From theory to practice - page 122
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
If we add non-stationarity, with which SanSanych is running around like a chicken with an egg, in a form of asymmetry coefficient of the current distribution, then yesterday we would gain more than 500 pips on AUDCAD. Only one deal of the week - but what a deal!
See the model for VisSim. The files must be located in the directory C:\Forex.
Look, Peter - the ticks make a nice physical and mathematical picture and it becomes clear what we are dealing with.
Let's say we get 15500 ticks as input with some non-random time sampling. What is Vladimir getting at here? He is essentially saying that the maximum deviation from the mean is the root of t. Essentially the root of 15500 = 124.5 pips. I.e. multiplying by some quantile of the Gaussian distribution e.g. = 3 we obtain that the maximal distance the price can deviate from the average for the Wiener model = 373.5 pips. When the price leaves this range, let's make a deal.
No, it doesn't! My calculations give an average deviation = 145 pips for AUDCAD. The occurrence of 1 unit of data out of 15500 outside the general probability space for the Student's distribution occurs at about 4*sigma. That is, the maximum distance the price is capable of deviating from the mean in reality = 580 pips for a sample of 15500 ticks.
I.e. it is convenient to build maths on ticks.
Give me a link where I said such nonsense: "the maximum deviation from the average equals the root of t". Or again are you just absolutely convinced that I would have said it?
What about the corridors of a certain Katya Savkina??? After all, they are the ones for the root of t - i.e. for the Wiener model!
Not "ie", keep them, but a link to my words with this nonsense.
Look, Peter - the ticks make a nice physical and mathematical picture and it becomes clear what we are dealing with.
Let's say we get 15500 ticks as input with some non-random time sampling. What are some people getting at here? They are essentially saying that the standard deviation from the mean equals the root of t. Essentially the root of 15500 = 124.5 pips. That is, multiplying by some quantile of the Gaussian distribution e.g. = 3, we get that the maximum distance that price can deviate from the mean with the Wiener model = 373.5 pips. When the price leaves this range, let's make a deal.
No, it doesn't! My calculations give an average deviation = 145 pips for AUDCAD. The occurrence of 1 unit of data out of 15500 outside the general probability space for the Student's distribution occurs at about 4*sigma. That is, the maximum distance the price is capable of deviating from the mean in reality = 580 pips for a sample of 15500 ticks.
I.e. it is convenient to build mathematics on ticks.
How is it better than ATR and primitive maths in a window of max 1000 five-minute bars when calculating by bar opening? You have one signal purely news - the meaning and mathematics are zero (nfp games - place pending orders in advance), because the spread is wild, the second signal is 15:45 "stochastic in the direction of the daily trend". VisSim, sigmas/delta to repeat the 1st em and one stochastic?
How is it better than ATR and primitive maths in a window of max 1000 five-minute bars when calculating by bar opening? You have one signal purely news - meaning and mathematics zero (nfp games - place pending orders in advance) as the spread is wild, the second signal 15:45 "stochastic in the direction of the daily trend"
Here's a look around the branch at some of Vladimir's research. Awesome! In what book can one find such a thing? There is no such thing - only here on the forum you can see cool stuff.
That's the thing, maybe other models are good, but I personally haven't seen rigorous sample volume calculations, quantiles or any other rigorous mathematics anywhere. Maybe they are working models - I don't know. But without mathematics and physics - nowhere, I'm afraid to use. I need to understand why things happen this way and not that way. In a good way every trader should be able to explain the result of each transaction. Do you agree?
No, why? For examplehttps://www.mql5.com/ru/forum/218573/6006792#comment_6006792 - ATR(13), what would be the benefit of applying a more rigorous/intricate/monstrous construct(s)? Trading sessions and trends are fait accompli (a set of constants) that have repeated so far and will continue to do so - why prove and justify it?
Link to EMA https://www.mql5.com/ru/forum/165546/6214086#comment_6214086, further read there about economics, a logical framework. Any increase in complexity must give quality: make it 1000 times more complex - accuracy should increase by at least 10-20%. You have complicated it 1000000 times and obtained stochastic +ema.
No, why? For examplehttps://www.mql5.com/ru/forum/218573/6006792#comment_6006792 - ATR(13), what would be the benefit of applying a more rigorous/intricate/monstrous construct(s)? Trading sessions and trends are fait accompli (a set of constants) that so far have repeated and will continue to do so - why prove and justify it?
Link to EMA https://www.mql5.com/ru/forum/165546/6214086#comment_6214086, further read there about economics, a logical framework. Any increase in complexity must give quality: make it 1000 times more complex - accuracy should increase by at least 10-20%. You have made it 1000000 times more complex and obtained stochastic +ema.
Now watch what I do.
It's my job to give out free EA to everyone who wants it. It has to work for everyone, on any data stream from any brokerage company. How to do this? The answer is to accept ticks according to a unified algorithm.
The new messiah announced to the locals that from now on they did not need to work, as the island would soon begin to receive an unlimited supply of kargo.
.
Look at the distribution of time intervals between ticks - it's almost exponential. But, still - it's different for different DCs due to different intensity of flows. I will unify it forcibly - now it is the same for all and it is exponential.
Now it turns out that we have simplified the model of a non-Markovian process to a Markovian one with pseudo-states. And for it it's just the expression S = sqrt(N*mean(|Ask(t)-Ask(t-1)|), where N is the number of ticks in the observation time t.
The emphases have shifted, compared to this :
Alexander_K:
Of course, this is the most important issue.
I think that in the case of a non-Markovian process we should trade against the trend, and for a Markovian process we should trade along the trend.
Next week I will investigate the probability distribution of tick arrival time - let's see what it is for different pairs.
If it is nonexponential - then the processes are not Markovian and vice versa.
I will post the results on the forum.
Does it mean progression?
But you already realise this is nonsense?(2017.11.06 04:01#64)
.
Do you understand anything of what I am writing about?
Of course I do. -- nonsense.
I also understand that you are completely missing the point.