Statistics as a way of looking into the future! - page 6
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
Prival wrote >>.
We build the model ourselves, based on certain assumptions. So what prevents us from adding or subtracting a parameter. Can you justify the appropriateness of the choice?
I'll have to check my "predictor" I've never done that, got curious. I'll try to post the results tonight.
Very interesting, let's wait for the results...
to m_a_sim
So, on what TF is this result obtained and what is the instrument volatility calculated?
Prival wrote >>.
We build the model ourselves, based on certain assumptions. So what prevents us from adding or subtracting a parameter. Can you justify the appropriateness of the choice?
Very interesting, let's wait for results...
to m_a_sim
So, on what TF is this result obtained and what is the instrument volatility calculated by you?
daily, didn't calculate volatility, don't want to)
daily, I didn't calculate the volatility, I don't want to)
For almost all instruments the daily volatility is in the range of 50-100 points. If you did not make a mistake in constructing your algorithm, you can be congratulated on creating a super profitable strategy. With tangent 0.3 it will provide the average yield of 50*0.3=15 points per transaction, minus spread. Total: about 10 pips per day with a spread of plus or minus 5 pips (see fig. increments).
A decent result. Is it also good at other symbols. For example EUR/USD?
I.e. having a difference forecast at each step, what is the decision-making algorithm in which your conclusions are adequate?
Come on: we have at the opening of each daily candle a prediction of its increment before the next one opens...
I made an inaccuracy in the previous post. The spread of the predicted value is defined as the root mean square of the prediction error (5 pips) and the instrument daily volatility (50 pips), i.e. the rate of profit growth can be estimated as 10+-50 pips per day on average. From this we can estimate the risks and determine the optimal capitalization of the position.
Presented version of the forecast of daily bars with the declared accuracy is good enough, if we consider the possibility of diversification of risks in a portfolio investment.
I want to ask the author one more time: is the result similar for other symbols?
Can you tell me more about your "fortune teller"?
It's all there in bits and pieces, but you can't tell it all fast enough in a Word-processed file.
Optional.
Present the result in Cartesian coordinate system, plot the increments of the predicted value on abscissa axis and model predictions of these increments as points on ordinate axis. Ideally (100% accurate prediction) you will get a cloud of points with a 45 degree slope. In reality you will get a low-pitch cloud (which is a measure of prediction accuracy) and a very thick one (this measure is a measure of the prediction scatter). To do it you should build a series of first differences for the tool (x[i]=Open[i]-Open[i-1]) and a series of first differences for the forecast (y[i]=Predict[i]-Predict[i-1]).
Then we can talk about the quality of your model.
Here's what I got
there is no slope at all (. May be I did something wrong.
Come on: we have at the opening of each daily candle a prediction of its increment before the next one opens...
I made an inaccuracy in the previous post. The variation of the predicted value is determined as the root mean square of the prediction error (5 pips) and the instrument daily volatility (50 pips), i.e. the profit growth rate can be estimated as 10+-50 pips per day on average. From this we can estimate the risks and determine the optimal capitalization of the position.
Presented version of the forecast of daily bars with the declared accuracy is good enough, if we consider the possibility of diversification of risks in a portfolio investment.
I want to ask the author one more time: is the result similar with other instruments?
I did not try it with other instruments because I do not know the factors affecting them