Indicators: Price prediction by Nearest Neighbor found by a weighted correlation coefficient - page 4

[Renat Akhtyamov]

Ryan L Johnson #:
This is what the blue line is for, which overlays the opening prices of the historical pattern.

You misunderstand me

At the moment the red line is drawn where there is no chart, as it is a forecast.

I propose to draw the red line where there is a chart.

To do this, we should calculate the red line not from the end of the chart, not on the current date, but for example a few days earlier.

As a result, we will get the forecast and immediately see the result, i.e. what really happened to the price and whether we should believe the forecast/indicator ?

[Ryan L Johnson]

Renat Akhtyamov #:

You misunderstand me

At the moment the red line is drawn where there is no chart, as it is a forecast.

I propose to draw the red line where there is a chart.

To do this, we should calculate the red line not from the end of the chart, not on the current date, but for example a few days earlier.

As a result, we will get the forecast and immediately see the result, i.e. what really happened to the price and whether we should believe the forecast/indicator ?

I'm afraid that "misunderstand" is an understatement here. The shifting of the beginning of the red forecast line into the past would do violence to the indicator. If you do some research regarding k-NN theory, you'll likely see what I mean. The correlation coefficient prints very clearly to the Experts tab of the terminal, so Vladimir has done the hard work for us. All that we have to do is deduce the correlation coefficient threshold that is acceptable. Stated more accurately, I have no idea what you're trying to achieve nor how.