Machine learning in trading: theory, models, practice and algo-trading - page 745
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
Good day, everyone.
I wanted to summarize a little bit... What do we know about the future candlestick, for example? We know the opening time and the closing time. We know that it may have 3 states: a white candle in the up direction, a black candle in the down direction and a doji. We know that the probability of a "long" or "big candle" you know)) - is small compared to the "average" candle or doji. We can find a channel, or call it a range, in which the price moves. That's it? We don't know anything else? It's too small to make a forecast even for a simple classification like a down candle or an up candle... If you don't try to predict directions... you can't enter a trade without predicting the direction... it won't work, so you have to predict it anyway. What else can we say about a future candlestick that would allow us to classify it? After all all predictions on past data give signs of past candlesticks. And the prediction on this data is presented as "today will be like yesterday" - this is not good....
Well, here's a class 1 candlestick in buy, from the moment of the opening went to buy 80% and to sell 20%, class 2 candlestick went to buy 60% and to sell 40%, .... all that didn't fit into those 4 classes and you have 5 classes of candlesticks with the prediction of stop positions.
Ok, here is the class 1 candlestick in buy, from the moment of opening it passed in the buy direction 80% and in the sell direction 20%, class 2 candlestick passed in the buy direction 60% and in the sell direction 40%,..... all that didn't fit into those 4 classes and you have 5 classes of candlesticks with the prediction of a stop variant.
This is all clear, the question was not to teach me, but to analyze together what we know about the future movement. I think all members of this branch of the forum will be able to classify the next candle and without us. The signs of a future candle - so that the neuron - the signs to classify a candle. Let's summarize these signs. I'm not writing to argue and find out who is smarter will be more useful to all of us to explain and think ...
it is all clear, the question was not to teach me, but together to analyze what WE KNOW about the FUTURE MOTION. I think all participants of this branch of the forum will be able to classify the next candle without us, but for this we need to determine what to supply to the input network. The signs of a future candle - so that the neuron - the signs to classify a candle. Let's summarize these signs. I'm not writing to argue and find out who's smarter will be more useful to all of us to explain and think ...
There are different signs: black, white, red, but all equally want to overfit something.
(Signs can be different, black white red, but all equally want to overfeed on something.
But this question is not solved, if we answer the question if the open high delta is greater than the open low delta - we may go long with more confidence.
We can consider it as a sign of the next candle. The prediction will be 50/50, like when flipping a candlestick, until we gather several signs of what candlestick or movement will be in the near future.
If I understand correctly, nobody uses trawls or at least break-even? If so, the parameters of the trawl do not correspond to the trained model?
Exactly right, It is necessary to have a model that as a person made the trade. A human trader
1) Makes a forecast
2) Assesses the risks
3) Meeting the risks of entering the trade
4) Exit from a trade
Everything approximately, sketches as they say)))))
Here's another heresy about the "non-stationarity problem"...
The return is stationary and almost Gaussian, if you straighten by volatility, and only it is needed, the price itself, which is not stationary, does not participate in the calculations. Once again, the number of samples proportional sdt(standard deviation, ie variation squared), the forecast error should be an order of magnitude less than the threshold, the market data PRIMARY TARGETS, 52-53% accuracy for these 2-3% with 0.1% variation, to get this 0.1% error needs hundreds of thousands of samples, if you take 1000 will be 1% error, almost like the forecast, it is not acceptable. Dreamers, children and imbeciles with prediction accuracy of 80% on ZZ, of course it is allowed to use 10 samples, even for very talented ones 1 sample, the "key" )))). Such should not trade more than $100, take psychology into account and do not double up.
This is the best student of Warlock wrote how a neuronet should work. No one reads it anyway :)))))
)))) smiled, but the question is not solved, if we answer that the open high delta will be higher than the open low delta - then we can enter long with more confidence.
We can consider it as a sign of the next candle. The prediction will be 50/50 like when you flip a candlestick, until we gather several signs of what will be the candlestick or movement in the near future.
As I see it, the NS should have at least 2 signs, or better 3. The short-term, medium-term, and long-term description of market conditions. The others can be added, if they have some extra information in addition, for example autoregression of n-th order of the sign on itself and the like, that would take into account the dynamics of the signs.
As for the outputs - it is stupid to feed fixed values. A better solution would be to feed probabilities of growth/decline by n points at given sl\tp levels that can also be dynamic, that is if we classify signals
For regression, i.e. for N-bar forecasting we just need an additional module which processes forecast results and adaptively defines sl\tp/trailing depending on the forecast
But as mentioned above, these are all outdated techniques and rather poorly working in the market because of the complexity (impossibility) of expert evaluation of the real, not temporal, relationship of the sign/target
But the question is not solved, if we answer the question that the open high delta will be higher than the open low delta - then we can enter long with more confidence.
We can consider it as a sign of the next candle. The prediction will be 50/50 like a flip of a candle, until we collect a few signs of what will be the next candle or movement in the near future.
To calculate the next candlestick is real, but it's not real to do it with every candlestick in a long series.