Suggest a topic for research - page 5

New comment
 
C-4:

Just understand one thing, the purpose of any research is to validate ideas.

And often the most idiotic ones at that. Most of today's so-called "scientific discoveries" are made by epigones using statistical packages.

The scheme is simple enough, namely, something needs to do a pairwise comparison of the two methods on the number of favorable and unfavorable outcomes.

For example: "Pairwise comparisons of TS effectiveness with (specify particular TA) and the strategy of market entry directions selection depending on the parity of the deal number".

  1. Let's take some TS with some TA specified in the topic with fixed take and loss.
  2. We make a copy of the first TS and remove TA entry from it. Instead of it we use the direction of the trade depending on whether the trade is even or odd: int x = (OrdersTotal() % 2), if x == 0, then it is long, x == 1, then it is short. Losses and takes are the same as in the first TS. Thus, we obtain the second TS for pairwise comparison.
  3. We optimize the first TS by the periods of indicators. And we run it on some part of the history. We also run the second system on the same part of the history (if it is on the other part, then the comparison is not paired, because the experiment conditions are different).

As a result, after the experiment we have four numbers:

  1. Number of profitable deals of the first TS - A
  2. Number of profitable deals of the second TS - B
  3. Number of losing trades of the first TS - C
  4. Number of losing trades of the second TS - D

Calculate the odds ratio:

Odds ratio = A * (B + D) / (B * (A + C))

If the odds ratio is at least 1.2, it means that we are on the verge of "scientific discovery".

It only remains to check the accumulated probability of the null hypothesis, i.e. how random our experiment is (probability of error of the first kind). To do this, we calculate the significance level p for the null hypothesis using the chi-square distribution with one degree of freedom and Yates's correction according to the contingency table (available in all statistical packages or online):

\
Basic TS
Control TP
Profitable trades
A
B
Loss trades
C
D


If p < 0.05, then the null hypothesis is rejected and the "discovery" is recognised as "scientific".

All that remains is to write an abstract on the above topic. The abstract must necessarily conclude with "scientific" conclusions, such as: "As a result of pairwise comparisons, we have scientifically proved that fitting the TS with (specify the specific TA) is more effective compared to the strategy of selecting the direction of market entry depending on the parity of the order number of the trade". The abstract should be published in some epigonous publication and wait for the dynamite inventor award.

This is how modern "science" is done.

 
Reshetov:

This is how modern "science" is done.

The main problem of Russian higher education is precisely that it does not teach people to think, it does not teach them to generate new ideas. They give us an excellent mathematical apparatus, a bunch of useless methods by themselves and say: "well done, now you are a real physicist/mathematician/econometrist!" - but all such a physicist-mathematicianeconometrist can do is mindlessly stir data balls in his head using abstruse algorithms and generate white noise at the output.
 
C-4:
The main problem of Russian higher education is that it does not teach people to think and does not teach them to generate new ideas. They give us an excellent mathematical apparatus, a bunch of useless methods by themselves and say: "well done, now you are a real physicist/mathematician/econometrist!" - but all such a physicist-mathematicianeconometrist can do is mindlessly stir balls of data in his head by abstruse algorithms and generate white noise at the output.

This is not a pan-Russian problem, nor is it of Russian origin. The fashion for "scientific" evidence using statistical packages came from the Wild West. Econometrics benefits big banks and biostatistics benefits pharmaceutical companies. They promote all this pseudo-scientific farming. Who pays, dances the girl.

The biostatistician or econometrician has no need to mix the balls in their head and they need their head only for eating, because they put one figure into some statistical package and receive other figures. All this stuff for a certain fee can be processed in ready-made abstracts and dissertations. Anybody can be engaged in a modern "science", if only money for experiments, registration of the dissertation and publications in epigonous magazines would be enough. The technology of riveting "scientific discoveries" is now a conveyor belt.

 
Reshetov:

Econometrics benefits big banks and biostatistics benefits pharmaceutical companies. They promote all this pseudo-scientific farming. Whoever pays, dances the girl.

But why promote and sponsor something that is unlivable. You can cheat a nuncular council, but you can't cheat the market. It always puts everything in its place. You won't make any money on all these digs at the "ACF UPCVy vpyazh? 82z garch arch CPh" (many letters, I can't remember all of them). So why should the banks sponsor this crap?
 
C-4:
But why promote and sponsor something that is not viable. You can fool the nouveau consulium, but you can't fool the market. It will always put everything in its place. You won't make any money on all these digs at the "ACF UPCVy vpyazh? 82z garch arch CPh" (many letters, I can't remember all of them). So why should the banks sponsor this crap?
And you go to a pharmacy and trivially calculate the approximate percentage of new drugs whose "effectiveness" has been "proven" by biostatistics and known before the fashion for statistical "proof". The results will surprise you. The reason is that when a person gets sick, he starts believing all sorts of advertised nonsense. It is the same with an economic crisis, the "science-intensive" econometric "predictions" become more in demand. I.e. the worse off we are, the better off they are. Filling the market with real medicines (if everyone is relatively healthy, who will buy patented drugs?) or fixing the economic crisis is not profitable for them
 
Eh .... writers .....
 
Reshetov:


The abstract should be published in some epigonous publication and wait for the award named after the inventor of dynamite.

The abstract is not published at all. Nowhere, Not supposed to.

This is how modern "science" is done.

This is how science is done by kindergarten graduates who finally manage to graduate at the age of 25.

 
Reshetov:
Go to a pharmacy and calculate the approximate percentage of new medicines whose "efficacy" has been "proven" by biostatistics and known before the fashion for statistical "evidence". The results will surprise you. The reason is that when a person gets sick, he starts believing all sorts of advertised nonsense. It is the same with an economic crisis, the "science-intensive" econometric "predictions" become more in demand. I.e. the worse off we are, the better off they are. Filling the market with real medicines (if everyone is relatively healthy, who will buy patented drugs?) or fixing the economic crisis is not profitable for them
But don't mix in a puddle of unscrupulous pharmacists who hide their shoddy practices behind scientifically based formulas from biostatistics, and evidence-based medicine, which is the standard in normal countries not because it benefits someone, but because only in this form it can be considered science. Sorry, but what we have in medicine in this country, where one doctor will tell you one thing, another will call them all assholes and then send you to treatment that won't get you killed, is called shamanism and is not science at all. People who have completed six years of medical school and a year of internship are our doctors and can decide whether you should live or not, prescribing treatment for patients solely on the basis of their diploma. There are no standards, no clear algorithms for prescribing medicines, nothing. In pharmacies there are antibiotics that are only formally sold on prescription, and the result is that people maim themselves because they simply cannot understand what this or that medicine does; resistance to antibiotics is rampant in our country.
 
Men, that's enough, strictly on the subject. Please.
 
faa1947:


The abstract is not published at all. Nowhere, it's not supposed to be.

You tell this tale to the authors and publishers of abstract journals, maybe they will believe it.

123456789
New comment