Fibonacci - How does it work - Does it really work?

I believe that as all other support/resistance levels, Fibonaci levels are points where something CAN happen. But, as there is no Holy Grail, it is only a CAN, in no way a certainty. Now if enough people believe something will happen at that level, it will actually happen.
Many times, those kinds of mystical numbers (mentioned in the Da Vinci Code) are used by trading gurus to look smart. Some Elliot Wave followers will refer to the Kondratieff winter and the Gann followers to some other esoteric concepts. They sound so mcuh smarter than the rest of us !
All in all, all those concepts should be only used, imo, with robust money management tools, namely, stop losses, when they do not work, which happens (!)
 
It's interesting how sometimes 'previous/prior' Projection fibo levels appear to become support/resistance lines when the price is travelling in the opposite direction, as well as when Retracement/Correction fibos are used — clustering.
 

Attachments

Quote from A Dashing Blade:

By using your coin-toss example as a critique of Fibonaci, you are assuming that markets are normally distributed.

They are not (although, to paraphrase Orwell, some markets are more normal than others).
More...

Excuse me: they are NOT normally distributed, because a computer immediately sees that the spectrum is uniform.
I will say that TA is not a good instrument and also the Fibonacci rules are leading you to nowhere.

In the picture, you should see a straight line or something so. What we see is the 'underground rumor': this is instead normally distributed.

Really, on the picture, there was no market at all: only random data.
 
Quote from seleukos:

Excuse me: they are NOT normally distributed, because a computer immediately sees that the spectrum is uniform.
I will say that TA is not a good instrument and also the Fibonacci rules are leading you to nowhere.

In the picture, you should see a straight line or something so. What we see is the 'underground rumor': this is instead normally distributed.

Really, on the picture, there was no market at all: only random data.
More...


We are the underground--We know who you are and what your trading strategies are--We are watching.
 
I predict my lunch today will be at 50% retracement of the previous day's range. This seems to be the natural order of things and can be easily forecast.
 
Quote from seleukos:
Excuse me: they are NOT normally distributed . . .
More...
er, that's what I said :confused:

Quote from seleukos:
In the picture, you should see a straight line or something so.
More...
I think you mean that on average you should see a straight line.

I'm not disputing your assertion about Fib analysis, I'm pointing out a flaw in your argument ie that of applying normal distribution theory (coin tossing) to a non-normal environment (financial markets)

.
 
Quote from A Dashing Blade:

er, that's what I said :confused:


I think you mean that on average you should see a straight line.

I'm not disputing your assertion about Fib analysis, I'm pointing out a flaw in your argument ie that of applying normal distribution theory (coin tossing) to a non-normal environment (financial markets)

.
More...

Coin tossing demonstrates that also in markets going to nowhere you can discover head and shoulders and so on. This implies that the matter is more difficult than we can think (see the last work of Mandelbrot).
When sentencing an head and shoulder, it could be meaningless at all. So, it is meaningless to discuss about it.
Markets are not normally distributed and Head and Shoulder are not really pointing somewhere.
If markets were normally distributed, on the last two centuries we should have seen one or two daily movements grater than 5%. October 1987 had 0 probabilites about to happen in a gaussian distribution. Only one thing is about certain: in the long, long, long run markets are going up.
But maximum for aluminium was in 1895 about.
 
Quote from seleukos:

Here we attach a random chart.
More...

That was a very interesting chart, but I think I found a problem with it. Its scale is extremely stretched to the movement (vertical) compared to the time (horizontal) line.

If you make a chart where the scale is 1:1, so for 1 unit of time you make 1 unit of movement in a direction you will see that the very little difference between heads and tails makes very little difference in moving away from the zero line. So after 1000 unit of time occurs the heads would be only 26 units higher from the zero points which is only 2.6 % of the whole chart. So no big rally occured. It is like saying the Dow is pretty much exactly (well 2.6 % away) where it was 4 years ago...

That is pretty much an untradable movement in such a timeframe...
 
Quote from Pekelo:

That was a very interesting chart, but I think I found a problem with it. Its scale is extremely stretched to the movement (vertical) compared to the time (horizontal) line.

If you make a chart where the scale is 1:1, so for 1 unit of time you make 1 unit of movement in a direction you will see that the very little difference between heads and tails makes very little difference in moving away from the zero line. So after 1000 unit of time occurs the heads would be only 26 units higher from the zero points which is only 2.6 % of the whole chart. So no big rally occured. It is like saying the Dow is pretty much exactly (well 2.6 % away) where it was 4 years ago...

That is pretty much an untradable movement in such a timeframe...
More...

Well: again, we will demonstrate that the random charts creates some patterns that have no reason.
For the dimension of the vertical measurement, the computer arranges the scale so that the top is at the maximum and viceversa.
 
Quote from seleukos:

the random charts creates some patterns that have no reason.
More...

I think what you are saying is equivalent of putting a piece of grass under a magnifying glass and discovering chartpatterns on it.

In a few hours I will post a chart with normal 1:1 scale. You will see that there won't be any discoverable chartpatterns on it....
 
You must log in or register to reply here.
Back
Top