Why is the trend our friend? - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Homepage.
As to really math basement for trend detection:
There are two approaches-
1). DSP approach.
Problem: very slow, extremely complex, very difficult to put into real computer program. Probably it is used by James Simons in Renaissance Technologies. They use very uncommon math formulas, totally unknown in trading/finance world.
Renaissance Technologies - Wikipedia, the free encyclopedia
James Harris Simons - Wikipedia, the free encyclopedia
2). Statistical/probability approach.
Problem: everything after 1945 is useless due to extreme distortion of math science after Kolmogorov's axioms.
Both approaches are very hard to put into working computer program.Pretty interesting. As far as i've read, Renaissance hires 2 types of people more than the others. One is voice recognition specialists & i recall somewhere that James Simons said that prices are non-random and that there are price patterns (not the usual kind). Some kind of DSP is probably involved here. The 2nd type are cryptanalysts & it is probable that the related field we are talking about here is about information theory, most of which has to do with entropy. But still even then, they suffer alot of gut-wrenching volatility even with their superior systems.
With regards to statistical theory, why do you say that everything is useless due to extreme distortion after Kolmogorov? Name some particular instances and explain yourself more? Interested to hear more with regards to this aspect
Yours sincerely,
Wintersky
Pretty interesting.
With regards to statistical theory, why do you say that everything is useless due to extreme distortion after Kolmogorov? Name some particular instances and explain yourself more? Interested to hear more with regards to this aspect
Yours sincerely,
WinterskyThe Kolmogorov's so-called axioms are not "axioms", they are all *** tautology ***.
This is a trap, explained by
Gödel's incompleteness theorems - Wikipedia, the free encyclopedia
and
https://en.wikipedia.org/wiki/Church%E2%80%93Turing_thesis
Adding that modern "statisticians" use formulas for Gauss' distribution everywhere, even if there is no Gauss' distribution on horizon, You will see exactly why modern "theory of probability" does not work on human system, such as trading.
All swings contain a trend element .the trick is to differentiate between a trend that is a corrective and a trend that is impulsive. ie moving with or against the higher degree swing or trend.
Trading a new trend is always best but make sure that new trend is not corrective ie moving against the higher degree swing or trend for best results.
Your probably have more success with what you consider to be counter trend trading because you are entering at the beginning of a new trend, if the swing in the opposite direction has hit a top then you get lucky and get some pips but quite often what traders things is a trend reversal is merely a new trend in a corrective swing.
Therefore the trend is not always your friend.
Hope this helps
This is one of the golden sayings in the forex trading business because if one ride the trend then there are all chances that they can find some profitable trades and thus ride huge number of pips if you get the trend at the early stages.
Adding that modern "statisticians" use formulas for Gauss' distribution everywhere, even if there is no Gauss' distribution on horizon, You will see exactly why modern "theory of probability" does not work on human system, such as trading.
Not exactly true perhaps. The current academic studies in statistical trading is that the market doesnt have infinite variance nor that it is impossible to give a general approximation to distribution. There was a time that the academic community thought that levy processes and Cauchy distributions were the only way to go. Debunking the old kurtosis measures and using robust kurtosis formulas, it is found that aside from the rare once-in-a-year (or more) or so extreme SD event, financial series are well-modeled with SDs less than 5 or so, but pinpointing it more precisely is a problem itself.
Personally, i would say that the distributions are non-stationary and vary within a limited range too, hence making it hard to nail down more precisely beyond this. & i can tell you that there are those who have used Gaussian distributions in trading systems successfully. The only thing that matters here is if you know what you are doing to make the necessary indicator complements to develop a full system. The theory is one thing but the applications can be tweaked if you make an effort. But if you are looking for the holy grail though, then you will probably not find it.
With regards to modern statistics, there are companies who have successfully used Levy or Cauchy distributions in trading, while there are also those which use improved versions of Bayesian probabilities. The issue is a problem of people not actively investigating the application possibilities & remaining stuck in mindset to rigorous application. As such, it is best that the application methods remain "trade secrets" to be profitable only for those who "think out of the box" & have their own insights and learning.
