Discussion of article "Exploring Seasonal Patterns of Financial Time Series with Boxplot" - page 2
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
like how to look for drifting cycles.
They will be found if you keep going. They are exactly drifting cycles in the market, and that is its great secret.
Just because some data a person sees on charts doesn't mean there's a pure pattern there.
It's not some data, but unambiguously interpretable data. That in certain hours the market is growing, they cannot be understood in any other way. Unlike if you optimise something there and don't find the ends afterwards
it's not some data, it's unambiguously interpretable. That in certain hours the market is growing, they cannot be understood in any other way. Unlike if you optimise something there and don't find the ends afterwards
I.e. if you have enough skill to interpret it, there is a pattern, but if you don't, it doesn't seem to exist?
I.e., if you have the skill to interpret it, there's a pattern, if not, there doesn't seem to be one?
In the second case, it would be impossible to understand why the TC broke down, no?
It seems to be there, but it is impossible to describe it in human words. It's all relative, it's a philosophical question.
I started to search in this way, because the neural network is almost always a curse of dimensionality and it's impossible to understand anything.
it's not some data, it's unambiguously interpretable. That in certain hours the market is growing, they cannot be understood in any other way. Unlike if you optimise something there and don't find the ends afterwards
If you put the horse before the cart, it is rather interpreted as "the general market movement is determined in specific hours (dates)".
It's just the way the world works - at this time important volumes go around on important business, and all other speculators have to adjust to it.
If you put the horse before the cart, it is more likely to be interpreted as "the general market movement is defined in specific hours (dates)".
It's just the way the world works - at this time important volumes are moving on important business, and all other speculators have to adjust to it.
it is interesting that the found pattern is purely a pullback pattern, i.e. the rest of the time the market was mostly falling for all 3 years
Didn't want to comment on the article, but still )))))
the article is certainly interesting, because the author gave a completely ready-made methodology for evaluating price series using Python - for this thank you very much, as they say from the heart! - time was saved very decently!
What I didn't want to comment on is the end of the article, the section "Checking regularities with trading logic" - if I don't describe it for a long time, it's just all drawn out by the ears - here is the target sample of 2015-2019, we apply filtering, we get... we get nothing! 2015-2017 did not drain the TS only because there were fewer and no more trades, i.e. the conclusion that "We can see that the system has become more stable in the interval of 2015-17 years". - well, it's just pulled by the ears?
OK with the bad period until 2017, let's talk about what had a profit: here is the source, here is the statement that if you trade on MA, then at certain hours there will be a profit..... well, yes the balance chart of the tester shows an upward trend, but the code of the EA, without stoplosses, maybe I read the code badly, but I do not see the opening of orders to sell (OP_SELL), there are only purchases (OP_BUY).
In general, testing in the tester does not show anything at all, imho, if there is a statement "The market is growing at certain hours, they cannot be understood in any other way". - then it unambiguously defines growth relative to something and accordingly then implies a decrease relative to growth - I again where the sales test????. As well as checking of regularities without limitation of loss (stoploss), well it is like an old Indian saying: "if you sit on the river bank for a long time, you can see the body of your enemy floating by" - i.e. there will be a profit some time anyway.
SZY: someone once wrote on forums that MA always returns to the price, or the price returns to MA, well, it's kind of a regularity too? ;)
I saw the regularity of buying with my eyes, there is no smell of selling there, you can't sell there. Put stops, it will be the same.
For sales, look at 11-15 hours, you should try under them. I didn't try it as the purpose was to give information on boxplots
15-17 years do not show the same pattern, I just tweaked the TS so that it does not pour and showed on them that in those years - differently!
This is not a bad pattern, by the way, which can be traded right now, with short stops. But I am not agitating anyone for this, the purpose of the article is boxplots and a handy research notebook
there's no question about it, it's well-prepared material
OK, then why did you choose the TS on MA? The TS should have just opened an order with take and stop loss - everything is clear and understandable, here is the methodology for evaluating the MD, here is the test
and trading on TS "price above/below MA"... well, this is the agents from DCs make up all sorts of nonsense and people already perceive it as an adequate TS ))))
on MAs only make sense TS on several MAs, you can use it even on the Dow to look for trends.
ok, to whom I wanted to write, I've read it, I'll delete the posts, it's not good - I don't do that usually, I'm sad in the middle of the night ))))
MA was used for detrend, the article says. you can use increments, which is the same almost. It would be a different article without it. You can use regression or something else. Detrend is necessary anyway, otherwise you don't get the statistics. Since the stats are obtained with a specific detrend, the ts is traded based on that.
I didn't think that such questions would arise... I wanted everyone to start using the notepad )).by the way, yes, there is a mistake with MA.
imho statistical analysis with de-trend cannot be done by moving. It should be something close (but not equal) to just average :-) The history is well known, it is possible not to use surrogates.
In this case, if the result is a "defining" area or pattern, then the goal is achieved - the mega thing has been found - after its appearance the trend is a bit predictable.
PS/ about the fact that "the pattern is purely rolling" - I will look at the weekend. I need to see if there is a pattern at all
by the way, yeah - MA is wrong.
imho statistical analysis with de-trend cannot be done by moving. It should be something close (but not equal) to just average :-) The history is well known, you can avoid using surrogates.
In this case, if the result is a "defining" area or pattern, then the goal has been achieved - a mega thing has been found - after its appearance the trend is a bit predictable.
PS/ about the fact that "the pattern is purely rolling" - I will look at the weekend. I need to see if there is a pattern at all
Detrend by MA is absolutely normal in econometrics, you can replace it with rolling regression or more complex analogues. I predict: the result will be +- the same