Machine learning in trading: theory, models, practice and algo-trading - page 403
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
If you trained on 3 months of data, you can't expect the model to last much longer. What market the model has seen, it will be able to trade. Your dataset is nonsense, trading on it is like guessing with coffee grounds. The same is true for "Reshetov's machine" that assigns coefficients for a linear model, while the data are not linear at all. You have to be not at all distant to believe in this nonsense that on a dataset of <500 points a linear model took weeks to learn, because it is "AI")))))))))..... I don't know.... this is more nonsense than martingale and "depo boosting"
If you have more experience, it can be felt, let the newbies build up their own bumps, maybe something interesting will come out. I personally will be able to say something clear about this model in about 2 weeks, when I've at least got an idea how it works :)
I will send you a model with 100% level of generalization....
Perhaps the guys fromLTCM, too, and so argued)) They say that if they had looked twice as far back in their models, they wouldn't have merged so fiercely.
They say that if they look twice further back in their models, they would bleed so much fieri. Learning in any case is not going on the whole sample in 5 years, it certainly takes a sliding window on the sample, with a constant learning, but it's important to know how quickly the model "understands" when something strongly changes in the market and this is not an outlier, not someone's stupidity, and will not boldly fill up against a sudden trend, the call of Kolya.
How did they manage to lose on hft when the risks are very tightly controlled there and any deviations just shut down the system with minimal losses. martingey or what).
Who are they LTCM? how did they manage to lose on hft, when there are very tightly controlled risks and at any deviations the system just shuts down with minimal losses. marting or what )
It was your brothers in common sense: they also believed that if today everything is profitable on the real market, then it will always be so. And if you consider that there were a bunch of Nobels, then you too are on the doorstep of the Nobel Committee.
These were your brothers in the mind: they also thought that if everything is profitable in the real today, then it will always be so. And if you consider that there were a lot of Nobels, then you are also on the doorstep of the Nobel Committee.
All righty then. I'll put the prize in a frame and hang it over the table ..... or better over the bed.... I'd love to see it :-)
These were your brothers in the mind: they also thought that if everything is profitable in the real today, then it will always be so. And if you consider that there were a bunch of Nobels, then you are also on the verge of the Nobel Committee.
I think you slightly overestimate my mind :) I never thought so, I have yet to grow and grow (in quotes)
Regarding the file posted in the first post. It's an interesting task. I think the optimizer of the solver will figure it out at once... Because there will be repeating strings with data, so there will be retraining. But I'll certainly calculate it, just for the sake of interest, and we'll see what the result will be. Although the number of lines there is not small. But we'll see...
Try it, but I would not do meaningless abstract rubbish and would write TC sitting in your place, time is fleeting and it can go to all sorts of meaningless tasks with some there invented predictors and targets, which have nothing to do with reality. Sometimes you need to concentrate on your ultimate goal for success and not listen to anyone :)
NS is only part of the system, and sometimes not even the most important. For them, it's just mind games, not real work, I guess. All we're interested in here is the development of TC and the launch into production. I've already run one and am happy with it, now I'm doing a more complex one.
Try it, but I would not do meaningless abstract stuff and write systems sitting in your place, time is fleeting and it can go on all sorts of meaningless tasks with some made-up predictors and targets that have nothing to do with reality. Sometimes you need to concentrate on your ultimate goal for success and not listen to anyone :)
NS is only a part of the system, and sometimes it's not even the most important.
Golden words Victor Benedictovich..... I still need for Alyosha model in 100? Well this is quick, and then you can calculate the file from the first post. What to do when the TS is trained and working? Right. Follow its signals. But this is terribly boring....
So how about a trading skeleton? Who has one???