Recognise changes in the "behaviour" of a financial time series (Trading on the news) - page 6
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
It doesn't have to predict - and it's useless.
It is possible to build such a model. Something like a [differential] parametric equation of motion. And a stochastic one, too.
At the moment of the break, some dipure parameters determining the character of the motion change by leaps and bounds. All that comes immediately after the break is relaxation, which is approximated by a second-order linear equation of motion similar to the equations of mechanics.
This is where one can try to make money on relaxations. But this is if we can detect kinks quickly.
The most important thing is to come up with a model of motion during the relaxation.
The market is moved by news, but this is not necessarily visible to the eye.
The market moves not necessarily because of news. In the complete absence of news it can still move.
All imho.
P.S. I suspect that some stones will be thrown at me - a mechanistic approach, pulling the market on formulas, etc. No, not really, although some "mechanistic" categories are present - tool mass, force, even the Lagrange function (essentially energy).
It has been forming in my mind for a long time, about 15 years ago. I gave it up when I thought the model was not concrete enough. But sometimes I remember it.
That's not going to cut it. The only thing the news does is cluster volatility - the news is strong (or many) - volatility is high, the news flow is weak - volatility is weak. The direction was 50/50 and remains so. You just will lose more and earn more on the news, but the sum is still 0.
I have argued and continue to argue that "the market is driven by news" and random drifting is also driven by what-if rumours, prejudices and maybe news. The news is varied and I know of very "smart" and thick books that tell how the market moves from this or that news.
There are all kinds of news and the lack of news is also news.
I won't count fundamental analysis as news - that's a type of news and a separate conversation.
.
But it is impossible to trade on news. For two reasons: for unexpected news we get an unexpected market reaction (Syria attacked Iran - what will happen to the market?). Funnily enough, the same reaction to expected news. For example, Greece withdrew from the euro. The reaction to this expected news will be obligatory, but it cannot be predicted and therefore cannot be traded. The result depends on the extent to which this expected news is taken into account by the market at the time the news is released. It could be anything from a jump forward to a pullback depending on whether the market is overbought or oversold relative to the particular expected news.
.
The TS should be able to react to the news. Testing of TS (which is possible using artificial data) should include two patterns of behavior: the news - the change of quotes, and the return to the previous movement line. The news - change of the quote and the movement on the new line.
I have argued and continue to argue that "the market is driven by news" and random drifting is also driven by what-if rumours, prejudices and maybe news. The news is varied and I know of very "smart" and thick books that tell how the market moves from this or that news.
There are all kinds of news and the lack of news is also news.
I won't count fundamental analysis as news - that's a type of news and a separate conversation.
.
But it is impossible to trade on news. For two reasons: for unexpected news we get an unexpected market reaction (Syria attacked Iran - what will happen to the market?). Funnily enough, the same reaction to expected news. For example, Greece withdrew from the euro. The reaction to this expected news will be obligatory, but it cannot be predicted and therefore cannot be traded. The result depends on the extent to which this expected news is taken into account by the market at the time the news is released. It could be anything from a jump forward to a pullback depending on whether the market is overbought or oversold relative to the particular expected news.
.
The TS should be able to react to the news. Testing of TS (which is possible using artificial data) should include two patterns of behavior: the news - the change of quotes, and the return to the previous movement line. The news - change of quotes and movement on the new line.
Then it turns out that the moment of news publication can be used as a reference point for measuring the parameters of a price series. Assuming that the series will be stationary for some time.
There is always a countdown, but how many bars does it take to calculate the parameters? Will you have time to trade or will there be new news?
There are TCs that can work on certain types of non-stationary quotes. One does not follow from the other.
News is different and a lack of news is also news.
I was just talking about my imho model. In it a jump change of parameter can pump energy into the system or vice versa. It's common for parametric diffusers.
But with me "energy" is not some abstract term, but something expressed by a very specific formula from the parameters of the diffura itself.
I respect your point of view and am not going to argue with you, SunSunich.
I was just talking about my imho model. In it, a jump change in a parameter can pump energy into the system, or vice versa. This is common in parametric diffusers.
But with me "energy" is not some abstract term, but something expressed by a very specific formula from the parameters of the diffura itself.
I respect your point of view and am not going to argue with you, SunSunich.
When news is released, the market changes its main characteristic - volatility. You can estimate the strength of the news flow pre-factually, by the same volatility and the slope of the regression, before a strong flow the market is usually in a thin flat. You can use the breakdown of the same volatility with close stops. But that's theory, in practice requotes, breakdowns, spread spread widening, fisheries watchdogs.
And there will be a failure here. You bet on a breakout. The strong news will come out, the order will be triggered, and then at the same strong volatility the price will successfully turn around and move in the opposite direction. Certainly, we can hope that the jump will be much larger than the level of a breakdown order, and in this case take profit might help against the reversal. But one must be able to predict the strength of volatility and the strength of the news flow will not help. To say that the market is in a thin flat before the strong newsflow is like saying that the market is so volatile that in the previous period all market movements seem to be small flat.
