Market patterns - page 25
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
Logically, I have a rough idea of how to do it, but all that's left to do is implement it))
This is my thought: noise is generated, for example with a constant probability density function values vs time, with MO =0. Then some differential (C(t)-C(t-1) or O(t)-C( t) or the average) is calculated, then the noise is fitted to a series of differentials from the FI, with MO and probability density depending on amplitude, which is approximately normal for FI. A mixture is made, sequentially with random lengths of individual sections, and then a cumulative plot from those increments is calculated. Then there will be no "seams" and it will look like one instrument. But there will be patterns in some places and no patterns in others.
simpler: you take a real series and with 0.5 probability you flip the candle, and with 0.5 probability you leave it unchanged. Flipping it means changing the sign of the increments for Close, High and Low.
The volatility structure of such series will be similar to that of a real series, but the direction will be random.
Congratulations.
Blue SB, red CVR. The key has been attached.
Thanks for the rows!
It was so from the author. Attached the version in numbers, if anything.
simpler: you take a real row and with probability 0.5 you flip the candle, and with probability 0.5 you leave it unchanged. Flipping it means changing the sign of the increments for Close, High and Low.
In such a series the volatility structure will remain as in a real series, but the direction will be random
Yeah, as an option. The aim is to understand on what minimum number of series to test the bot and how to interpret the results, to maximize the reliability of the bonus test.
Well, in general, with the measure of reliability that suits me, I approximately came to the method of remote verification of bots.
Sort out what the crowd is doing at the moment (there are only two options: buying or selling) and do the opposite. :)
Gentlemen seekers, there is one stationary law in the market - the Crowd ALWAYS LOSES !!!
Figure out what the crowd is doing at the moment (there are only two options: buying or selling) and do the opposite. :)
This meta-principle implicitly underlies all rollback strategies. Formalised in oscillators.
If we understand how oscillators work, we can see their common factor, it is analogous to the 2nd derivative for discrete series. Simply, it's kind of like "acceleration". We should act not by "speed" but by "acceleration", the crowd is the first derivative - "speed", it often becomes obvious (determinedby trend indicators) when the "acceleration" is directed to the opposite side and we need to exit or turn over, taking into account the lag of filters. That is why it seems to be a play against the trend but only in this way it seems, in fact, because state detection is lagging and when the "acceleration" is detected, it is in fact an extremum or the beginning of a micro-trend, but how long it will last only God knows. For SB such technology is meaningless.
For tsvr only statistics can give an answer as far as what makes sense.
I won't talk about this particular situation, because I don't care about exceptions, but in general:
Only the price change potential that occurs after some pattern of states, which occurs more often than average, should be taken into account.
Those structures that cause reversal/reinforcement/decrease/continuation of the trend but that either cannot be identified (not formalized as a vector) or cannot be clustered do not make sense to consider.
Two candlesticks and volume are too small a vector, if we collect statistics of outcomes after such a condition, it will be a random outcome for sure. As my little experience shows, it is at least 10 components, preferably closer to thirty.
Also worth mentioning about OS-relativity. I haven't trusted candles for a long time. And that the functionality of the volume depends on its source and is constantly changing.
In general the essence of this topic is how to describe patterns, clustering, finding on the TzVR and comparing with the benchmarks.
But nobody has yet said anything sensible on this topic and you can understand why.
This picture is really getting to me...
In general Nanex are good, they clearly represent the data. They should also display the glass data (level2) in the same way, in a similar way, because the digits are not so quickly interacting with the brain.
I think that in order to search for something one should first present the data in a convenient way.
Actually, my post is about psychology (crowd moods), not patterns.
In the example I showed one possible way to identify the moods that are currently present in the market. Everyone has to define their own criteria for this evaluation. But the law remains the same - Crowd ALWAYS LOSES.
"The crowd always loses," Fred Kelly wrote in his 1930 work on the stock market, "because the crowd is always wrong. They are wrong because they behave normally.
THE CROWD ALWAYS LOSES !!!
Actually, the crowd is the market, just as sand is the sandbox, saying that the crowd is always wrong means that the market is wrong, and this contradicts the postulate "the market is always right".
If we say that on an evident trend one must act in the opposite direction, we must agree with each other. But even here there are many nuances regarding timeframes and so on. This does not apply to one-minute bars for sure, at least from one-hour bars we can talk about such a pattern.
How do you determine the trend? When it has gone a long way. It does not just goes up for a couple of hours, but when it has crossed the boundary of the previous channel and has not stopped, all stops and margin call have triggered and added oil to the fire. Well, this is usually the last recharge, and it either corrects or collapses.
What's the moral? Firstly, don't wait for the channel to be broken through. Secondly, as it was mentioned above, consider the first signs of the price movement.
In general, get on the train when it just starts, and get off when it starts to slow down. Calculate quantitatively, which timeframe is convenient for you and gives statistical advantage at tests.
So the crowd cannot be wrong. The part of it that follows the initial impulse is not right. The part that creates the initial impulse is the one that takes the cream off.
The momentum by itself can't appear, and market makers can't make it on Forex because of its huge liquidity. Only a spark can be thrown in at the right moment. Those who recognise the spark and start the fire are the winners.
A momentum by itself cannot arise, and market makers cannot generate it in forex because of its enormous liquidity. Only a spark can be thrown in at the right moment. Those, who recognised the spark and started the fire, are the winners.
What are you saying, "instantly"?! What bars are you thinking with? And anyway, it was a very small swing.
And as for "Trading Chaos", that's the psychologist's demagoguery. His "Alligator" is a fey! Shifting the already laggy mash-ups forward is genius! Especially as a result of 20 years of practice.