As some food for thought, many people consider 30% accuracy to be the minimum for a profitable system. In japan, there is an old trader who only has 10% accuracy & yet has been consistently profitable for decades & being one of the top profitable traders there. I am certain there are statistical experts with regard to profitable trading here on this forum itself & i am pretty sure that what i am blabbering so far is a fraction of their vast knowledge and experience for they choose to remain lurking and secretive in the background. Sharing out a system will only mean it's death some day. The best expert in statistical trading is probably William Eckhardt but sadly, we have no way of catching up to his brilliance.
Yours sincerely,
Wintersky
Not exactly true perhaps. The current academic studies in statistical trading ............. Wintersky
I dont see where what I said is contradicting to what You said about "academic studies" or hidden know-how. Therefore I dont get why You call me a liar.
As a proof I can easily show up 100 000+ posts on this trading forum with Moving Average, which is first, clear and perfect exploit of Gauss (normal) distribution.
I can show up 100000+ scientific studies on a probability theory published EACH YEAR. Only 1-5% from them are using non-Gauss distribution. Non-parametric or robust methods' software packages like SPSS
SPSS - Wikipedia, the free encyclopedia
are almost unknown in trading world and have very limited application in economy.
Why am I a "liar"? Please, explain.
The trend is your friend. That is true.
But I like to do counter trend as well to get quick pips.
Oh the counter trend strategy is good if you can enter at exact point because if you got it wrong then there will be many red pips unless you have used a stop loss. I hope you are using a tight SL in counter trend trades.
As a proof I can easily show up 100 000+ posts on this trading forum with Moving Average, which is first, clear and perfect exploit of Gauss (normal) distribution.
I can show up 100000+ scientific studies on a probability theory published EACH YEAR. Only 1-5% from them are using non-Gauss distribution.No one's calling you a liar. As i said before, the market doesn't have very high kurtosis as previously thought & the exact PDF varies slightly/moderately over time too in a limited range much much closer to Gaussian as we know now, but certainly not 100% Gaussian too, but more like a forcibly mixture of 2 Gaussians, the composition of which varies with time too even.
To put it more explicitly, the markets will never be totally Gaussian nor totally non-Gaussian on a real time basis. Markets are widely acknowledged to be approximately 15% trending and 85% non-trending (mean reverting/Gaussian-like/Gaussian). It is the 15% of the time that makes the PDF so non-Gaussian. & that, partially is why there is so much academic studies on Gaussian stuff and traders and indicators here on this forum still using Gaussian-category indicators. In terms of time, the market is Gaussian type more more often, while in terms of overall distribution it is a a limited non-Gaussian distribution. On a real-time basis in trading, we shall see a mix of both in different times of the day & have to have ways to handle both.
You said in post #24: "Adding that modern "statisticians" use formulas for Gauss' distribution everywhere, even if there is no Gauss' distribution on horizon, You will see exactly why modern "theory of probability" does not work on human system, such as trading""//////
In terms of overall distribution, yes you are right. But that does not preclude us from using Gaussian-type indicators because, in terms of TIME, the markets are more Gaussian-like. At the end of the day, it is up to the individual which type of strategy (mean-reverting/trending) he wants to use, which dictates the use of which type of distribution, each strategy with their pros and cons.
Overall non-Gaussian PDF in financial time series doesn't mean that you cant use Gaussian-type indicators. There is always more behind the story if we look from different angles. & that i hope, solves our issue here. I am not a good mathematical student myself & whatever is said here is pretty much my personal opinion after reading through various available papers so do correct me if anyone spots anything incorrect i've said.
Hope you have a nice day. It's too hot at my side here now, making me melt away almost. I might become ice-cream soon LOLLL
Yours sincerely,
Wintersky
UsdZar when "trend is your friend". Long on UsdZar whenever it corrects is the only secure lucrative trade on this cross.