expect these levels occur more often than random chance. It will be perfect if they explain more about the meaning for their result.
http://www.socionomics.net/pdf/Fibo_Statistics.pdf

 

[Liquid validity]

Each Fibo value has its p value. The smaller p value, the more significant in statistics the Fibo value has. So we expect these levels occur more often than random chance. It will be perfect if they explain more about the meaning for their result.
http://www.socionomics.net/pdf/Fibo_Statistics.pdf

I'd read that paper again if I were you, the sample sizes used in that paper are so low they are irrelevant:

For Fibonacci ratios less than 1, “bullish” and “bearish” retracements measured on log scale produced
ratios of 0.786, 0.618 and 0.382 more often than would be expected on average in a random environment
(hereafter called “positive outcome”). Two of them are significant at the 1% level and one at the 5% level.
So half of these six Fibonacci occurrence rates are significant, despite limited data, and all six have positive
outcomes. Finding positive outcomes six out of six times itself is statistically significant at a level of 1.6%.
Finding significant outcomes three out of six times is also statistically significant, at the 2% level. Projec-
tions also produced positive outcomes in five out of six tests, a result that is significant at the 11% level. So,
price trends on log scale produced positive outcomes in 11 out of 12 tests for Fibonacci ratios less than 1.
This amount is highly statistically significant (p =0.003).
For Fibonacci ratios greater than 1, price trends on
log scale produced positive outcomes in 10 out of 16 tests (p= 0.2).

In all, price trends on log scale produced positive outcomes in 21 out of 28 independent tests,
which is a highly significant result (p= 0.006).
Price trends on arithmetic scale produced positive outcomes in 18 out of 27 independent tests,
3 a result significant at the 6% level.
Time lengths and percentage price moves also produced positive outcomes in more than half
the tests: 16 times out of 28 and 16 times out of 27, respectively.

On the whole, 71 out of 110 Fibonacci occurrence rates exceed the 50th percentile,
a number greater than the 55 expected by random chance.
Similarly, 22 out of 112, more than the 11 expected, achieve the 90th
percentile; 19 out of 112, more than the 6 expected, achieve the 95th
percentile; and fully 16 out of 112, many more than the 1 expected, achieve the 99th
percentile. If these results could be assumed independent,
the observed amounts would be highly significant for each of these percentiles:
50th (p =0.001),
90th (p =0.003).

How anyone can possibly think those sample sizes are large enough to account for
variance and deviation beats me...
Basically, its completely worthless.

 

[NVP]

Hi, I try to find out "why Fibonacci ratio can become the Support or Resistance" in our forum, as links below
Trading with Fibonacci
http://www.trade2win.com/boards/training-education/16387-why-successful-traders-use-fibonacci-golden-ratio.html

Here is the link which says that "reasons that are unclear"
What is Fibonacci retracement, and where do the ratios that are used come from?

They dont mention the reasons why Fibonacci ratio works well. Is there any scientific proof or explainations by behavioral finance?

Thank in advance!

no ..........if you load enough possible numbers in a set they are bound to be similar to reality........

N

 

[Shakone]

Yeah that socionomics paper isn't very convincing.

Back on the original question, if you define support and resistance as areas where price has been 'turned around', and may do again, then it doesn't really make sense to say that any fib retracement can be support or resistance, since wherever the fib is taken, price must have moved through all retracements in between.

 

[valentinestock]

Yeah that socionomics paper isn't very convincing.
Back on the original question, if you define support and resistance as areas where price has been 'turned around', and may do again, then it doesn't really make sense to say that any fib retracement can be support or resistance, since wherever the fib is taken, price must have moved through all retracements in between.

Yes. No all Fib retracements can be support and resistance. Each value has different p value. Usually, Fib retracements with p value < 0.005 are considered important one and they will appear more often (than random chance).

I'd read that paper again if I were you, the sample sizes used in that paper are so low they are irrelevant:
How anyone can possibly think those sample sizes are large enough to account for
variance and deviation beats me...
Basically, its completely worthless.

Again, thank Liquid validity for your answer. This paper doesn't convince me but it shows the right direction for my problem
Assume we do stats on a large sample. Who guarantee it always true for testing data.
Another case, a stats on SP500 and on NASDAQ may also show different p values for Fibo retracement. This means support and resistance of SP500 can't use in NASDAQ
These Fibo values are popular numbers so that it is true for many securities, everyone can easily remember and apply.

 

[Liquid validity]

Yes. No all Fib retracements can be support and resistance. Each value has different p value. Usually, Fib retracements with p value < 0.005 are considered important one and they will appear more often (than random chance).



Again, thank Liquid validity for your answer. This paper doesn't convince me but it shows the right direction for my problem
Assume we do stats on a large sample. Who guarantee it always true for testing data.
Another case, a stats on SP500 and on NASDAQ may also show different p values for Fibo retracement. This means support and resistance of SP500 can't use in NASDAQ
These Fibo values are popular numbers so that it is true for many securities, everyone can easily remember and apply.

If you are going to use them, I'd suggest reading through some of bbmac's stuff:
http://www.trade2win.com/boards/members/2135-bbmac.html
http://www.trade2win.com/boards/general-trading-chat/137020-fibs.html

TBH, in isolation I don't see any statistical value.
You may however find a way of incorporating them into
a broader plan.
I don't use them as I find the logic patchy at best.
Others do use them, so its up to you really

 

[valentinestock]

If you are going to use them, I'd suggest reading through some of bbmac's stuff:
http://www.trade2win.com/boards/members/2135-bbmac.html
http://www.trade2win.com/boards/general-trading-chat/137020-fibs.html

TBH, in isolation I don't see any statistical value.
You may however find a way of incorporating them into
a broader plan.
I don't use them as I find the logic patchy at best.
Others do use them, so its up to you really

Maybe, I'll include it in my system. However, I need more time to understand and test strategies . Hope it works!