Adaptive lookback indicators - page 69

mntiwana
4557
mntiwana  
krelian99:

"More complex trend smoothing method"? An + and an - more in the equation and it is 'complex' Wow, these guys are great. I don't want to see their purchase list for the weekend dinner. 5 positions max on the list, for more they have to go twice or more The funny thing is, the shown 'adaptive' MA looks actually better, but the guy who wrote this is blind like a mole.

Fresh_Prince, every bonehead can write something in the internet and there are people who think because it is written there it is true. Make your own experience and see with your own eyes that this guy is totally wrong.

Dear Krelian,

well,your read and noticed,how peoples were writing 10 years ago,may be that time knowledge and technology was at this level or may be that specific person was at this level.

William Snyder
9492
William Snyder  

This is vhf adaptive price trender of averages, the color change is based on the price trend crossing the user price selection.

vhf_adaptive__price_trender_of_averages.ex4

AFT1
145
AFT1  
mntiwana:

Dear Krelian,

well,your read and noticed,how peoples were writing 10 years ago,may be that time knowledge and technology was at this level or may be that specific person was at this level.

mladen mntiwana krelian99 Actually from the very beginning I somehow finded this book: https://cdn.preterhuman.net/texts/finance_and_marketing/stock_market/Encyclopedia%20Of%20Trading%20Strategies.pdf There're lots of tests of different MA's - i was tryin' to find which one is the most effective. And on the page 125 and 130 we can see the results of testing each an every MA strategies (crosses and slope-leading) both trend and counter-trend entries - ALL the systems, based on MA's, according to authors - are worthless)

And i qoute page 131: "For the slope-based models, we thought the adaptive moving average, with its faster response, would provide the best performance; in fact, it provided one of the worst."

Then i found investopedia article which is mentioned above. That's why i decided to ask Mladen what he thinks about that.

Mladen Rakic
160260
Mladen Rakic  
Fresh_Prince:

mladen mntiwana krelian99 Actually from the very beginning I somehow finded this book: https://cdn.preterhuman.net/texts/fi...Strategies.pdf There're lots of tests of different MA's - i was tryin' to find which one is the most effective. And on the page 125 and 130 we can see the results of testing each an every MA strategies (crosses and slope-leading) both trend and counter-trend entries - ALL the systems, based on MA's, according to authors - are worthless)

And i qoute page 131: "For the slope-based models, we thought the adaptive moving average, with its faster response, would provide the best performance; in fact, it provided one of the worst."

Then i found investopedia article which is mentioned above. That's why i decided to ask Mladen what he thinks about that.

I would recommend STAD books with similar matter but with very different results. Taking into account that STAD books were written by tradestation professionals and traders documented with thorough back and forward tests on 100s of symbols and those systems and books took 13 years, year after year, each and every year with new tests, system improvements and testings,, and also taking into account my personal experience with trending strategies, I would say that the "encyclopedia of trading systems" is a superficial kind of "I know everything" type of "And I am going to sell my book to all" thing

It is, unfortunately, rare to find real tests, real knowledge about automatic systems, and real knowledge about components of automatic (and manual) systems in books written to have 1000 pages of whatever. The only 1, and I really mean only 2, dealing with issues like those, are the STAD books and Perry Kaufman's "Trading Systems and Methods" that I could recommend. There are some other books that are dealing with some other stuff, but the "Encyclopedia of ..." is not on my list

krelian99
1202
krelian99  
mntiwana:

Dear Krelian,

well,your read and noticed,how peoples were writing 10 years ago,may be that time knowledge and technology was at this level or may be that specific person was at this level.

After some time you can divide the adepts from the pseudo-scientists, the good from the bad ones. Famous is often not equal good, but only famous. Investopia is too superficial to me. They should rewrite it, but not in 10 years.

Mladen Rakic
160260
Mladen Rakic  

Just a quick update : the adapting of pdf ma will probably have to be avoided (due to the way how weights are calculated, it would need to be calculated on each new tick and it would make it highly impractical)

AFT1
145
AFT1  
mladen:
Just a quick update : the adapting of pdf ma will probably have to be avoided (due to the way how weights are calculated, it would need to be calculated on each new tick and it would make it highly impractical)

So we could never have adaptive step indicator? Or just with this particullar function?

Mladen Rakic
160260
Mladen Rakic  
Fresh_Prince:
So we could never have adaptive step indicator? Or just with this particullar function?

Pdf ma is bound to bars (it can calculate only integer periods) and that makes it a poor candidate for adapting (best for adapting is if the calculation allows fractional periods - like ema, T3 and similar ones). But also, since the coefficients/weights of pdf ma need to be adjusted for each period according to probability density function, for each new period that would be needed and different that the previous calculation period, the whole process of coefficients calculation would need to be done, and that would put some burden to CPU. Testing if some of the other 2 parameters of pdf ma (variance and mean) can be used for adapting in a more logical way, but so far the results are not promising

Mladen Rakic
160260
Mladen Rakic  

VHF adaptive ssm velocity (much smoother momentum) : ssm_velocity_vhf_adaptive_-_floating_levels.mq4

AFT1
145
AFT1  
mladen:
Pdf ma is bound to bars (it can calculate only integer periods) and that makes it a poor candidate for adapting (best for adapting is if the calculation allows fractional periods - like ema, T3 and similar ones). But also, since the coefficients/weights of pdf ma need to be adjusted for each period according to probability density function, for each new period that would be needed and different that the previous calculation period, the whole process of coefficients calculation would need to be done, and that would put some burden to CPU. Testing if some of the other 2 parameters of pdf ma (variance and mean) can be used for adapting in a more logical way, but so far the results are not promising

And what about XO? Can it be adaptive and doublesmoothed (134 averages)?