It is an interesting idea to forecast something within (inside) a market model.
Forexample, we have the bolinger as a market model, and we make forecasts within this model...
The bolinger has three indicators: upper boundary, lower boundary and average.
We predict what line the price will cross in the near future.
crossing means that the line will be between the high and low of the candle.
So we forecast three classes.
The model learns with such data quite well, on new data the error does not fall and behaves stably.
Confusion Matrix and Statistics
Reference
Prediction -1 0 1
-1 3647 721 10
0 817 5013 948
1 22 655 3167
Overall Statistics
Accuracy : 0.7885
95% CI : (0.7818, 0.795)
No Information Rate : 0.4259
P-Value [Acc > NIR] : < 2.2e-16
Kappa : 0.6745
Mcnemar's Test P-Value : 8.019e-14
Statistics by Class:
Class: -1 Class: 0 Class: 1
Sensitivity 0.8130 0.7846 0.7678
Specificity 0.9305 0.7950 0.9377
Pos Pred Value 0.8330 0.7396 0.8239
Neg Pred Value 0.9210 0.8326 0.9141
Prevalence 0.2991 0.4259 0.2750
Detection Rate 0.2431 0.3342 0.2111
Detection Prevalence 0.2919 0.4519 0.2563
Balanced Accuracy 0.8717 0.7898 0.8528
Also on the market the model sees trends very well and never makes a mistake in the direction of the trend, although it lags, but behaves much, MUCH more adequately than the same zig-zag, I highly recommend to experiment...
I've already experimented with the Donchian channel instead of the Bolinger.
So try it, you'll like it.
============
It is difficult to explain, but the point is that we do not forecast abstract up-down, but "lock" AMO in the framework of some model and forecast inside, as if we reduce the degrees of freedom for AMO....
Well? From now on, the prediction is which of the three lines will be the first to be crossed in the future.
Yes yes yes yes yes... of course... but ZZ Akourasi is 55-58 and I'm at 79, but there's no difference...
Well, if the goal is just numbers...
If the forecast on each bar, there is still a distortion, because it is clear that the trend is more likely to continue than to end, i.e. finding above/below the average value will automatically give 60% of the correct outcome.
On the Donchian channel I put 80% and 20% up to the boundary (the upper boundary is 100% and the lower boundary is 0%) - these are points for forecasting and making a trading decision. Stops and trawl on the opposite border of the channel.
Allow the use of cookies to log in to the MQL5.com website.
Please enable the necessary setting in your browser, otherwise you will not be able to log in.
You are signed in. Two-factor authentication has been disabled. Set it up again and scan a new QR code, or continue with two-factor authentication disabled.
What can I say?
Sandbox!
No specifics).
Goodbye.
It is an interesting idea to forecast something within (inside) a market model.
Forexample, we have the bolinger as a market model, and we make forecasts within this model...
The bolinger has three indicators: upper boundary, lower boundary and average.
We predict what line the price will cross in the near future.
crossing means that the line will be between the high and low of the candle.
So we forecast three classes.
The model learns with such data quite well, on new data the error does not fall and behaves stably.
Also on the market the model sees trends very well and never makes a mistake in the direction of the trend, although it lags, but behaves much, MUCH more adequately than the same zig-zag, I highly recommend to experiment...
I've already experimented with the Donchian channel instead of the Bolinger.
So try it, you'll like it.
============
It is difficult to explain, but the point is that we do not forecast abstract up-down, but "lock" AMO in the framework of some model and forecast inside, as if we reduce the degrees of freedom for AMO....
Here's another variant of what and how you can predict
Predicting which line the price will cross in the near future...
At what point do you make the prediction?
I've already experimented with the Donchian channel instead of the bolinger.
The Donchian channel is almost synchronous with the standard ZZ - the moment of crossing the border == the moment of appearance of a new ZZ segment.
As it is, I've been using this strategy since the beginning in MO.
Apparently the problem here is that no one hears each other, so everyone goes round in circles....
Apparently the problem here is that no one hears each other, so everyone goes round in circles....
You can say a thought in 1000 ways, but only 1 is understood by everyone)))))
You can say a thought in 1000 ways, but is only 1 understandable to everyone)))))
So the policy should be reversed - not on the efforts of the speaker, but on the efforts of the listener.
It is not so difficult to ask a question or ask a question - it is just a megalomaniacal attitude of many people who think of their uniqueness....
At what point do you make a prediction?
Well? from the current moment you make a forecast - which of the three lines will be the first to be crossed in the future.
Donchian channel is almost synchronous with the standard ZZ - the moment of border crossing == the moment of appearance of a new ZZ segment.
And so, I used this strategy from the very beginning in MO.
Apparently the problem here is that no one hears each other, so everyone goes round in circles....
Yeah, yeah, yeah, yeah, yeah, sure. only ZZ has akurasi 55-58 and I have 79, but there is no difference... of course... ))
Well? From now on, the prediction is which of the three lines will be the first to be crossed in the future.
Yes yes yes yes yes... of course... but ZZ Akourasi is 55-58 and I'm at 79, but there's no difference...
Well, if the goal is just numbers...
If the forecast on each bar, there is still a distortion, because it is clear that the trend is more likely to continue than to end, i.e. finding above/below the average value will automatically give 60% of the correct outcome.
On the Donchian channel I put 80% and 20% up to the boundary (the upper boundary is 100% and the lower boundary is 0%) - these are points for forecasting and making a trading decision. Stops and trawl on the opposite border of the channel.
Nah, well, if the goal is just numbers.....
no comment))))
Nah, well, if the goal is just numbers.....
why, the intersection of a function and its derivatives can also be a goal)