I have seen many times, when some strong news is released and the market stays almost motionless. And vice versa, when seemingly insignificant news causes an avalanche-like price movement. To explain this effect, we will have to discard the EMH theory and admit the fact that the market considers news instantly, but reacts to it only when the sum of such news becomes critical. It is like an overflowing reservoir. The new equilibrium price will be determined when the past level is no longer capable of describing the totality of the situation. Take the "reasonable investor". By EMH logic, he should instantly change the structure of his portfolio correctly and rationally reacting to each news. But this is nonsense! How, for example, a manager of a multi-billion-dollar portfolio can balance his portfolio every second (there are a lot of news)? He cannot. Even with the questionable assumption that the investor is able to "correctly and rationally" react to the news. It's too big and large for such frequent changes. Instead, another, more realistic approach is seen. An investor learns new negative information about a stock in his portfolio. Instead of rushing headlong to sell it, he takes the news "on his pencil" and decides to wait for additional confirmation. When there are enough confirmations he starts to react and get rid of the ugly duckling. This last confirmation, after which he will say "I've had enough" does not have to be significant. Remember the movie of the same name? The man was refused breakfast because the clock was 10:02, and breakfast was given out before 10:00. It was a small thing, but it was the impetus for the dramatic development of the story.
I was just talking about my imho model. In it, a jump change in a parameter can pump energy into the system, or vice versa. It's common for parametric diffusers.
But my "energy" is not some abstract term, but something expressed by a very specific formula from the parameters of the diphur itself.
There is also this interpretation. Any transaction (well, almost any transaction)) is influenced by a certain amount of information available to the agent, and absolutely any transaction, depending on the volume even by a millimeter, has an impact on the price movement. Thus, the price movement can be represented as a result of processing by agents of various information coming from the news, from an insider, from a neighbour or directly from the space. The specific source is not really important, what is important is that any information upon which the decisions are based can be conditionally divided into "information noise" (i.e. information that on average gives zero expectation of price movement after processing by agents /for example, short-term gossip in the inner circle/) and "useful information signal" (i.e. information that finally results in the non-zero shift). Here it should be noted that any information trigger for a trade will not be instantly worked out by all agents: in the simplest example, the data on GDP volume in the past quarter can be used to enter immediately at the time of exit, or it can be used a month later (this consideration, by the way, immediately leads us to the non-markoving nature of the process). Each input turns out to be "smeared" on the time axis, so to speak, and at different sections the movement under its influence can occur either in one or in the other direction.
The next thing is this reasoning. If we know how the market is spherical in vacuum (i.e. isolated from any influences other than the one in question) reacts to a single news signal, then we can calculate the response function (and thus the structure of the responding black box) and use it to detect useful signals (i.e., as we remember, leading to a non-zero shift) already in the noisy version (here DSP apparently already works).
Thus, the problem comes down to describing the structure of the system correctly and defining its parameters (albeit floating ones). But this is precisely the most wicked practical problem: we do not know how the market is stratified, hence the structure has to be guessed. I remember once gave an analogy with a kind of a fractal oscillatory loop, where agents with different horizon are grouped, roughly speaking, by timeframes (read: by the size of available funds) and make oscillations in response to information, each at its own scale and frequency. Incidentally, this analogy came to mind after decomposing into empirical modes by NNT.
... if you are interested, give me a kick in which direction to reason further))
I have seen many times when mega strong news is released and the market does not move at all. And vice versa, seemingly insignificant news causes an avalanche-like price movement. To explain this effect, we will have to discard the EMH theory and admit the fact that the market considers news instantly, but reacts to it only when the sum of such news becomes critical. It is like an overflowing reservoir. The new equilibrium price will be determined when the past level is no longer capable of describing the totality of the situation. Take the "reasonable investor". By EMH logic, he should instantly change the structure of his portfolio correctly and rationally reacting to every news. But this is nonsense! How, for example, a manager of a multi-billion-dollar portfolio can balance his portfolio every second (there are a lot of news)? He cannot. Even with the questionable assumption that the investor is able to "correctly and rationally" react to the news. He's too big and large for such frequent changes. Instead, another, more realistic approach is seen. An investor learns new negative information about a stock in his portfolio. Instead of rushing headlong to sell it, he takes the news "on his pencil" and decides to wait for additional confirmation. When there are enough confirmations he starts to react and get rid of the ugly duckling. This last confirmation, after which he will say "I've had enough" does not have to be significant. Remember the movie of the same name? The man was refused breakfast because the clock was 10:02, and breakfast was given out before 10:00. It was a small thing, but it was the impetus for this dramatic development of the plot.