MACHINE period with minus value - page 33

 
alsu:

It makes no difference, you can write it backwards using the same principle and you will get /in the same notation/: (X1 + X2 +X3)/3 = (X3 - d1 -d2 + X3 - d2 +X3)/3 = X3 - 2/3*d2 - 1/3 * d1 (you have X1 + 2/3*d1 + 1/3*d2), i.e. following your logic about greater or lesser consideration of increments, exactly the opposite. And that's because the formula mixes signal values and incremental values, you can't draw any conclusions from such expressions.

In general, the input signal for a filter is usually not incremental, but directly the value of the input. If you want to filter the increments, you must first use a differentiating element.



Regarding the influence of increments, I agree, it depends on what price it is counted relative to. If from the price of the beginning of the period, then the weight of the last increments decreases. If from the last price, vice versa. In practice it doesn't matter what to change the coefficients in relation to. Only whether they should be changed - or to differentiate - I do not know. Therefore I do not need))

 
The LRlinear regression (which is the difference between linearly weighted and regular dashes with the right coefficients) is a filter, in which the latter weights can even be negative. And the leading weights (which are closer to the zero bar) are much higher than those of the regular dash. Therefore the LR lags behind the price much less than the dash. But as a consequence, it reacts more strongly to price fluctuations.
 
Zhunko:
Not everyone. I'm the only one. Somebody has to take advantage of the lowering of supply before demand. Forex taught me that.

: )
 
sultonov:
Bought minus 10 apples = sold 10 apples. The grandmother lost 10+3+5 = 18 apples. How many remain is unknown.

Nice anecdotes! I'm getting more and more in control of the fact that this isn't a comic portal after all, seeing as it fires such cruel and accurate bullets that hit you right in the nutshell. : )
Reason: