Machine learning in trading: theory, models, practice and algo-trading - page 2754
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
We need to describe what we want to get in the output through a window with features and labels.
it can freeze on these reference points to the next, narrow-expand depending on conditions and with it
classification goal
I was just suggesting an alternative to the classic sliding window, since it's so annoying for everyone.
"what we want to get as an output" - this is a stumbling block of this thread (and many others too)
nobody can formulate the goal in objective terms.
(they are inclined to "eternal profit exponentially", through the search for a methodology of approach to the hole to pull out the right pike).
"what do we want to get as an output" is the stumbling block of this thread (and many others too).
No one can formulate the goal in objective terms.
(they are inclined to "eternal profit exponentially", through the search for a methodology of approach to the hole to pull out the right pike).
We're getting sidetracked. I suggest we focus on the drunken stumbling window. It promises new insights and practically covers the need to search for patterns, levels, events and other nonsense.
Overshooting and classification error on new data will show the rest.
For simplicity, you can start stumbling on reference points. That's it, I'm gone.
These repertoires will be bifurcation repertoires, and after them we should look for attractors and adjust the window to them, until the next bifurcation
in terms of nonlinear dynamics or whatever.
to put this window in the MO and forecast for the near future.
Attractors will be clear only somewhere after the middle of this self-affine piece of the chart, before that nothing will be clear, or a larger or smaller one will be in effect
there is probably an algorithmically simple way to take everything into account in the training examples, without having to worry about it.
It is quite interesting to determine in advance the price/time limits when it will be necessary to rebuild this construct, if the repertoires change significantly. But this is purely academic interest.
In practice - once a week (the minimum natural cycle of business planning) recalculated, recalculated. Because you can't get away from cycles and Fourier.
Everything is conservative in this respect. There is a suspicion that the supply of negroes to the Caribbean every six months still affects the quotations :-)
Which of the global optimisation algorithms converges fastest, who knows?
I see)))
And here he is, replacing my concept of " Event" with "Repertory Points" and pretending he wasn't told about it a dozen days ago.... Yeah.
Yeah, I like that word (something that's native to event-driven programming) -
in the variant voiced by SanSanych Fomenko -- something like this seems to be implemented: ejection -> means input (or output)... I've mixed a mess of dim_reduction and multidimensional classification methods (LDA, clustering) above.... but the essence of Mahalanobis has probably always been primarily in multidimensional space as outliers/novelty detection anyway... so the option of trading on outliers looks very nice (only feature_engineering should be done correctly, not a dumb set of initial data to be searched for fs)...
but still the "sliding window" is confusing (although the usual autoregressive model is common for timeseries-following-trading).... - there can be a mess there too (in the window)... -- I assume the window boundaries are the entry into the market of smarts, who incidentally use in their portfolio_management -- Mean-Variance Optimisation... we don't know their portfolio of course (only roughly from SoTs retrospectively), but within this variance they probably do rebalancing of their portfolio --- while fixing or loading about retail...
trading on outliers is certainly an interesting option, but considering who is up against it - OTF or DTF (day traders) -- is also important to interpret the outlier correctly...
p.s..
well, or just take not Mahalanobis outliers, but extreme deciles of prediction distribution -- for risk-acceptance (vs. risk-aversion) behaviour, for example, of an actor (with corresponding switching of its state according to env. parameters).
I see.)
Yeah, I like that word (something so native to event-driven programming) -
Reference points are usually a consequence of a cause event, so reference points or areas or segments as a consequence of events from the real world are closer in meaning.
Of course, you can consider a price series as a separate world and define events in this world, fantasising that pattern events influence further price behaviour, but it's not the same)))))
I still do not understand why a one-dimensional array is a vector, and vector algebra, there the direction of array indices is almost never used and does not carry such a semantic load, as in a regular vector)))) Here I am tormented)))))
pattern events influence further price behaviour, but it's not the same)))))
Not always, and not always the same. There are identical patterns caused by events that are not identical, we have a cumulative impact beyond our control. I.e. the case error will always be without controlling FA events. That's why it's not like that.
In general, I am inclined to the idea that the target is the behaviour of decisions/our actions on the price behaviour. I find the price function a bit cramped, after all, a function is something definite, even with some probabilities, corridors, but with rules and interrelations.
I am looking for an idea of defining behaviour through a grid of levels, uniform, or levels of historical extrema) As if it is necessary to apply both approaches, while I am solving the problem, what should be calculated from a number that would set the right grid step, and how often it is necessary to change this grid)))))