The market is a controlled dynamic system. - page 137
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
I take it you're giving your permission to test the model in real life?
2. With a well-designed TS strategy you can do without stop orders, but they should be virtual anyway.
a permanent lot?
Yes.
If one accepts the hypothesis that patterns in the market are changing/transforming all the time, there is no escaping the adoption of a sliding window. I don't see this as a disadvantage. Eternal patterns in the market???
Floating window to do - how do you pick the optimum size at each step?
A fixed sliding window is neither a disadvantage nor an advantage.
It is simply a forced voluntaristic method which everyone uses due to the lack of anything better.
The idea of a variable sliding window is very old and I have seen some descriptions of indicators based on it, but I have not tested them yet.
Regularities on the market live some time.
You have to exploit them while they are alive.
As they change/transform, you need to take timely actions to update the dynamic model (you need to keep current statistics),
adapt the model in use on the fly.
The beginning of a transient (signal) should be identified by some indication and this point should become the "left boundary of the window" (zero), but the "right boundary of the window" is sliding
(up to the end of the transient
process).
By taking "snapshots" of the transients over and over in the history, we obtain a kind of renewable typical reaction (in Oleg's case it is an equivalent object) which we use to make a prediction.
The fixed sliding window is neither a disadvantage nor an advantage.
It is simply a forced voluntarist technique that everyone uses for lack of anything better.
Regularities in the market live for some time.
You have to exploit them while they are alive.
As they change/transform, the dynamic model should be updated in time (keep current statistics) and the model used should be adapted as you go along.
The beginning of a transient (signal) should be identified by some indication and this point should become the "left window boundary" (zero), but the "right window boundary" is sliding (until the end of the transient
process).
By taking snapshots of the transients over and over again, we get some kind of typical reaction (in Oleg's case it is an equivalent object) which we use to make a forecast.
All that is correct only if you have a pattern that appears in quotes from time to time - in that case you can first identify it and then define "the beginning of a transient" and make the right border of the window a sliding one.
Regression models in general and Yusuf's model in particular work on a different basis - they continuously "adjust" to a changing market based on the moving window. They have a different problem - they identify a pattern at the end of its life cycle or already after its death.
Your approach has the advantage of identifying a pattern at the beginning of its life cycle in the market, but only if the pattern is valid in the future. This is unrealistic.
Question for the moderators:
Can I open two branches:
1. Doctoral dissertation on: OPTIMAL TRADE
2. Participants' comments and suggestions for the thesis.
Topic 1 is for participants in read-only mode.
Theme 2 is a regular theme.
There will be chapters in the thesis (large):
1. The entire NSP model under consideration, including the regression model with the proof of its omnivorousness and universality, converting to many known functions;
2. the Indicator and Expert Advisors based on it;
3. Profit equation;
4. Conditions for optimum trading in an ordinary market depending on selling price, trade mark-up, income, and quantity of products sold;
5. The level of competition on the market and the numerical way of evaluating it;
6. The optimum selling price of the goods;
7. The marginal purchase price of the commodity;
8. The concept of "market price", its numerical definition, its place in the price hierarchy of the market;
9. The two break-even points and the method of their numerical definition. 10;
10. Marginal selling price of goods. 11;
Price hierarchy on the market. 12;
12. Estimating the coefficients of a linear regression equation with multiple variables (simpler than Gauss and Cramer methods);
Yes.
The fixed sliding window is neither a disadvantage nor an advantage.
It is simply a forced voluntarist technique that everyone uses for lack of anything better.
The idea with variable sliding window is an old one and I've seen some descriptions of indicators based on it, but I haven't checked it myself and haven't used it.
Regularities in the market live for some time.
You have to exploit them while they are alive.
As they change/transform, they need to update the dynamic model in time (up-to-date statistics should be kept),
adapt the model in use on the fly.
The beginning of a transient (signal) should be identified by some sign and this point should become the "left boundary of the window" (zero), but the "right boundary of the window" is a sliding
(Till the end of the transition.
process).
By taking "snapshots" of the transients over and over again we get a certain renewable typical reaction (in Oleg's case it's an equivalent object) which is used to make a forecast.
There are, after all, only 8 trades.
Here's a month's work from a year ago!
http://forum.finam.ru/index.php?showtopic=13324&hl=%D1%E5%E7%EE%ED%ED%E0%FF&st=0
"The final 'chord':
http://forum.finam.ru/index.php?showtopic=13324&hl=%D1%E5%E7%EE%ED%ED%E0%FF&st=150
======================
So, according to the results of monthly commodity trades with the estimated amount of initial capital of about [b]25 thousand dollars - a profit of +$4900 was obtained. Portfolio of trades was made in such a way as to minimize possible risks. And I think it was successful - when summing up weekly results, we saw that the funds have steadily and progressively increased!
---------------------------
It may seem to some that the result is random. "Well, like - lucky! Lucky today - and not lucky tomorrow!"
I won't argue! Maybe it was! I only note that I've hardly seen anywhere else an analogue of such open, informed stock trading, practically online with a preliminary (!) description of each trade! If anyone has seen it, please give me a link!
Besides, the following fact also speaks about the non-randomness of the result. During last year and a half or two years I was trading almost on-line with preliminary (!) description of my deals on my contest account in a special thread on one of the forums. And every month I received profit from +5 to +15 percents of profit!
I used to sum up results weekly. Articles, analysis, etc. are available there: http://www.procapital.ru/showthread.php?t=34876 The description of arbitrage commodity-fund trading and explaining charts read (without exaggeration) as a thriller from any page!
----------------------------"
All of this is only correct if you have a pattern that appears from time to time in the quotes - then you can first identify it and then determine the "start of the transient" and make the right-hand edge of the window sliding.
Regression models in general and Yusuf's model in particular work on a different basis - they continuously "adjust" to a changing market based on the moving window. They have a different problem - they identify a pattern at the end of its life cycle or already after its death.
Your approach has the advantage of identifying a pattern at the beginning of its life cycle in the market, but only if the pattern is valid in the future. This is unrealistic.
This is only correct if you have a pattern that appears from time to time in the quotes - then you can first identify it and then determine the "start of the transient" and make the right-hand edge of the window a sliding one.
It takes some time for it (the pattern) to emerge "in your hands", i.e. to accumulate data with the help of the model created.
The model should be run for a while to collect data and identify a pattern, and then try to make a forecast.
Regression models in general and the Yusuf model in particular work on a different basis - they continually "adjust" to a changing market based on the sliding window. They have
The problem is something else - they identify the pattern at the end of its life cycle or already after death
Enough has been said and said about the shortcomings of Yusuf's model and its flawed application.
Probably something can be corrected, here Oleg has everything in his hands.
Are your patterns realistic?
Or what?