You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Are there pullback systems that backtest well?
- Thread starter jimcrist
- Start date
- Nothing astonishing backtests have much chance to be bullshit for statistical reason ... -
harry
trusting so much on technology that uses past data and at the same time diminishing backtesting is a unique point of view. there is only one reason why you do not do backtesting: you save the effort. took me some time to extract that essence from your many posts.
peace
harry
trusting so much on technology that uses past data and at the same time diminishing backtesting is a unique point of view. there is only one reason why you do not do backtesting: you save the effort. took me some time to extract that essence from your many posts.
peace
Quote from nitro:
You know,
It is funny that you should say that. I have an idea for just for using a 3D "view" of the markets. But every time I try to use some of my math background in anything but the most trivial way, the markets just hands me a bunch of losers in backtest.
But I am getting better at it.
Thanks for the kind words.
nitro
that is a real sad thing about this business. whenever i thought something is good, smart and beautiful it turned out to be completely useless. acrary had that once: don't expect any reward out of this business but ... money. you probably won't get modelling satisfaction. at least i did not.
Quote from mind:
that is a real sad thing about this business. whenever i thought something is good, smart and beautiful it turned out to be completely useless. acrary had that once: don't expect any reward out of this business but ... money. you probably won't get modelling satisfaction. at least i did not.
Sad but true... but we still seek for that...
At least I'm satisfied with my Gann Analysis... he he he
qoute from WDGann:
Sad but true... but we still seek for that...
At least I'm satisfied with my Gann Analysis... he he he
---------------------------------------------------------------------------------
Unfortunately some of the Gann stuff does not lead itself easily into objective backtesting
Sad but true... but we still seek for that...
At least I'm satisfied with my Gann Analysis... he he he
---------------------------------------------------------------------------------
Unfortunately some of the Gann stuff does not lead itself easily into objective backtesting
hahahaha the validity of a backtest has nothing to do per se with the use of past datas. It has to do with the ADEQUATION of the model, the SAMPLING, the Degree of Freedom etc. I can't make a whole engineering course to you on that here.
And you lack some logic because I didn't say that ALL backtests are bullshit I said it has much chance that they are since many of them are based of INADEQUATE models because they are based on STOCHASTICS and since there is no AUTOCORRELATION in datas they are just inadequate models.
And you lack some logic because I didn't say that ALL backtests are bullshit I said it has much chance that they are since many of them are based of INADEQUATE models because they are based on STOCHASTICS and since there is no AUTOCORRELATION in datas they are just inadequate models.
Quote from mind:
- Nothing astonishing backtests have much chance to be bullshit for statistical reason ... -
harry
trusting so much on technology that uses past data and at the same time diminishing backtesting is a unique point of view. there is only one reason why you do not do backtesting: you save the effort. took me some time to extract that essence from your many posts.
peace
Quote from harrytrader:
hahahaha the validity of a backtest has nothing to do per se with the use of past datas. It has to do with the ADEQUATION of the model, the SAMPLING, the Degree of Freedom etc. I can't make a whole engineering course to you on that here.
And you lack some logic because I didn't say that ALL backtests are bullshit I said it has much chance that they are since many of them are based of INADEQUATE models because they are based on STOCHASTICS and since there is no AUTOCORRELATION in datas they are just inadequate models.
embarrassed?
Quote from harrytrader:
hahahaha the validity of a backtest has nothing to do per se with the use of past datas. It has to do with the ADEQUATION of the model, the SAMPLING, the Degree of Freedom etc. I can't make a whole engineering course to you on that here.
And you lack some logic because I didn't say that ALL backtests are bullshit I said it has much chance that they are since many of them are based of INADEQUATE models because they are based on STOCHASTICS and since there is no AUTOCORRELATION in datas they are just inadequate models.
