A class is predicted. Prediction error is 25 to 30%, and it is the same in training and out of sample. the model is NOT retrained.
This is very cool, grail, that is, accuracy 100% - error = 70-75% on such a prediction is a sin not to squeeze the Sharp in 4-5, so if nowhere accidentally peeked then boldly to battle!
I finally started tweeter, downloaded messages (in the most primitive form) so to speak "buy character" and "sell character" by eur, I want to look at these messages as a time series on the euro chart, but I do not know how to turn these messages into a time series and then synchronize with eur, help the code who understands time functions in R-ka...
I attached a file in RData below...
the file has a sheet in which there are two sheets :) one with the messages "buy character" the second "sat down
Since I did not see a vigorous response to my question, it once again proves that you do not understand in principle what the market is, a couple of this is the reason for failures. By the way, my article was accepted for editing. The main thing is that I cannot make any profit if I do not pay too much. But in order to raise the status of the branch I can announce the result here before the publication of the article. So what is the market? More precisely, who is a market participant???? Yes Market makers, big players, small spokesmen, various funds.... you will answer and I will say YES, that's right. BUT all these participants are divided into two categories BUYERS and SELLERS!!!!!!! When they make transactions between themselves, they form the volume that pushes the price, but in addition to the volume, we have information available to us about the difference between buyers and sellers. So called DELTA. This is what causes the price to change. As well as the reason for the result of many TC. But only this secret, I hope you understand :-)
Do you think it's a coincidence that the "panic like" peak coincided with the reversal, because there were as many as 8 messages in all this panic :))))
Now I had time to double-check - I pulled the texts of those tweets from your rdata file.
22 2017-02-22 13:59:41 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/Et9um07c4g #forex #eurusd #fx #news 23 2017-02-22 13:56:14 RT WhyLose "EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://tco/Mg2Lj85Wd7" 24 2017-02-22 13:56:08 $EURUSD EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/jfaGQM4hsI S/R Levels https://t.co/3XxB9fiazN #E 25 2017-02-22 13:50:49 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/EZBQAp5w1i 26 2017-02-22 13:50:48 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/qpD3KWOYms 27 2017-02-22 13:45:37 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/dS8lcooPc3 $EURUSD #Forex https://t.co/ohDZQygceH 28 2017-02-22 13:45:16 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/KQOHnaR8oJ #EURUSD #Politics #DollarIndex #Fed 29 2017-02-22 13:45:16 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/IiTe3R5hTT #EURUSD #Politics #DollarIndex #Fed 30 2017-02-22 13:10:00 RT @ElieWaked: #EURUSD SELLING AGAINST 1.0650 IS A GOOD TRADE TO 1.03 #RISKMANAGEMENT
The green tweets -selling pressure abates- translates as "sellers pressure is dropping". That's more of a signal for a flat. In fact, they say, that the price rebounded from 1.05000 two bars ago, and it is not even a forecast but a statement of fact. The valid sell signal is the last one. Some kind of strange situation, we need to watch more situations with such cloudy tweets and further price development, common sense says that they will be right 50% of the time.
Some strange situation in general, need to observe more situations with such murky tweets and further price development, common sense says that they will be right 50% of the time.
You need more time to download, like not three days and a few weeks, for example, and then look on the spot, also a huge role played by the setting of keywords, if you score something fairly common there all tweets will be all solid advertising ... you're a slag.
The problem is that I haven't figured out how to download a few weeks of tweets there
Well of course it's not that simple, you first need to know how to apply it and not only it, the volume itself, or rather changing the volume will also be a good reason. I, using a delta, and can not get good enough models, and not using, it's like flipping a coin, it may work, and maybe not...
Allow the use of cookies to log in to the MQL5.com website.
Please enable the necessary setting in your browser, otherwise you will not be able to log in.
You are signed in. Two-factor authentication has been disabled. Set it up again and scan a new QR code, or continue with two-factor authentication disabled.
A class is predicted. Prediction error is 25 to 30%, and it is the same in training and out of sample. the model is NOT retrained.
I finally started tweeter, downloaded messages (in the most primitive form) so to speak "buy character" and "sell character" by eur, I want to look at these messages as a time series on the euro chart, but I do not know how to turn these messages into a time series and then synchronize with eur, help the code who understands time functions in R-ka...
I attached a file in RData below...
the file has a sheet in which there are two sheets :) one with the messages "buy character" the second "sat down
to see the time of each message
[1] "2017-02-24 17:06:54 UTC"
message itself
[1] "Buying Pressure seen building at Support EURUSD 1.05894000 https://t.co/XN9MZcnRc8"
Kaoroch help visualize and synchronize the eurora, because it is very interesting what can happen
Kaoroch help visualize and synchronize with the euro, because it's very interesting what can happen
What I got is what I got :)
What came out, came out :)
Thanks for your help!
Do you think that the "panic-like" peak coincided with the reversal, because there were 8 messages in all that panic :))))
Do you think it's a coincidence that the "panic like" peak coincided with the reversal, because there were as many as 8 messages in all this panic :))))
22 2017-02-22 13:59:41 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/Et9um07c4g #forex #eurusd #fx #news
23 2017-02-22 13:56:14 RT WhyLose "EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://tco/Mg2Lj85Wd7"
24 2017-02-22 13:56:08 $EURUSD EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/jfaGQM4hsI S/R Levels https://t.co/3XxB9fiazN #E
25 2017-02-22 13:50:49 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/EZBQAp5w1i
26 2017-02-22 13:50:48 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/qpD3KWOYms
27 2017-02-22 13:45:37 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/dS8lcooPc3 $EURUSD #Forex https://t.co/ohDZQygceH
28 2017-02-22 13:45:16 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/KQOHnaR8oJ #EURUSD #Politics #DollarIndex #Fed
29 2017-02-22 13:45:16 EUR/USD selling pressure abates near 1.0500 handle as Fed minutes loom https://t.co/IiTe3R5hTT #EURUSD #Politics #DollarIndex #Fed
30 2017-02-22 13:10:00 RT @ElieWaked: #EURUSD SELLING AGAINST 1.0650 IS A GOOD TRADE TO 1.03 #RISKMANAGEMENT
The green tweets -selling pressure abates- translates as "sellers pressure is dropping". That's more of a signal for a flat. In fact, they say, that the price rebounded from 1.05000 two bars ago, and it is not even a forecast but a statement of fact.
The valid sell signal is the last one.
Some kind of strange situation, we need to watch more situations with such cloudy tweets and further price development, common sense says that they will be right 50% of the time.
The so-called DELTA. That is the reason for the price change.
Some strange situation in general, need to observe more situations with such murky tweets and further price development, common sense says that they will be right 50% of the time.
You need more time to download, like not three days and a few weeks, for example, and then look on the spot, also a huge role played by the setting of keywords, if you score something fairly common there all tweets will be all solid advertising ... you're a slag.
The problem is that I haven't figured out how to download a few weeks of tweets there
That's a little too naive.
Somehow too naive
Seems too naive
Well of course it's not that simple, you first need to know how to apply it and not only it, the volume itself, or rather changing the volume will also be a good reason. I, using a delta, and can not get good enough models, and not using, it's like flipping a coin, it may work, and maybe not...