A basic but HUGE REMARK about interpretation of statistical correlation: statistical correlation DOESN'T IMPLY CAUSAL/EFFECT RELATIONSHIP (this is general) it is just an indicator that can be a fake signal (see for example a statistical lesson here http://deming.eng.clemson.edu/pub/tutorials/qctools/scatm.htm#Interpretations) that is to say, when there is a statistical correlation as it is said in the tutorial above "relationship could be <FONT COLOR=RED><B>caused</B></FONT> by something <FONT COLOR=RED><B>totally different</B></FONT>. For instance, the two variables could be related to a third. Theoretically, if x is controlled, we have a chance of controlling y. " notwithstanding that the correlation must be not only statistically significant but also checked by design experiment of the controlling factor. Since here it is a natural phenomena that is normally difficult. But let's even suppose that it is statistically validated - which is not with only one statistical study - since there is absolutly no reason why the moon cycle by himself would impact the crowd, the reason should come from a supposed third hidden factor I mentioned above. This would be in fact consistent with some supposed statistical correlation with other astronomical phenomenas like solar winds. That is too say this third hidden factor must be common to all these phenomenas that are of different nature. We can exclude that this third factor has anything to do with astrology but rather with physiology of human body. If this third factor can be found and the physical biological effect on human body proved then it would be rational to believe in "moon effect" on stock market but not more in the mystical sense of belief of crowd in moon effect - that is to say a pure psychological fad and autorealisation of this fad - but a true physiological impact. Now this impact itself has also to be proved because it is an other chain in the whole chains of causalities: all the chains must be proved. That is proving something scientifically takes time but this is a necessary process or you would believe in anything that just shows some correlations - that are sometimes even not proved statistically. At the end of the causalities chains you must prove that the physiology of a person has also an impact on market price. So good luck . But thanks this will suggest me a future article on my site: "About the epistemolgy of market's research: example of the moon effect study" where I will develop all the necessary rigors of a scientific approach that every research in finance should apply if they want to be rigorous and avoid charlatanism
Even some astrologers do not in fact believe that human psychology play a crucial role for stock market cycle since I just read this on a financial astrologer site "human nature does not control the market, only the planetary energies do. "
I have enquired and already found the existence of this third hidden factor. I will tell you more soon .
you are insulting your own intelligence i am afraid. this is why you make statistics like a ttest. of course there could be tons of other reasons. but as william ockham said quite some years ago: given all possible solutions the most simple is to be preferred. (at least i think this was the essence). i guess, as i witnessed before, particpiating in a dicussion with you does not necessarily mean that one is talking about the chosen subject ... no offense intended, but i think you are discussing sometimes for the sake of discussion. peace
I'm afraid that it is you that are insulting your own intelligence since you said it "given all possible solutions" for I remind you the basic principle of parametric statistical tests (t test and so like): in such tests you have an hypothesis H0 tested against an other hypothesis H1. But there are two kinds of structures of such tests: one structure is said simple structure and another one is said to be composite structure when the law is unknown that is to say all possible solutions for H1 can't be enumerated. So don't mix this reason with ockham razor's "law". And if you have read the tutorial link given above you would probably learn that this kind of statistical test is far from enough that you have to isolate the causal factors and use other kind of tests which are design experiment tests which are controlled tests which is only possible once you know this factor. Only after that could you elaborate a law and think about using Okam razor. If you are like these journalists that publish "pseudo" scientific papers where everything and their contrary is said at convenience it's up to you but I doubt that the scientific community will consider such test as valid proof since many statistical tests in sociology and economitrics fields have been denied numerous time after they have been published and most of the time because someone finally found the existence of such hidden variables.
1. if you referred to that website, i agree with what you are talking about to some extent. if you refer to those tests i did, the case is different. i do not claim that the results are indicating anything, but to argue that what they indicate is not valid because there could be other hidden variables is poor (i hope you get the intention of this rather strange sentence). 2. without having talked to william intensively recently, i think he meant that his "law" is valid for the set of data at hand. let us draw the most obvious conclusion out of current information. if the market moves significantly different on full moon than on an average day, the most obvious conclusion is that there is a direct relationship. period. does that mean you must not dig deeper? not at all. once you have additional data the same is valid for the new data - draw the most direct conclusion. in the very end you will or will not have proof of whatsoever, but that has nothing to do with occam. and do not forget that occam lived at a time when none of our scientific vocabulary existed. he definitely did not mean multi regression factor analysis - simply because it did not exist. but besides all this there is some other element contained within your post which i would like to adress. reading your post makes me feel like cause and effect, correlation and materia-rational reasoning is the final judge in the world of science. i would say this has been contrasting with quantum physics. what we call cause and effect does not seem completely applicable to this field. at least for my decent understanding it seems that some experiments indicate that our world is different from what it seems to be. btw yes i am probably much more on the level of a journalist than yourself. but i have the feeling that this has advantages too ... peace
Strange sentence ? sorry this is not an invention from me, this a well known classical flaw that every professional statistician is aware of especially for psychology, sociology and economical fields, and specifically the expression "hidden variable" is the OFFICIAL TERM used in statistical field. These hidden variables rarely appear in physical field since the laws are generally known or at least deterministic enough whereas they are predominant in human activities fields. The rest of what you answered is just the illustration that you ignore the potential existence of these hidden variables. If these hidden variables were evident, be assured that they wouldn't be invented and that so many statistical studies in the human fields although conducted as seriously as they could be as for their data collection and hypothesis tests wouldn't be invalidated as for their conclusion. It is true that at school, teachers wil rarely learn you about them since they only give the basics and illustrate them in physical field. As for myself I first learn and experience them in food industries where we were accustomed to do psychological tests for assessing (or rather trying to assess) consumers preferences. At that time, for professional reason, I used to read a few books about psychological statistical tests and there were illustrating the difficulty of using statistical tests in such field and give many cases of flaw due to the reason above and warned that any conclusion shouldn't be drawn without multiple checks and wait for possible invalidation by someone. That's why I pretend to say what I said.