Machine learning in trading: theory, models, practice and algo-trading - page 1633
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
They initially form market expectations. Options on SI form the market expectations for this particular instrument. That is, what option traders think about the future price of the underlying asset. Whether they think correctly or not is not known, but their expectations are known through the smile curve and the set of certain positions on the option. Then there is a trading volume in accordance with this expectation or not. No one said it was that simple. Just by using this data and look at the market from this angle, you get into a group of professionals, who also cheat each other and try to cheat, but not using this information you just play roulette. IMHO of course.
Here's a tip. Follow during the day for this board, even if the pound, even with a delay of 15 minutes for free users and at dexterity you will be extremely surprised how it turns out to be simple.
This screen is for an option on the pound on the CME, if I am not mistaken
Interesting interpretation of options,
can you elaborate?
On what basis do you get expectations up or down?
I want to use the options data myself. Maybe it works because they say so - psychology, not logic.
I would like to give trade orders there as well, but it costs money - no wish to unite?
Interesting interpretation of options,
can you elaborate?
On what basis are the expectations up or down?
And for what it's worth. Since I collected OM for a sufficient period of time, I decided to increase training sample to 60 signals + 10 test and finally to 70. At first I saved a training file without OM and obtained approximately 150 useful inputs for 70 vectors. That seems to be normal, there are twice as many explanatory variables as vectors they explain. There is a lot to choose from. I have started Mile Shaitan machine, but it did not manage to get the quality of learning more than 88%, which is not good in principle. Well, I think that's okay, the OI are collected. I will turn them on. As a result, training file turned out 190 columns for the same 70 examples. I have to confess that there's a lot of trouble with OM, which I still have to work out. It writes everything to a file and use it in the tester is extremely difficult. Well, I think the hell with it. As a result, I start the shaitan machine and after a while it gives the following result: 95% coverage in training and 100% in the test (the best result), I think normal. Just on the delta volume, etc. Well that's how to be? :-) And when you are an exceptionally practical person you come across similar phenomena once in a while, which have yet to be explained.
The blue is the OOS, as I said leave 2 signal indentation. The only indication of the quality of TC is that it does not err even in small things. The point is that with this approach errors are not allowed. The time interval of polynomial survivorship is extremely small. It is literally 5-8 trades and the minus is not relevant here. Of course, the polynomial may work longer and as a rule it does that, but I only place guaranteed orders up to 10 trades and consider anything more than 10 as a super profit. Let's see how this sample works.
Mikhaylo, write in voice or iambic.)
Mikhaylo, write in voice or iambic.
Exactly, poetry in prose. It reminds me of Andrei Platonov.