Machine learning in trading: theory, models, practice and algo-trading - page 1779
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
show the chart with the trades
You can see the bars and areas of good classification.
This is if you activate on every bar.
You can see the jams and areas of good grading.
This is if you activate on every bar.
Well, yes, it looks like mine...
Well, we have to think and reduce the error...
I see two ways
1) improve quality of signs, what I wrote about two pages ago
2) change/smooth out the price, but before it reaches the training, post factum smoothening is of no use
Well, yes, it seems to be the same as mine...
Well, we have to think and reduce the error...
I see two ways
1) improve the quality of the signs, something I wrote about two pages ago
2) change/smooth out the price, but before it gets to the training, post factum smoothening is useless
Do you think we have the same predictors? :)
It is possible to squeeze something out of the current situation - think of a strategy.
I have added a condition - we activate only against the current segment ZZ, i.e. we catch reversals.
Do you think we have the same predictors? :)
and you think you can generate a lot of different things from OHLC ? :)
the target is the same...
same acuracies ...
Do you think you can generate a lot of different things from OHLC? :)
the target is the same...
the same akurasi...
For example, I don't have returns in pure form.
I can drop OHLC, generate the predictors and check the synergy like you wanted?
For example, I don't have pure returns.
I can drop OHLC, generate predictors and check the synergy, as you wanted?
Drop it off, let's see .... I will definitely not look at it today, sorry, I am in my head in another topic
This is not an option. There is too much hoarding in the flat.
You could of course shift the activation threshold to 0.65 - buy, and 0.35 - sell.
Do you always have more continuous losses, reports on the history that did not participate in the training?
>You can, of course, shift the activation threshold of 0.65 - buy, and 0.35 - sell.
And how often and when will you change these parameters, i.e. what will be the sign that the data for activation have changed? Will you really test the system after the National Operator and select the optimal values for filters (activation)?
Do you always have a greater continuous loss, reports on the history that did not participate in the training?
>You can, of course, shift the activation threshold to 0.65 - buy, and 0.35 - sell.
And how often and when will you change these parameters, i.e. what will be the sign that the data for activation have changed? Is it really after the TC you will test the system and select the optimal values for the filters (activation)?
The reports were not involved in the training.
This is called balancing, it is done once.
The problem of targeting is that there is no strategy as such, so if the classification is highly accurate in its bare form, you get a plum.
Reports that did not participate in the training.
This is called balancing, it is done once.
The problem with targeting is that it lacks strategy as such, so high accuracy in the naked classification results in a plum.
High accuracy is not 70%.
The problem with the target is that it lacks strategy as such, so with high classification accuracy in the naked form you get a plum.
Strategy... So after MO, you still need a whole strategy? It seemed to me that the purpose of MO is to finally give you an advice to trade (long/short) or not.