From theory to practice - page 31

[Renat Akhtyamov]

Yuriy Asaulenko:
No, but I can give you where it started in 2008. It's here.

What do you mean?

[Yuriy Asaulenko]

Renat Akhtyamov:

What do you mean?

You said here? I said here. In the kodobase, in short.) Sometime in 2009.

[Yuriy Asaulenko]

bas:

Very smooth, visually looks like a trace from the right-hand end of the approximation, not a muving)

SMA(8) will do, or LWMA(12). Although the muwings are of course less smooth.

The advantage of the approximation is not that, imho, but that it keeps up with the price, in relation to it (during the window) one can get variance more or less adequately.

It's not an approximation. Unless considered in a generalised sense.

On large periods it will lag far behind anyway - it is impossible to do real-time without re-tuning, but the delay is much less than standard MA and WMA.

[Renat Akhtyamov]

Yuriy Asaulenko:
You asked - here? I said - here. In kodobaz, in short.)) Sometime in 2009.

found some opera, I don't know if it'll fit

so I'm sitting here reading...

So far, so good.

http://sernam.ru/lect_math2.php?id=84

[Yuriy Asaulenko]

Renat Akhtyamov:

found some opera, I don't know if it'll fit

so I'm sitting here reading...

http://sernam.ru/lect_math2.php?id=84

I'll email you the link. But it's very old, it's from 2008.

[СанСаныч Фоменко]

Renat Akhtyamov:

found some opera, I don't know if it'll fit

so I'm sitting here reading...

nice one so far

http://sernam.ru/lect_math2.php?id=84

If you are interested in smooth curves approximation of cotier, then pay attention to the splines, which have the question of existence of derivatives in places of gluing pieces of spline is in all its glory.

For splines there are ready-made packages, if only you wanted to.

[Renat Akhtyamov]

Yuriy Asaulenko:
I'll send you a link in private. But ochrevaya old, already in 2008.

SanSanych Fomenko:

If you are interested in approximating a quotient by smooth curves, then look out for splines, which have the issue of the existence of derivatives in places where the spline pieces are glued together in all their glory.

There are ready-made packages for splines, if you'd like.

Thank you!!!

[secret]

Yuriy Asaulenko:

On large periods it will lag far behind at any rate - you can't do real-time here without rearranging anything, but the delay is considerably less than standard MA and WMA.

The trace of the right-hand end of the approximation will lag of course, but I mean the approximation line itself - it doesn't lag behind anything throughout the window.

[Alexander_K2]

bas:

Alexander, this may surprise you, but the regular SMA, and on 5-minute bars (right), goes almost identical to your contrived one on ticks (left). On the scale of your trades, the difference is almost imperceptible. Where is the "special precision in the behaviour of the average" here?

As far as I remember - that model simply uses moving average arithmetic MA. it is now I use WMA. Although I don't insist on that particular choice and am now reading carefully - what traders use and why.

[secret]

Alexander_K2:
As far as I remember - that model just uses moving average arithmetic MA. it's now I use WMA. Although I don't insist on that particular choice and am now reading carefully - what traders use and why.

So you are misleading us - you post one model, but write about another)

Well then give the same picture with the WMA.