As already said here (most market indicators are just the most primitive forms of time series techniques) :
http://www.elitetrader.com/vb/showthread.php?s=&postid=444764&highlight=autocorrelation#post444764
Quote from harrytrader:
The very basic reason why stock market time series are reputed to be one of the most difficult arena of forecast is because these classical time series techniques don't work, mathematically these techniques are based on autocorrelation of errors since these autocorrelations are very low in stock market time series they are not worth at least used in traditional way. Now low autocorrelation is not equivalent to independancy, it has been showned for a long time since Mandelbrott that the Market exhibits "long term memory effect" so that the latest kind of stochastic model taking into account that effect is ARFIMA's model. But the performance still is poor. All in all I say it is an error to use stochastic models to do market's forecast because only a deterministic model can do it (ie mine of course ), the problem is to find it and the reason that researchers didn't find it is because they try to extract knowledge from pure datas which is an idiocy from paradigm point of view because the model's knowledge is transcendant to the datas that is to say you cannot deduce it from the datas alone but only if you have the idea of how market really works or you will play with datas and stochastic models much like a monkey see:
http://www.elitetrader.com/vb/showt...&threadid=28614
ANNs (Neural Net): A Little Knowledge Can Be A Dangerous Thing
http://www.secondmoment.org/articles/ann.php
ANNs: A Little Knowledge Can Be A Dangerous Thing
Posted by Dr. Halbert White
When under backtest such indicators seem to perform well, it's almost due to persistency of trend (see Arcsinus law: distinguishing trend from persistency of chance http://www.elitetrader.com/vb/showthread.php?s=&threadid=22256&highlight=persistency+and+trend) that is to say that they are meaningless inferentially speaking as for pertaining to their performance in the future ! That's why the backtest on these indicators are flawed from their very ground and all datas about returns and decorations to appear "scientific" are just cosmetics that won't erase that ! Only those who confuse EMPIRICAL statistics which concerns past performance only and ignore the INFERENTIAL statistics would be fooled.
So pure MECHANICAL SYSTEM based SOLELY on these kind of indicators are due to be DOOMED SOONER or LATER. Of course there is always a chance that they don't fail but it is due to chance, for example such a system for daily scale could outperform during the past 20 years of Bull market but on the next "sampling" of twenty years you don't know. Being impressed by such a performance, especially if it is compounded, is to be fooled by probability fallacies.
So pure MECHANICAL SYSTEM based SOLELY on these kind of indicators are due to be DOOMED SOONER or LATER. Of course there is always a chance that they don't fail but it is due to chance, for example such a system for daily scale could outperform during the past 20 years of Bull market but on the next "sampling" of twenty years you don't know. Being impressed by such a performance, especially if it is compounded, is to be fooled by probability fallacies.
Quote from harrytrader:
As already said here (most market indicators are just the most primitive forms of time series techniques) :
http://www.elitetrader.com/vb/showthread.php?s=&postid=444764&highlight=autocorrelation#post444764
Quote from harrytrader:
The very basic reason why stock market time series are reputed to be one of the most difficult arena of forecast is because these classical time series techniques don't work, mathematically these techniques are based on autocorrelation of errors since these autocorrelations are very low in stock market time series they are not worth at least used in traditional way. Now low autocorrelation is not equivalent to independancy, it has been showned for a long time since Mandelbrott that the Market exhibits "long term memory effect" so that the latest kind of stochastic model taking into account that effect is ARFIMA's model. But the performance still is poor. All in all I say it is an error to use stochastic models to do market's forecast because only a deterministic model can do it (ie mine of course ), the problem is to find it and the reason that researchers didn't find it is because they try to extract knowledge from pure datas which is an idiocy from paradigm point of view because the model's knowledge is transcendant to the datas that is to say you cannot deduce it from the datas alone but only if you have the idea of how market really works or you will play with datas and stochastic models much like a monkey see:
http://www.elitetrader.com/vb/showt...&threadid=28614
ANNs (Neural Net): A Little Knowledge Can Be A Dangerous Thing
http://www.secondmoment.org/articles/ann.php
ANNs: A Little Knowledge Can Be A Dangerous Thing
Posted by Dr. Halbert White