Crude light future prob 99.5% bearish today

[science_trader]

Quote from seleukos:

We have a list of patterns.
In the last 400 days, when a certain configuration has happened in crude light (it happedn 8 times about) for height times (always in this reality) the next day crude light went down.
So we say that it 'always' goes down.
Obviously, as written elsewhere, we are waiting that opening is in a correct range (64.08 | 62.08) and this not in the true opening, but exactly after 40 minutes (10:40 am New York time)
moreover, stop loss is in place.

Nevertheless in the last 400 days, it always went down. Surely it is a sample, but the problem is: why not more than 400 days?

Really, with a Chi square analysis, you know that you cannot go on the past too much, otherwise the Chi square refuses your sample.

So last 400 days are somewhat a population and not a sample: but the boundary is not clear.

We prefer to operate in instruments that in the last 400 days always reacted in a certain manner to a pattern.
In every case we are not sure (crystal ball...) We thing that in this forum every one should present ideas: terminology is not clear in statistics, because the Great Numbers Law is not yet clear. (We are not english so probably Great Numbers Law is told differently)

Which is exactly what I say : your 99.5% statement has no statistical significance...

 

[empee]

clearly, this is being fooled by randomness

for the other guy who questions whether i took the trades,

http://www.elitetrader.com/vb/showthread.php?threadid=63802

 

[seleukos]

Quote from science_trader:

Which is exactly what I say : your 99.5% statement has no statistical significance...

We are facing a pattern that in the last 400 days always went down. For 8 times. 400 days is somewhat the population and not a sample.
You cannot infer in financial markets using normal distribution. If so (read Mandelbrot) October 1987 never should have occurred.

If instead of 8 patterns all bearish they were 4, we had 50% probabilities that today it will go down. In this context, obviously.
Good trading.

 

[Ebo]

Who is "We", do you have a turd in your pocket?

 

[science_trader]

Quote from seleukos:

We are facing a pattern that in the last 400 days always went down. For 8 times. 400 days is somewhat the population and not a sample.
You cannot infer in financial markets using normal distribution. If so (read Mandelbrot) October 1987 never should have occurred.

If instead of 8 patterns all bearish they were 4, we had 50% probabilities that today it will go down. In this context, obviously.
Good trading.

If I understand right, I used normal distribution in my reasoning and Mandelbrot is the king of fat tails...it says it all.

 

[Buy1Sell2]

Quote from empee:

clearly, this is being fooled by randomness

for the other guy who questions whether i took the trades,

http://www.elitetrader.com/vb/showthread.php?threadid=63802

Just to clarify, this was not directed at you. I have not had time to look and see if you have been posting in realtme.

 

[seleukos]

Quote from science_trader:

If I understand right, I used normal distribution in my reasoning and Mandelbrot is the king of fat tails...it says it all.

Please do not use normal distribution. Fat tails are a reality.
We are in finance from 1965 and in our experience, when using normal distribution, problems were on the corner.

Our approaches now tend to use shorter and shorter periods where an instrument should go up if it is 'in debt' of bull days and so on.
Moreover: possibly do not go overnight, enter the market half an hour after opening and exit the market half an hour before the end.
This is what we think. We think that a Forum is a place where everyone offers some idea, perhaps not correct, obviously.

 

[science_trader]

Quote from seleukos:

Please do not use normal distribution. Fat tails are a reality.
We are in finance from 1965 and in our experience, when using normal distribution, problems were on the corner.

Our approaches now tend to use shorter and shorter periods where an instrument should go up if it is 'in debt' of bull days and so on.
Moreover: possibly do not go overnight, enter the market half an hour after opening and exit the market half an hour before the end.
This is what we think. We think that a Forum is a place where everyone offers some idea, perhaps not correct, obviously.

OK, I say it clearly : YOU HAVE NOT A CLUE IN BASIC STATISTICS. Please stop using words you don't even understand, even 'normal distribution'.

 

[seleukos]

Quote from science_trader:

OK, I say it clearly : YOU HAVE NOT A CLUE IN BASIC STATISTICS. Please stop using words you don't even understand, even 'normal distribution'.

October 1987 and about twenty other situations are not comtemplated by Normal Gaussian Distribution: they are completely out and unforecastable. Nevertheless, shall we use a model that is incongruent with the happening? Why?
Please explain.

 

[science_trader]

Quote from seleukos:

October 1987 and about twenty other situations are not comtemplated by Normal Gaussian Distribution: they are completely out and unforecastable. Nevertheless, shall we use a model that is incongruent with the happening? Why?
Please explain.

Where did I use normal distribution in my reasoning ?

What you said in your first post is exactly the same as if I say the following : "Everytime today's temperature is over yesterday's temperature, I have eaten pasta for lunch, I got up at 7AM and did not have a shag before leaving to work, then market is going down with a 99.5% certainty."

However, I won't loose my day trying to explain basic probs and stats to you. Answers from others in this thread do confirm what I told you.